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 CString macro_equal_buf
;
47 static TokenString tokstr_buf
;
48 static unsigned char isidnum_table
[256 - CH_EOF
];
49 static int pp_debug_tok
, pp_debug_symv
;
51 static void tok_print(const char *msg
, const int *str
);
53 static struct TinyAlloc
*toksym_alloc
;
54 static struct TinyAlloc
*tokstr_alloc
;
55 static struct TinyAlloc
*cstr_alloc
;
57 static TokenString
*macro_stack
;
59 static const char tcc_keywords
[] =
60 #define DEF(id, str) str "\0"
65 /* WARNING: the content of this string encodes token numbers */
66 static const unsigned char tok_two_chars
[] =
68 "<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
69 "-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";
90 '.','.', 0xa8, // C++ token ?
91 '#','#', TOK_TWOSHARPS
,
95 static void next_nomacro_spc(void);
97 ST_FUNC
void skip(int c
)
100 tcc_error("'%c' expected (got \"%s\")", c
, get_tok_str(tok
, &tokc
));
104 ST_FUNC
void expect(const char *msg
)
106 tcc_error("%s expected", msg
);
109 /* ------------------------------------------------------------------------- */
110 /* Custom allocator for tiny objects */
115 #define tal_free(al, p) tcc_free(p)
116 #define tal_realloc(al, p, size) tcc_realloc(p, size)
117 #define tal_new(a,b,c)
118 #define tal_delete(a)
120 #if !defined(MEM_DEBUG)
121 #define tal_free(al, p) tal_free_impl(al, p)
122 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size)
123 #define TAL_DEBUG_PARAMS
126 //#define TAL_INFO 1 /* collect and dump allocators stats */
127 #define tal_free(al, p) tal_free_impl(al, p, __FILE__, __LINE__)
128 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size, __FILE__, __LINE__)
129 #define TAL_DEBUG_PARAMS , const char *file, int line
130 #define TAL_DEBUG_FILE_LEN 15
133 #define TOKSYM_TAL_SIZE (768 * 1024) /* allocator for tiny TokenSym in table_ident */
134 #define TOKSTR_TAL_SIZE (768 * 1024) /* allocator for tiny TokenString instances */
135 #define CSTR_TAL_SIZE (256 * 1024) /* allocator for tiny CString instances */
136 #define TOKSYM_TAL_LIMIT 256 /* prefer unique limits to distinguish allocators debug msgs */
137 #define TOKSTR_TAL_LIMIT 128 /* 32 * sizeof(int) */
138 #define CSTR_TAL_LIMIT 1024
140 typedef struct TinyAlloc
{
146 struct TinyAlloc
*next
, *top
;
155 typedef struct tal_header_t
{
158 int line_num
; /* negative line_num used for double free check */
159 char file_name
[TAL_DEBUG_FILE_LEN
+ 1];
163 /* ------------------------------------------------------------------------- */
165 static TinyAlloc
*tal_new(TinyAlloc
**pal
, unsigned limit
, unsigned size
)
167 TinyAlloc
*al
= tcc_mallocz(sizeof(TinyAlloc
));
168 al
->p
= al
->buffer
= tcc_malloc(size
);
175 static void tal_delete(TinyAlloc
*al
)
183 fprintf(stderr
, "limit=%5d, size=%5g MB, nb_peak=%6d, nb_total=%8d, nb_missed=%6d, usage=%5.1f%%\n",
184 al
->limit
, al
->size
/ 1024.0 / 1024.0, al
->nb_peak
, al
->nb_total
, al
->nb_missed
,
185 (al
->peak_p
- al
->buffer
) * 100.0 / al
->size
);
188 if (al
->nb_allocs
> 0) {
190 fprintf(stderr
, "TAL_DEBUG: memory leak %d chunk(s) (limit= %d)\n",
191 al
->nb_allocs
, al
->limit
);
194 tal_header_t
*header
= (tal_header_t
*)p
;
195 if (header
->line_num
> 0) {
196 fprintf(stderr
, "%s:%d: chunk of %d bytes leaked\n",
197 header
->file_name
, header
->line_num
, header
->size
);
199 p
+= header
->size
+ sizeof(tal_header_t
);
207 tcc_free(al
->buffer
);
213 static void tal_free_impl(TinyAlloc
*al
, void *p TAL_DEBUG_PARAMS
)
218 if (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
) {
220 tal_header_t
*header
= (((tal_header_t
*)p
) - 1);
221 if (header
->line_num
< 0) {
222 fprintf(stderr
, "%s:%d: TAL_DEBUG: double frees chunk from\n",
224 fprintf(stderr
, "%s:%d: %d bytes\n",
225 header
->file_name
, (int)-header
->line_num
, (int)header
->size
);
227 header
->line_num
= -header
->line_num
;
232 } else if (al
->next
) {
240 static void *tal_realloc_impl(TinyAlloc
**pal
, void *p
, unsigned size TAL_DEBUG_PARAMS
)
242 tal_header_t
*header
;
245 unsigned adj_size
= (size
+ 3) & -4;
246 TinyAlloc
*al
= *pal
;
249 is_own
= (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
);
250 if ((!p
|| is_own
) && size
<= al
->limit
) {
251 if (al
->p
+ adj_size
+ sizeof(tal_header_t
) < al
->buffer
+ al
->size
) {
252 header
= (tal_header_t
*)al
->p
;
253 header
->size
= adj_size
;
255 { int ofs
= strlen(file
) - TAL_DEBUG_FILE_LEN
;
256 strncpy(header
->file_name
, file
+ (ofs
> 0 ? ofs
: 0), TAL_DEBUG_FILE_LEN
);
257 header
->file_name
[TAL_DEBUG_FILE_LEN
] = 0;
258 header
->line_num
= line
; }
260 ret
= al
->p
+ sizeof(tal_header_t
);
261 al
->p
+= adj_size
+ sizeof(tal_header_t
);
263 header
= (((tal_header_t
*)p
) - 1);
264 memcpy(ret
, p
, header
->size
);
266 header
->line_num
= -header
->line_num
;
272 if (al
->nb_peak
< al
->nb_allocs
)
273 al
->nb_peak
= al
->nb_allocs
;
274 if (al
->peak_p
< al
->p
)
281 ret
= tal_realloc(*pal
, 0, size
);
282 header
= (((tal_header_t
*)p
) - 1);
283 memcpy(ret
, p
, header
->size
);
285 header
->line_num
= -header
->line_num
;
292 TinyAlloc
*bottom
= al
, *next
= al
->top
? al
->top
: al
;
294 al
= tal_new(pal
, next
->limit
, next
->size
* 2);
302 ret
= tcc_malloc(size
);
303 header
= (((tal_header_t
*)p
) - 1);
304 memcpy(ret
, p
, header
->size
);
306 header
->line_num
= -header
->line_num
;
308 } else if (al
->next
) {
312 ret
= tcc_realloc(p
, size
);
321 /* ------------------------------------------------------------------------- */
322 /* CString handling */
323 static void cstr_realloc(CString
*cstr
, int new_size
)
327 size
= cstr
->size_allocated
;
329 size
= 8; /* no need to allocate a too small first string */
330 while (size
< new_size
)
332 cstr
->data
= tal_realloc(cstr_alloc
, cstr
->data
, size
);
333 cstr
->size_allocated
= size
;
337 ST_INLN
void cstr_ccat(CString
*cstr
, int ch
)
340 size
= cstr
->size
+ 1;
341 if (size
> cstr
->size_allocated
)
342 cstr_realloc(cstr
, size
);
343 ((unsigned char *)cstr
->data
)[size
- 1] = ch
;
347 ST_FUNC
void cstr_cat(CString
*cstr
, const char *str
, int len
)
351 len
= strlen(str
) + 1 + len
;
352 size
= cstr
->size
+ len
;
353 if (size
> cstr
->size_allocated
)
354 cstr_realloc(cstr
, size
);
355 memmove(((unsigned char *)cstr
->data
) + cstr
->size
, str
, len
);
359 /* add a wide char */
360 ST_FUNC
void cstr_wccat(CString
*cstr
, int ch
)
363 size
= cstr
->size
+ sizeof(nwchar_t
);
364 if (size
> cstr
->size_allocated
)
365 cstr_realloc(cstr
, size
);
366 *(nwchar_t
*)(((unsigned char *)cstr
->data
) + size
- sizeof(nwchar_t
)) = ch
;
370 ST_FUNC
void cstr_new(CString
*cstr
)
372 memset(cstr
, 0, sizeof(CString
));
375 /* free string and reset it to NULL */
376 ST_FUNC
void cstr_free(CString
*cstr
)
378 tal_free(cstr_alloc
, cstr
->data
);
382 /* reset string to empty */
383 ST_FUNC
void cstr_reset(CString
*cstr
)
389 static void add_char(CString
*cstr
, int c
)
391 if (c
== '\'' || c
== '\"' || c
== '\\') {
392 /* XXX: could be more precise if char or string */
393 cstr_ccat(cstr
, '\\');
395 if (c
>= 32 && c
<= 126) {
398 cstr_ccat(cstr
, '\\');
400 cstr_ccat(cstr
, 'n');
402 cstr_ccat(cstr
, '0' + ((c
>> 6) & 7));
403 cstr_ccat(cstr
, '0' + ((c
>> 3) & 7));
404 cstr_ccat(cstr
, '0' + (c
& 7));
409 /* ------------------------------------------------------------------------- */
410 /* allocate a new token */
411 static TokenSym
*tok_alloc_new(TokenSym
**pts
, const char *str
, int len
)
413 TokenSym
*ts
, **ptable
;
416 if (tok_ident
>= SYM_FIRST_ANOM
)
417 tcc_error("memory full (symbols)");
419 /* expand token table if needed */
420 i
= tok_ident
- TOK_IDENT
;
421 if ((i
% TOK_ALLOC_INCR
) == 0) {
422 ptable
= tcc_realloc(table_ident
, (i
+ TOK_ALLOC_INCR
) * sizeof(TokenSym
*));
423 table_ident
= ptable
;
426 ts
= tal_realloc(toksym_alloc
, 0, sizeof(TokenSym
) + len
);
428 ts
->tok
= tok_ident
++;
429 ts
->sym_define
= NULL
;
430 ts
->sym_label
= NULL
;
431 ts
->sym_struct
= NULL
;
432 ts
->sym_identifier
= NULL
;
434 ts
->hash_next
= NULL
;
435 memcpy(ts
->str
, str
, len
);
441 #define TOK_HASH_INIT 1
442 #define TOK_HASH_FUNC(h, c) ((h) + ((h) << 5) + ((h) >> 27) + (c))
445 /* find a token and add it if not found */
446 ST_FUNC TokenSym
*tok_alloc(const char *str
, int len
)
454 h
= TOK_HASH_FUNC(h
, ((unsigned char *)str
)[i
]);
455 h
&= (TOK_HASH_SIZE
- 1);
457 pts
= &hash_ident
[h
];
462 if (ts
->len
== len
&& !memcmp(ts
->str
, str
, len
))
464 pts
= &(ts
->hash_next
);
466 return tok_alloc_new(pts
, str
, len
);
469 /* XXX: buffer overflow */
470 /* XXX: float tokens */
471 ST_FUNC
const char *get_tok_str(int v
, CValue
*cv
)
476 cstr_reset(&cstr_buf
);
484 /* XXX: not quite exact, but only useful for testing */
486 sprintf(p
, "%u", (unsigned)cv
->i
);
488 sprintf(p
, "%llu", (unsigned long long)cv
->i
);
492 cstr_ccat(&cstr_buf
, 'L');
494 cstr_ccat(&cstr_buf
, '\'');
495 add_char(&cstr_buf
, cv
->i
);
496 cstr_ccat(&cstr_buf
, '\'');
497 cstr_ccat(&cstr_buf
, '\0');
501 return (char*)cv
->str
.data
;
503 cstr_ccat(&cstr_buf
, 'L');
505 cstr_ccat(&cstr_buf
, '\"');
507 len
= cv
->str
.size
- 1;
509 add_char(&cstr_buf
, ((unsigned char *)cv
->str
.data
)[i
]);
511 len
= (cv
->str
.size
/ sizeof(nwchar_t
)) - 1;
513 add_char(&cstr_buf
, ((nwchar_t
*)cv
->str
.data
)[i
]);
515 cstr_ccat(&cstr_buf
, '\"');
516 cstr_ccat(&cstr_buf
, '\0');
520 cstr_cat(&cstr_buf
, "<float>", 0);
523 cstr_cat(&cstr_buf
, "<double>", 0);
526 cstr_cat(&cstr_buf
, "<long double>", 0);
529 cstr_cat(&cstr_buf
, "<linenumber>", 0);
532 /* above tokens have value, the ones below don't */
541 return strcpy(p
, "...");
543 return strcpy(p
, "<<=");
545 return strcpy(p
, ">>=");
548 /* search in two bytes table */
549 const unsigned char *q
= tok_two_chars
;
555 return cstr_buf
.data
;
560 sprintf(cstr_buf
.data
, "<%02x>", v
);
561 return cstr_buf
.data
;
566 } else if (v
< tok_ident
) {
567 return table_ident
[v
- TOK_IDENT
]->str
;
568 } else if (v
>= SYM_FIRST_ANOM
) {
569 /* special name for anonymous symbol */
570 sprintf(p
, "L.%u", v
- SYM_FIRST_ANOM
);
572 /* should never happen */
577 return cstr_buf
.data
;
580 /* return the current character, handling end of block if necessary
582 ST_FUNC
int handle_eob(void)
584 BufferedFile
*bf
= file
;
587 /* only tries to read if really end of buffer */
588 if (bf
->buf_ptr
>= bf
->buf_end
) {
590 #if defined(PARSE_DEBUG)
595 len
= read(bf
->fd
, bf
->buffer
, len
);
602 bf
->buf_ptr
= bf
->buffer
;
603 bf
->buf_end
= bf
->buffer
+ len
;
604 *bf
->buf_end
= CH_EOB
;
606 if (bf
->buf_ptr
< bf
->buf_end
) {
607 return bf
->buf_ptr
[0];
609 bf
->buf_ptr
= bf
->buf_end
;
614 /* read next char from current input file and handle end of input buffer */
615 ST_INLN
void inp(void)
617 ch
= *(++(file
->buf_ptr
));
618 /* end of buffer/file handling */
623 /* handle '\[\r]\n' */
624 static int handle_stray_noerror(void)
631 } else if (ch
== '\r') {
645 static void handle_stray(void)
647 if (handle_stray_noerror())
648 tcc_error("stray '\\' in program");
651 /* skip the stray and handle the \\n case. Output an error if
652 incorrect char after the stray */
653 static int handle_stray1(uint8_t *p
)
658 if (p
>= file
->buf_end
) {
665 if (handle_stray_noerror()) {
666 if (!(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
667 tcc_error("stray '\\' in program");
668 *--file
->buf_ptr
= '\\';
675 /* handle just the EOB case, but not stray */
676 #define PEEKC_EOB(c, p)\
687 /* handle the complicated stray case */
693 c = handle_stray1(p);\
698 /* input with '\[\r]\n' handling. Note that this function cannot
699 handle other characters after '\', so you cannot call it inside
700 strings or comments */
701 ST_FUNC
void minp(void)
708 /* single line C++ comments */
709 static uint8_t *parse_line_comment(uint8_t *p
)
717 if (c
== '\n' || c
== CH_EOF
) {
719 } else if (c
== '\\') {
728 } else if (c
== '\r') {
746 ST_FUNC
uint8_t *parse_comment(uint8_t *p
)
755 if (c
== '\n' || c
== '*' || c
== '\\')
759 if (c
== '\n' || c
== '*' || c
== '\\')
763 /* now we can handle all the cases */
767 } else if (c
== '*') {
773 } else if (c
== '/') {
775 } else if (c
== '\\') {
780 tcc_error("unexpected end of file in comment");
782 /* skip '\[\r]\n', otherwise just skip the stray */
788 } else if (c
== '\r') {
805 /* stray, eob or eof */
810 tcc_error("unexpected end of file in comment");
811 } else if (c
== '\\') {
821 ST_FUNC
void set_idnum(int c
, int val
)
823 isidnum_table
[c
- CH_EOF
] = val
;
828 static inline void skip_spaces(void)
830 while (isidnum_table
[ch
- CH_EOF
] & IS_SPC
)
834 static inline int check_space(int t
, int *spc
)
836 if (t
< 256 && (isidnum_table
[t
- CH_EOF
] & IS_SPC
)) {
845 /* parse a string without interpreting escapes */
846 static uint8_t *parse_pp_string(uint8_t *p
,
847 int sep
, CString
*str
)
855 } else if (c
== '\\') {
861 /* XXX: indicate line number of start of string */
862 tcc_error("missing terminating %c character", sep
);
863 } else if (c
== '\\') {
864 /* escape : just skip \[\r]\n */
869 } else if (c
== '\r') {
872 expect("'\n' after '\r'");
875 } else if (c
== CH_EOF
) {
876 goto unterminated_string
;
879 cstr_ccat(str
, '\\');
885 } else if (c
== '\n') {
888 } else if (c
== '\r') {
892 cstr_ccat(str
, '\r');
908 /* skip block of text until #else, #elif or #endif. skip also pairs of
910 static void preprocess_skip(void)
912 int a
, start_of_line
, c
, in_warn_or_error
;
919 in_warn_or_error
= 0;
940 } else if (c
== '\\') {
941 ch
= file
->buf_ptr
[0];
942 handle_stray_noerror();
949 if (in_warn_or_error
)
951 p
= parse_pp_string(p
, c
, NULL
);
955 if (in_warn_or_error
)
962 p
= parse_comment(p
);
963 } else if (ch
== '/') {
964 p
= parse_line_comment(p
);
974 (tok
== TOK_ELSE
|| tok
== TOK_ELIF
|| tok
== TOK_ENDIF
))
976 if (tok
== TOK_IF
|| tok
== TOK_IFDEF
|| tok
== TOK_IFNDEF
)
978 else if (tok
== TOK_ENDIF
)
980 else if( tok
== TOK_ERROR
|| tok
== TOK_WARNING
)
981 in_warn_or_error
= 1;
982 else if (tok
== TOK_LINEFEED
)
984 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
985 p
= parse_line_comment(p
- 1);
986 } else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
987 p
= parse_line_comment(p
- 1);
1000 /* ParseState handling */
1002 /* XXX: currently, no include file info is stored. Thus, we cannot display
1003 accurate messages if the function or data definition spans multiple
1006 /* save current parse state in 's' */
1007 ST_FUNC
void save_parse_state(ParseState
*s
)
1009 s
->line_num
= file
->line_num
;
1010 s
->macro_ptr
= macro_ptr
;
1015 /* restore parse state from 's' */
1016 ST_FUNC
void restore_parse_state(ParseState
*s
)
1018 file
->line_num
= s
->line_num
;
1019 macro_ptr
= s
->macro_ptr
;
1024 /* return the number of additional 'ints' necessary to store the
1026 static inline int tok_size(const int *p
)
1041 return 1 + ((sizeof(CString
) + ((CString
*)(p
+1))->size
+ 3) >> 2);
1047 return 1 + LDOUBLE_SIZE
/ 4;
1053 /* token string handling */
1055 ST_INLN
void tok_str_new(TokenString
*s
)
1059 s
->allocated_len
= 0;
1060 s
->last_line_num
= -1;
1063 ST_FUNC TokenString
*tok_str_alloc(void)
1065 TokenString
*str
= tal_realloc(tokstr_alloc
, 0, sizeof *str
);
1070 ST_FUNC
int *tok_str_dup(TokenString
*s
)
1074 str
= tal_realloc(tokstr_alloc
, 0, s
->len
* sizeof(int));
1075 memcpy(str
, s
->str
, s
->len
* sizeof(int));
1079 ST_FUNC
void tok_str_free_str(int *str
)
1081 tal_free(tokstr_alloc
, str
);
1084 ST_FUNC
void tok_str_free(TokenString
*str
)
1086 tok_str_free_str(str
->str
);
1087 tal_free(tokstr_alloc
, str
);
1090 ST_FUNC
int *tok_str_realloc(TokenString
*s
, int new_size
)
1094 size
= s
->allocated_len
;
1097 while (size
< new_size
)
1099 if (size
> s
->allocated_len
) {
1100 str
= tal_realloc(tokstr_alloc
, s
->str
, size
* sizeof(int));
1101 s
->allocated_len
= size
;
1107 ST_FUNC
void tok_str_add(TokenString
*s
, int t
)
1113 if (len
>= s
->allocated_len
)
1114 str
= tok_str_realloc(s
, len
+ 1);
1119 ST_FUNC
void begin_macro(TokenString
*str
, int alloc
)
1122 str
->prev
= macro_stack
;
1123 str
->prev_ptr
= macro_ptr
;
1124 macro_ptr
= str
->str
;
1128 ST_FUNC
void end_macro(void)
1130 TokenString
*str
= macro_stack
;
1131 macro_stack
= str
->prev
;
1132 macro_ptr
= str
->prev_ptr
;
1133 if (str
->alloc
== 2) {
1134 str
->alloc
= 3; /* just mark as finished */
1140 static void tok_str_add2(TokenString
*s
, int t
, CValue
*cv
)
1147 /* allocate space for worst case */
1148 if (len
+ TOK_MAX_SIZE
>= s
->allocated_len
)
1149 str
= tok_str_realloc(s
, len
+ TOK_MAX_SIZE
+ 1);
1158 str
[len
++] = cv
->tab
[0];
1165 /* Insert the string into the int array. */
1167 1 + (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1168 if (len
+ nb_words
>= s
->allocated_len
)
1169 str
= tok_str_realloc(s
, len
+ nb_words
+ 1);
1170 str
[len
] = cv
->str
.size
;
1171 memcpy(&str
[len
+ 1], cv
->str
.data
, cv
->str
.size
);
1178 #if LDOUBLE_SIZE == 8
1181 str
[len
++] = cv
->tab
[0];
1182 str
[len
++] = cv
->tab
[1];
1184 #if LDOUBLE_SIZE == 12
1186 str
[len
++] = cv
->tab
[0];
1187 str
[len
++] = cv
->tab
[1];
1188 str
[len
++] = cv
->tab
[2];
1189 #elif LDOUBLE_SIZE == 16
1191 str
[len
++] = cv
->tab
[0];
1192 str
[len
++] = cv
->tab
[1];
1193 str
[len
++] = cv
->tab
[2];
1194 str
[len
++] = cv
->tab
[3];
1195 #elif LDOUBLE_SIZE != 8
1196 #error add long double size support
1205 /* add the current parse token in token string 's' */
1206 ST_FUNC
void tok_str_add_tok(TokenString
*s
)
1210 /* save line number info */
1211 if (file
->line_num
!= s
->last_line_num
) {
1212 s
->last_line_num
= file
->line_num
;
1213 cval
.i
= s
->last_line_num
;
1214 tok_str_add2(s
, TOK_LINENUM
, &cval
);
1216 tok_str_add2(s
, tok
, &tokc
);
1219 /* get a token from an integer array and increment pointer
1220 accordingly. we code it as a macro to avoid pointer aliasing. */
1221 static inline void TOK_GET(int *t
, const int **pp
, CValue
*cv
)
1240 cv
->str
.size
= *p
++;
1242 p
+= (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1250 #if LDOUBLE_SIZE == 16
1252 #elif LDOUBLE_SIZE == 12
1254 #elif LDOUBLE_SIZE == 8
1257 # error add long double size support
1270 /* Calling this function is expensive, but it is not possible
1271 to read a token string backwards. */
1272 static int tok_last(const int *str0
, const int *str1
)
1274 const int *str
= str0
;
1279 TOK_GET(&tok
, &str
, &cval
);
1283 static int macro_is_equal(const int *a
, const int *b
)
1292 /* first time preallocate macro_equal_buf, next time only reset position to start */
1293 cstr_reset(¯o_equal_buf
);
1294 TOK_GET(&t
, &a
, &cv
);
1295 cstr_cat(¯o_equal_buf
, get_tok_str(t
, &cv
), 0);
1296 TOK_GET(&t
, &b
, &cv
);
1297 if (strcmp(macro_equal_buf
.data
, get_tok_str(t
, &cv
)))
1303 /* defines handling */
1304 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1309 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1311 s
->next
= first_arg
;
1312 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1314 if (o
&& !macro_is_equal(o
->d
, s
->d
))
1315 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1318 /* undefined a define symbol. Its name is just set to zero */
1319 ST_FUNC
void define_undef(Sym
*s
)
1322 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1323 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1326 ST_INLN Sym
*define_find(int v
)
1329 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1331 return table_ident
[v
]->sym_define
;
1334 /* free define stack until top reaches 'b' */
1335 ST_FUNC
void free_defines(Sym
*b
)
1337 while (define_stack
!= b
) {
1338 Sym
*top
= define_stack
;
1339 define_stack
= top
->prev
;
1340 tok_str_free_str(top
->d
);
1345 /* restore remaining (-D or predefined) symbols */
1348 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1349 Sym
**d
= &table_ident
[v
- TOK_IDENT
]->sym_define
;
1358 ST_FUNC Sym
*label_find(int v
)
1361 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1363 return table_ident
[v
]->sym_label
;
1366 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1369 s
= sym_push2(ptop
, v
, 0, 0);
1371 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1372 if (ptop
== &global_label_stack
) {
1373 /* modify the top most local identifier, so that
1374 sym_identifier will point to 's' when popped */
1376 ps
= &(*ps
)->prev_tok
;
1383 /* pop labels until element last is reached. Look if any labels are
1384 undefined. Define symbols if '&&label' was used. */
1385 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
)
1388 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1390 if (s
->r
== LABEL_DECLARED
) {
1391 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1392 } else if (s
->r
== LABEL_FORWARD
) {
1393 tcc_error("label '%s' used but not defined",
1394 get_tok_str(s
->v
, NULL
));
1397 /* define corresponding symbol. A size of
1399 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1403 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1409 /* eval an expression for #if/#elif */
1410 static int expr_preprocess(void)
1415 str
= tok_str_alloc();
1416 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1417 next(); /* do macro subst */
1418 if (tok
== TOK_DEFINED
) {
1423 c
= define_find(tok
) != 0;
1428 } else if (tok
>= TOK_IDENT
) {
1429 /* if undefined macro */
1433 tok_str_add_tok(str
);
1435 tok_str_add(str
, -1); /* simulate end of file */
1436 tok_str_add(str
, 0);
1437 /* now evaluate C constant expression */
1438 begin_macro(str
, 1);
1446 /* parse after #define */
1447 ST_FUNC
void parse_define(void)
1449 Sym
*s
, *first
, **ps
;
1450 int v
, t
, varg
, is_vaargs
, spc
;
1451 int saved_parse_flags
= parse_flags
;
1455 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1456 /* XXX: should check if same macro (ANSI) */
1459 /* We have to parse the whole define as if not in asm mode, in particular
1460 no line comment with '#' must be ignored. Also for function
1461 macros the argument list must be parsed without '.' being an ID
1463 parse_flags
= ((parse_flags
& ~PARSE_FLAG_ASM_FILE
) | PARSE_FLAG_SPACES
);
1464 /* '(' must be just after macro definition for MACRO_FUNC */
1470 if (tok
!= ')') for (;;) {
1474 if (varg
== TOK_DOTS
) {
1475 varg
= TOK___VA_ARGS__
;
1477 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1481 if (varg
< TOK_IDENT
)
1483 tcc_error("bad macro parameter list");
1484 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1489 if (tok
!= ',' || is_vaargs
)
1499 parse_flags
|= PARSE_FLAG_ACCEPT_STRAYS
| PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
;
1500 /* The body of a macro definition should be parsed such that identifiers
1501 are parsed like the file mode determines (i.e. with '.' being an
1502 ID character in asm mode). But '#' should be retained instead of
1503 regarded as line comment leader, so still don't set ASM_FILE
1505 set_idnum('.', (saved_parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0);
1506 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1507 /* remove spaces around ## and after '#' */
1508 if (TOK_TWOSHARPS
== tok
) {
1515 } else if ('#' == tok
) {
1517 } else if (check_space(tok
, &spc
)) {
1520 tok_str_add2(&tokstr_buf
, tok
, &tokc
);
1525 parse_flags
= saved_parse_flags
;
1527 --tokstr_buf
.len
; /* remove trailing space */
1528 tok_str_add(&tokstr_buf
, 0);
1531 tcc_error("'##' cannot appear at either end of macro");
1532 define_push(v
, t
, tok_str_dup(&tokstr_buf
), first
);
1535 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
, int add
)
1537 const unsigned char *s
;
1543 s
= (unsigned char *) filename
;
1546 h
= TOK_HASH_FUNC(h
, toup(*s
));
1548 h
= TOK_HASH_FUNC(h
, *s
);
1552 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1554 i
= s1
->cached_includes_hash
[h
];
1558 e
= s1
->cached_includes
[i
- 1];
1559 if (0 == PATHCMP(e
->filename
, filename
))
1566 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1567 strcpy(e
->filename
, filename
);
1568 e
->ifndef_macro
= e
->once
= 0;
1569 dynarray_add((void ***)&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1570 /* add in hash table */
1571 e
->hash_next
= s1
->cached_includes_hash
[h
];
1572 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1574 printf("adding cached '%s'\n", filename
);
1579 static void pragma_parse(TCCState
*s1
)
1582 if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1586 if (next(), tok
!= '(')
1588 if (next(), tok
!= TOK_STR
)
1590 v
= tok_alloc(tokc
.str
.data
, tokc
.str
.size
- 1)->tok
;
1591 if (next(), tok
!= ')')
1593 if (t
== TOK_push_macro
) {
1594 while (NULL
== (s
= define_find(v
)))
1595 define_push(v
, 0, NULL
, NULL
);
1596 s
->type
.ref
= s
; /* set push boundary */
1598 for (s
= define_stack
; s
; s
= s
->prev
)
1599 if (s
->v
== v
&& s
->type
.ref
== s
) {
1605 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1607 tcc_warning("unbalanced #pragma pop_macro");
1608 pp_debug_tok
= t
, pp_debug_symv
= v
;
1610 } else if (tok
== TOK_once
) {
1611 search_cached_include(s1
, file
->filename
, 1)->once
= pp_once
;
1613 } else if (s1
->ppfp
) {
1614 /* tcc -E: keep pragmas below unchanged */
1616 unget_tok(TOK_PRAGMA
);
1618 unget_tok(TOK_LINEFEED
);
1620 } else if (tok
== TOK_pack
) {
1622 #pragma pack(1) // set
1623 #pragma pack() // reset to default
1624 #pragma pack(push,1) // push & set
1625 #pragma pack(pop) // restore previous */
1628 if (tok
== TOK_ASM_pop
) {
1630 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1632 tcc_error("out of pack stack");
1634 s1
->pack_stack_ptr
--;
1638 if (tok
== TOK_ASM_push
) {
1640 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1642 s1
->pack_stack_ptr
++;
1645 if (tok
!= TOK_CINT
)
1648 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1652 *s1
->pack_stack_ptr
= val
;
1657 } else if (tok
== TOK_comment
) {
1667 file
= tcc_strdup((char *)tokc
.str
.data
);
1668 dynarray_add((void ***)&s1
->pragma_libs
, &s1
->nb_pragma_libs
, file
);
1674 if (s1
->warn_unsupported
)
1675 tcc_warning("#pragma %s is ignored", get_tok_str(tok
, &tokc
));
1680 tcc_error("malformed #pragma directive");
1684 /* is_bof is true if first non space token at beginning of file */
1685 ST_FUNC
void preprocess(int is_bof
)
1687 TCCState
*s1
= tcc_state
;
1688 int i
, c
, n
, saved_parse_flags
;
1692 saved_parse_flags
= parse_flags
;
1693 parse_flags
= PARSE_FLAG_PREPROCESS
1694 | PARSE_FLAG_TOK_NUM
1695 | PARSE_FLAG_TOK_STR
1696 | PARSE_FLAG_LINEFEED
1697 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
1706 pp_debug_symv
= tok
;
1712 pp_debug_symv
= tok
;
1713 s
= define_find(tok
);
1714 /* undefine symbol by putting an invalid name */
1719 case TOK_INCLUDE_NEXT
:
1720 ch
= file
->buf_ptr
[0];
1721 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1726 } else if (ch
== '\"') {
1731 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1732 if ((q
- buf
) < sizeof(buf
) - 1)
1735 if (handle_stray_noerror() == 0)
1743 /* eat all spaces and comments after include */
1744 /* XXX: slightly incorrect */
1745 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1750 /* computed #include : concatenate everything up to linefeed,
1751 the result must be one of the two accepted forms.
1752 Don't convert pp-tokens to tokens here. */
1753 parse_flags
= (PARSE_FLAG_PREPROCESS
1754 | PARSE_FLAG_LINEFEED
1755 | (parse_flags
& PARSE_FLAG_ASM_FILE
));
1758 while (tok
!= TOK_LINEFEED
) {
1759 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1763 /* check syntax and remove '<>|""' */
1764 if ((len
< 2 || ((buf
[0] != '"' || buf
[len
-1] != '"') &&
1765 (buf
[0] != '<' || buf
[len
-1] != '>'))))
1766 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1768 memmove(buf
, buf
+ 1, len
- 2);
1769 buf
[len
- 2] = '\0';
1772 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1773 tcc_error("#include recursion too deep");
1774 /* store current file in stack, but increment stack later below */
1775 *s1
->include_stack_ptr
= file
;
1776 i
= tok
== TOK_INCLUDE_NEXT
? file
->include_next_index
: 0;
1777 n
= 2 + s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1778 for (; i
< n
; ++i
) {
1779 char buf1
[sizeof file
->filename
];
1784 /* check absolute include path */
1785 if (!IS_ABSPATH(buf
))
1789 } else if (i
== 1) {
1790 /* search in file's dir if "header.h" */
1793 path
= file
->filename
;
1794 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1797 /* search in all the include paths */
1798 int j
= i
- 2, k
= j
- s1
->nb_include_paths
;
1799 path
= k
< 0 ? s1
->include_paths
[j
] : s1
->sysinclude_paths
[k
];
1800 pstrcpy(buf1
, sizeof(buf1
), path
);
1801 pstrcat(buf1
, sizeof(buf1
), "/");
1804 pstrcat(buf1
, sizeof(buf1
), buf
);
1805 e
= search_cached_include(s1
, buf1
, 0);
1806 if (e
&& (define_find(e
->ifndef_macro
) || e
->once
== pp_once
)) {
1807 /* no need to parse the include because the 'ifndef macro'
1808 is defined (or had #pragma once) */
1810 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1815 if (tcc_open(s1
, buf1
) < 0)
1818 file
->include_next_index
= i
+ 1;
1820 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1822 /* update target deps */
1823 dynarray_add((void ***)&s1
->target_deps
, &s1
->nb_target_deps
,
1825 /* push current file in stack */
1826 ++s1
->include_stack_ptr
;
1827 /* add include file debug info */
1829 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1830 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1831 ch
= file
->buf_ptr
[0];
1834 tcc_error("include file '%s' not found", buf
);
1841 c
= expr_preprocess();
1847 if (tok
< TOK_IDENT
)
1848 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1852 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1854 file
->ifndef_macro
= tok
;
1857 c
= (define_find(tok
) != 0) ^ c
;
1859 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1860 tcc_error("memory full (ifdef)");
1861 *s1
->ifdef_stack_ptr
++ = c
;
1864 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1865 tcc_error("#else without matching #if");
1866 if (s1
->ifdef_stack_ptr
[-1] & 2)
1867 tcc_error("#else after #else");
1868 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1871 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1872 tcc_error("#elif without matching #if");
1873 c
= s1
->ifdef_stack_ptr
[-1];
1875 tcc_error("#elif after #else");
1876 /* last #if/#elif expression was true: we skip */
1880 c
= expr_preprocess();
1881 s1
->ifdef_stack_ptr
[-1] = c
;
1884 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1885 file
->ifndef_macro
= 0;
1894 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1895 tcc_error("#endif without matching #if");
1896 s1
->ifdef_stack_ptr
--;
1897 /* '#ifndef macro' was at the start of file. Now we check if
1898 an '#endif' is exactly at the end of file */
1899 if (file
->ifndef_macro
&&
1900 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1901 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1902 /* need to set to zero to avoid false matches if another
1903 #ifndef at middle of file */
1904 file
->ifndef_macro
= 0;
1905 while (tok
!= TOK_LINEFEED
)
1907 tok_flags
|= TOK_FLAG_ENDIF
;
1912 n
= strtoul((char*)tokc
.str
.data
, &q
, 10);
1916 if (tok
!= TOK_CINT
)
1918 tcc_error("wrong #line format");
1922 if (tok
!= TOK_LINEFEED
) {
1924 pstrcpy(file
->filename
, sizeof(file
->filename
), (char *)tokc
.str
.data
);
1925 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
1932 total_lines
+= file
->line_num
- n
;
1935 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1940 ch
= file
->buf_ptr
[0];
1943 while (ch
!= '\n' && ch
!= CH_EOF
) {
1944 if ((q
- buf
) < sizeof(buf
) - 1)
1947 if (handle_stray_noerror() == 0)
1954 tcc_error("#error %s", buf
);
1956 tcc_warning("#warning %s", buf
);
1964 /* ignore gas line comment in an 'S' file. */
1965 if (saved_parse_flags
& PARSE_FLAG_ASM_FILE
)
1967 if (tok
== '!' && is_bof
)
1968 /* '!' is ignored at beginning to allow C scripts. */
1970 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
1972 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
- 1);
1975 /* ignore other preprocess commands or #! for C scripts */
1976 while (tok
!= TOK_LINEFEED
)
1979 parse_flags
= saved_parse_flags
;
1982 /* evaluate escape codes in a string. */
1983 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
1998 case '0': case '1': case '2': case '3':
1999 case '4': case '5': case '6': case '7':
2000 /* at most three octal digits */
2005 n
= n
* 8 + c
- '0';
2009 n
= n
* 8 + c
- '0';
2014 goto add_char_nonext
;
2022 if (c
>= 'a' && c
<= 'f')
2024 else if (c
>= 'A' && c
<= 'F')
2034 goto add_char_nonext
;
2058 goto invalid_escape
;
2068 if (c
>= '!' && c
<= '~')
2069 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
2071 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
2078 cstr_ccat(outstr
, c
);
2080 cstr_wccat(outstr
, c
);
2082 /* add a trailing '\0' */
2084 cstr_ccat(outstr
, '\0');
2086 cstr_wccat(outstr
, '\0');
2089 static void parse_string(const char *s
, int len
)
2091 uint8_t buf
[1000], *p
= buf
;
2094 if ((is_long
= *s
== 'L'))
2098 if (len
>= sizeof buf
)
2099 p
= tcc_malloc(len
+ 1);
2103 cstr_reset(&tokcstr
);
2104 parse_escape_string(&tokcstr
, p
, is_long
);
2110 /* XXX: make it portable */
2114 char_size
= sizeof(nwchar_t
);
2115 if (tokcstr
.size
<= char_size
)
2116 tcc_error("empty character constant");
2117 if (tokcstr
.size
> 2 * char_size
)
2118 tcc_warning("multi-character character constant");
2120 tokc
.i
= *(int8_t *)tokcstr
.data
;
2123 tokc
.i
= *(nwchar_t
*)tokcstr
.data
;
2127 tokc
.str
.size
= tokcstr
.size
;
2128 tokc
.str
.data
= tokcstr
.data
;
2136 /* we use 64 bit numbers */
2139 /* bn = (bn << shift) | or_val */
2140 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
2144 for(i
=0;i
<BN_SIZE
;i
++) {
2146 bn
[i
] = (v
<< shift
) | or_val
;
2147 or_val
= v
>> (32 - shift
);
2151 static void bn_zero(unsigned int *bn
)
2154 for(i
=0;i
<BN_SIZE
;i
++) {
2159 /* parse number in null terminated string 'p' and return it in the
2161 static void parse_number(const char *p
)
2163 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
2165 unsigned int bn
[BN_SIZE
];
2176 goto float_frac_parse
;
2177 } else if (t
== '0') {
2178 if (ch
== 'x' || ch
== 'X') {
2182 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
2188 /* parse all digits. cannot check octal numbers at this stage
2189 because of floating point constants */
2191 if (ch
>= 'a' && ch
<= 'f')
2193 else if (ch
>= 'A' && ch
<= 'F')
2201 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
2203 tcc_error("number too long");
2209 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
2210 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
2212 /* NOTE: strtox should support that for hexa numbers, but
2213 non ISOC99 libcs do not support it, so we prefer to do
2215 /* hexadecimal or binary floats */
2216 /* XXX: handle overflows */
2228 } else if (t
>= 'a') {
2230 } else if (t
>= 'A') {
2235 bn_lshift(bn
, shift
, t
);
2242 if (t
>= 'a' && t
<= 'f') {
2244 } else if (t
>= 'A' && t
<= 'F') {
2246 } else if (t
>= '0' && t
<= '9') {
2252 tcc_error("invalid digit");
2253 bn_lshift(bn
, shift
, t
);
2258 if (ch
!= 'p' && ch
!= 'P')
2265 } else if (ch
== '-') {
2269 if (ch
< '0' || ch
> '9')
2270 expect("exponent digits");
2271 while (ch
>= '0' && ch
<= '9') {
2272 exp_val
= exp_val
* 10 + ch
- '0';
2275 exp_val
= exp_val
* s
;
2277 /* now we can generate the number */
2278 /* XXX: should patch directly float number */
2279 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2280 d
= ldexp(d
, exp_val
- frac_bits
);
2285 /* float : should handle overflow */
2287 } else if (t
== 'L') {
2289 #ifdef TCC_TARGET_PE
2294 /* XXX: not large enough */
2295 tokc
.ld
= (long double)d
;
2302 /* decimal floats */
2304 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2309 while (ch
>= '0' && ch
<= '9') {
2310 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2316 if (ch
== 'e' || ch
== 'E') {
2317 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2321 if (ch
== '-' || ch
== '+') {
2322 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2327 if (ch
< '0' || ch
> '9')
2328 expect("exponent digits");
2329 while (ch
>= '0' && ch
<= '9') {
2330 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2342 tokc
.f
= strtof(token_buf
, NULL
);
2343 } else if (t
== 'L') {
2345 #ifdef TCC_TARGET_PE
2347 tokc
.d
= strtod(token_buf
, NULL
);
2350 tokc
.ld
= strtold(token_buf
, NULL
);
2354 tokc
.d
= strtod(token_buf
, NULL
);
2358 unsigned long long n
, n1
;
2359 int lcount
, ucount
, must_64bit
;
2362 /* integer number */
2365 if (b
== 10 && *q
== '0') {
2372 /* no need for checks except for base 10 / 8 errors */
2382 tcc_error("invalid digit");
2385 /* detect overflow */
2386 /* XXX: this test is not reliable */
2388 tcc_error("integer constant overflow");
2391 /* Determine the characteristics (unsigned and/or 64bit) the type of
2392 the constant must have according to the constant suffix(es) */
2393 lcount
= ucount
= must_64bit
= 0;
2399 tcc_error("three 'l's in integer constant");
2400 if (lcount
&& *(p
- 1) != ch
)
2401 tcc_error("incorrect integer suffix: %s", p1
);
2403 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
2408 } else if (t
== 'U') {
2410 tcc_error("two 'u's in integer constant");
2418 /* Whether 64 bits are needed to hold the constant's value */
2419 if (n
& 0xffffffff00000000LL
|| must_64bit
) {
2427 /* Whether type must be unsigned to hold the constant's value */
2428 if (ucount
|| ((n1
>> 31) && (b
!= 10))) {
2429 if (tok
== TOK_CLLONG
)
2433 /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
2434 } else if (n1
>> 31) {
2435 if (tok
== TOK_CINT
)
2438 tcc_error("integer constant overflow");
2444 tcc_error("invalid number\n");
2448 #define PARSE2(c1, tok1, c2, tok2) \
2459 /* return next token without macro substitution */
2460 static inline void next_nomacro1(void)
2462 int t
, c
, is_long
, len
;
2475 if (parse_flags
& PARSE_FLAG_SPACES
)
2476 goto keep_tok_flags
;
2477 while (isidnum_table
[*p
- CH_EOF
] & IS_SPC
)
2486 /* first look if it is in fact an end of buffer */
2487 c
= handle_stray1(p
);
2494 TCCState
*s1
= tcc_state
;
2495 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2496 && !(tok_flags
& TOK_FLAG_EOF
)) {
2497 tok_flags
|= TOK_FLAG_EOF
;
2499 goto keep_tok_flags
;
2500 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2502 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2503 tcc_error("missing #endif");
2504 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2505 /* no include left : end of file. */
2508 tok_flags
&= ~TOK_FLAG_EOF
;
2509 /* pop include file */
2511 /* test if previous '#endif' was after a #ifdef at
2513 if (tok_flags
& TOK_FLAG_ENDIF
) {
2515 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2517 search_cached_include(s1
, file
->filename
, 1)
2518 ->ifndef_macro
= file
->ifndef_macro_saved
;
2519 tok_flags
&= ~TOK_FLAG_ENDIF
;
2522 /* add end of include file debug info */
2523 if (tcc_state
->do_debug
) {
2524 put_stabd(N_EINCL
, 0, 0);
2526 /* pop include stack */
2528 s1
->include_stack_ptr
--;
2537 tok_flags
|= TOK_FLAG_BOL
;
2540 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2543 goto keep_tok_flags
;
2548 if ((tok_flags
& TOK_FLAG_BOL
) &&
2549 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2551 preprocess(tok_flags
& TOK_FLAG_BOF
);
2557 tok
= TOK_TWOSHARPS
;
2559 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2560 p
= parse_line_comment(p
- 1);
2569 /* dollar is allowed to start identifiers when not parsing asm */
2571 if (!(isidnum_table
[c
- CH_EOF
] & IS_ID
)
2572 || (parse_flags
& PARSE_FLAG_ASM_FILE
))
2575 case 'a': case 'b': case 'c': case 'd':
2576 case 'e': case 'f': case 'g': case 'h':
2577 case 'i': case 'j': case 'k': case 'l':
2578 case 'm': case 'n': case 'o': case 'p':
2579 case 'q': case 'r': case 's': case 't':
2580 case 'u': case 'v': case 'w': case 'x':
2582 case 'A': case 'B': case 'C': case 'D':
2583 case 'E': case 'F': case 'G': case 'H':
2584 case 'I': case 'J': case 'K':
2585 case 'M': case 'N': case 'O': case 'P':
2586 case 'Q': case 'R': case 'S': case 'T':
2587 case 'U': case 'V': case 'W': case 'X':
2593 h
= TOK_HASH_FUNC(h
, c
);
2594 while (c
= *++p
, isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2595 h
= TOK_HASH_FUNC(h
, c
);
2600 /* fast case : no stray found, so we have the full token
2601 and we have already hashed it */
2602 h
&= (TOK_HASH_SIZE
- 1);
2603 pts
= &hash_ident
[h
];
2608 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2610 pts
= &(ts
->hash_next
);
2612 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2616 cstr_reset(&tokcstr
);
2617 cstr_cat(&tokcstr
, p1
, len
);
2621 while (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2623 cstr_ccat(&tokcstr
, c
);
2626 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2632 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2634 goto parse_ident_fast
;
2637 if (c
== '\'' || c
== '\"') {
2641 cstr_reset(&tokcstr
);
2642 cstr_ccat(&tokcstr
, 'L');
2643 goto parse_ident_slow
;
2648 case '0': case '1': case '2': case '3':
2649 case '4': case '5': case '6': case '7':
2653 /* after the first digit, accept digits, alpha, '.' or sign if
2654 prefixed by 'eEpP' */
2656 cstr_reset(&tokcstr
);
2658 cstr_ccat(&tokcstr
, t
);
2659 if (!((isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2661 || ((c
== '+' || c
== '-')
2662 && (((t
== 'e' || t
== 'E')
2663 && !(parse_flags
& PARSE_FLAG_ASM_FILE
2664 /* 0xe+1 is 3 tokens in asm */
2665 && ((char*)tokcstr
.data
)[0] == '0'
2666 && toup(((char*)tokcstr
.data
)[1]) == 'X'))
2667 || t
== 'p' || t
== 'P'))))
2672 /* We add a trailing '\0' to ease parsing */
2673 cstr_ccat(&tokcstr
, '\0');
2674 tokc
.str
.size
= tokcstr
.size
;
2675 tokc
.str
.data
= tokcstr
.data
;
2680 /* special dot handling because it can also start a number */
2685 } else if ((isidnum_table
['.' - CH_EOF
] & IS_ID
)
2686 && (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))) {
2688 goto parse_ident_fast
;
2689 } else if (c
== '.') {
2695 *--p
= '.'; /* may underflow into file->unget[] */
2706 cstr_reset(&tokcstr
);
2708 cstr_ccat(&tokcstr
, 'L');
2709 cstr_ccat(&tokcstr
, c
);
2710 p
= parse_pp_string(p
, c
, &tokcstr
);
2711 cstr_ccat(&tokcstr
, c
);
2712 cstr_ccat(&tokcstr
, '\0');
2713 tokc
.str
.size
= tokcstr
.size
;
2714 tokc
.str
.data
= tokcstr
.data
;
2723 } else if (c
== '<') {
2740 } else if (c
== '>') {
2758 } else if (c
== '=') {
2771 } else if (c
== '=') {
2784 } else if (c
== '=') {
2797 } else if (c
== '=') {
2800 } else if (c
== '>') {
2808 PARSE2('!', '!', '=', TOK_NE
)
2809 PARSE2('=', '=', '=', TOK_EQ
)
2810 PARSE2('*', '*', '=', TOK_A_MUL
)
2811 PARSE2('%', '%', '=', TOK_A_MOD
)
2812 PARSE2('^', '^', '=', TOK_A_XOR
)
2814 /* comments or operator */
2818 p
= parse_comment(p
);
2819 /* comments replaced by a blank */
2821 goto keep_tok_flags
;
2822 } else if (c
== '/') {
2823 p
= parse_line_comment(p
);
2825 goto keep_tok_flags
;
2826 } else if (c
== '=') {
2846 case '@': /* only used in assembler */
2852 if (c
>= 0x80 && c
<= 0xFF) /* utf8 identifiers */
2853 goto parse_ident_fast
;
2854 if (parse_flags
& PARSE_FLAG_ASM_FILE
)
2856 tcc_error("unrecognized character \\x%02x", c
);
2862 #if defined(PARSE_DEBUG)
2863 printf("token = %s\n", get_tok_str(tok
, &tokc
));
2867 /* return next token without macro substitution. Can read input from
2869 static void next_nomacro_spc(void)
2875 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2876 if (tok
== TOK_LINENUM
) {
2877 file
->line_num
= tokc
.i
;
2886 ST_FUNC
void next_nomacro(void)
2890 } while (tok
< 256 && (isidnum_table
[tok
- CH_EOF
] & IS_SPC
));
2894 static void macro_subst(
2895 TokenString
*tok_str
,
2897 const int *macro_str
,
2901 /* substitute arguments in replacement lists in macro_str by the values in
2902 args (field d) and return allocated string */
2903 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
2915 TOK_GET(&t
, ¯o_str
, &cval
);
2920 TOK_GET(&t
, ¯o_str
, &cval
);
2923 s
= sym_find2(args
, t
);
2926 cstr_ccat(&cstr
, '\"');
2930 TOK_GET(&t
, &st
, &cval
);
2931 if (t
!= TOK_PLCHLDR
2933 && 0 == check_space(t
, &spc
)) {
2934 const char *s
= get_tok_str(t
, &cval
);
2936 if (t
== TOK_PPSTR
&& *s
!= '\'')
2937 add_char(&cstr
, *s
);
2939 cstr_ccat(&cstr
, *s
);
2945 cstr_ccat(&cstr
, '\"');
2946 cstr_ccat(&cstr
, '\0');
2948 printf("\nstringize: <%s>\n", (char *)cstr
.data
);
2951 cval
.str
.size
= cstr
.size
;
2952 cval
.str
.data
= cstr
.data
;
2953 tok_str_add2(&str
, TOK_PPSTR
, &cval
);
2957 expect("macro parameter after '#'");
2959 } else if (t
>= TOK_IDENT
) {
2960 s
= sym_find2(args
, t
);
2964 /* if '##' is present before or after, no arg substitution */
2965 if (*macro_str
== TOK_PPJOIN
|| t1
== TOK_PPJOIN
) {
2966 /* special case for var arg macros : ## eats the ','
2967 if empty VA_ARGS variable. */
2968 if (t1
== TOK_PPJOIN
&& t0
== ',' && gnu_ext
&& s
->type
.t
) {
2970 /* suppress ',' '##' */
2973 /* suppress '##' and add variable */
2980 TOK_GET(&t1
, &st
, &cval
);
2983 tok_str_add2(&str
, t1
, &cval
);
2989 /* NOTE: the stream cannot be read when macro
2990 substituing an argument */
2991 macro_subst(&str
, nested_list
, st
, 0);
2993 if (str
.len
== l0
) /* exanded to empty string */
2994 tok_str_add(&str
, TOK_PLCHLDR
);
2996 tok_str_add(&str
, t
);
2999 tok_str_add2(&str
, t
, &cval
);
3003 tok_str_add(&str
, 0);
3007 static char const ab_month_name
[12][4] =
3009 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
3010 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
3013 /* peek or read [ws_str == NULL] next token from function macro call,
3014 walking up macro levels up to the file if necessary */
3015 static int next_argstream(Sym
**nested_list
, int can_read_stream
, TokenString
*ws_str
)
3023 p
= macro_ptr
, t
= *p
;
3025 while (is_space(t
) || TOK_LINEFEED
== t
)
3026 tok_str_add(ws_str
, t
), t
= *++p
;
3028 if (t
== 0 && can_read_stream
) {
3030 /* also, end of scope for nested defined symbol */
3032 while (sa
&& sa
->v
== 0)
3041 while (is_space(ch
) || ch
== '\n' || ch
== '/') {
3044 uint8_t *p
= file
->buf_ptr
;
3047 p
= parse_comment(p
);
3048 file
->buf_ptr
= p
- 1;
3049 } else if (c
== '/') {
3050 p
= parse_line_comment(p
);
3051 file
->buf_ptr
= p
- 1;
3056 if (!(ch
== '\f' || ch
== '\v' || ch
== '\r'))
3057 tok_str_add(ws_str
, ch
);
3071 /* do macro substitution of current token with macro 's' and add
3072 result to (tok_str,tok_len). 'nested_list' is the list of all
3073 macros we got inside to avoid recursing. Return non zero if no
3074 substitution needs to be done */
3075 static int macro_subst_tok(
3076 TokenString
*tok_str
,
3079 int can_read_stream
)
3081 Sym
*args
, *sa
, *sa1
;
3082 int parlevel
, *mstr
, t
, t1
, spc
;
3089 /* if symbol is a macro, prepare substitution */
3090 /* special macros */
3091 if (tok
== TOK___LINE__
) {
3092 snprintf(buf
, sizeof(buf
), "%d", file
->line_num
);
3096 } else if (tok
== TOK___FILE__
) {
3097 cstrval
= file
->filename
;
3099 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
3104 tm
= localtime(&ti
);
3105 if (tok
== TOK___DATE__
) {
3106 snprintf(buf
, sizeof(buf
), "%s %2d %d",
3107 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
3109 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
3110 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
3117 cstr_cat(&cstr
, cstrval
, 0);
3118 cval
.str
.size
= cstr
.size
;
3119 cval
.str
.data
= cstr
.data
;
3120 tok_str_add2(tok_str
, t1
, &cval
);
3123 int saved_parse_flags
= parse_flags
;
3126 if (s
->type
.t
== MACRO_FUNC
) {
3127 /* whitespace between macro name and argument list */
3129 tok_str_new(&ws_str
);
3132 parse_flags
|= PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
3133 | PARSE_FLAG_ACCEPT_STRAYS
;
3135 /* get next token from argument stream */
3136 t
= next_argstream(nested_list
, can_read_stream
, &ws_str
);
3138 /* not a macro substitution after all, restore the
3139 * macro token plus all whitespace we've read.
3140 * whitespace is intentionally not merged to preserve
3142 parse_flags
= saved_parse_flags
;
3143 tok_str_add(tok_str
, tok
);
3144 if (parse_flags
& PARSE_FLAG_SPACES
) {
3146 for (i
= 0; i
< ws_str
.len
; i
++)
3147 tok_str_add(tok_str
, ws_str
.str
[i
]);
3149 tok_str_free_str(ws_str
.str
);
3152 tok_str_free_str(ws_str
.str
);
3154 next_nomacro(); /* eat '(' */
3156 /* argument macro */
3159 /* NOTE: empty args are allowed, except if no args */
3162 next_argstream(nested_list
, can_read_stream
, NULL
);
3163 } while (is_space(tok
) || TOK_LINEFEED
== tok
);
3165 /* handle '()' case */
3166 if (!args
&& !sa
&& tok
== ')')
3169 tcc_error("macro '%s' used with too many args",
3170 get_tok_str(s
->v
, 0));
3173 /* NOTE: non zero sa->t indicates VA_ARGS */
3174 while ((parlevel
> 0 ||
3176 (tok
!= ',' || sa
->type
.t
)))) {
3177 if (tok
== TOK_EOF
|| tok
== 0)
3181 else if (tok
== ')')
3183 if (tok
== TOK_LINEFEED
)
3185 if (!check_space(tok
, &spc
))
3186 tok_str_add2(&str
, tok
, &tokc
);
3187 next_argstream(nested_list
, can_read_stream
, NULL
);
3192 tok_str_add(&str
, 0);
3193 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
3197 /* special case for gcc var args: add an empty
3198 var arg argument if it is omitted */
3199 if (sa
&& sa
->type
.t
&& gnu_ext
)
3207 tcc_error("macro '%s' used with too few args",
3208 get_tok_str(s
->v
, 0));
3211 parse_flags
= saved_parse_flags
;
3213 /* now subst each arg */
3214 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
3219 tok_str_free_str(sa
->d
);
3225 sym_push2(nested_list
, s
->v
, 0, 0);
3226 parse_flags
= saved_parse_flags
;
3227 macro_subst(tok_str
, nested_list
, mstr
, can_read_stream
| 2);
3229 /* pop nested defined symbol */
3231 *nested_list
= sa1
->prev
;
3234 tok_str_free_str(mstr
);
3239 static int paste_tokens(int t1
, CValue
*v1
, int t2
, CValue
*v2
)
3245 if (t1
!= TOK_PLCHLDR
)
3246 cstr_cat(&cstr
, get_tok_str(t1
, v1
), -1);
3248 if (t2
!= TOK_PLCHLDR
)
3249 cstr_cat(&cstr
, get_tok_str(t2
, v2
), -1);
3250 cstr_ccat(&cstr
, '\0');
3252 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3253 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3256 if (0 == *file
->buf_ptr
)
3260 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid"
3261 " preprocessing token", n
, cstr
.data
, (char*)cstr
.data
+ n
);
3266 //printf("paste <%s>\n", (char*)cstr.data);
3271 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
3272 return the resulting string (which must be freed). */
3273 static inline int *macro_twosharps(const int *ptr0
)
3277 TokenString macro_str1
;
3278 int start_of_nosubsts
= -1;
3281 /* we search the first '##' */
3282 for (ptr
= ptr0
;;) {
3283 TOK_GET(&t
, &ptr
, &cval
);
3284 if (t
== TOK_PPJOIN
)
3290 tok_str_new(¯o_str1
);
3292 //tok_print(" $$$", ptr0);
3293 for (ptr
= ptr0
;;) {
3294 TOK_GET(&t
, &ptr
, &cval
);
3297 if (t
== TOK_PPJOIN
)
3299 while (*ptr
== TOK_PPJOIN
) {
3301 /* given 'a##b', remove nosubsts preceding 'a' */
3302 if (start_of_nosubsts
>= 0)
3303 macro_str1
.len
= start_of_nosubsts
;
3304 /* given 'a##b', remove nosubsts preceding 'b' */
3305 while ((t1
= *++ptr
) == TOK_NOSUBST
)
3307 if (t1
&& t1
!= TOK_PPJOIN
) {
3308 TOK_GET(&t1
, &ptr
, &cv1
);
3309 if (t
!= TOK_PLCHLDR
|| t1
!= TOK_PLCHLDR
) {
3310 if (paste_tokens(t
, &cval
, t1
, &cv1
)) {
3311 t
= tok
, cval
= tokc
;
3313 tok_str_add2(¯o_str1
, t
, &cval
);
3319 if (t
== TOK_NOSUBST
) {
3320 if (start_of_nosubsts
< 0)
3321 start_of_nosubsts
= macro_str1
.len
;
3323 start_of_nosubsts
= -1;
3325 tok_str_add2(¯o_str1
, t
, &cval
);
3327 tok_str_add(¯o_str1
, 0);
3328 //tok_print(" ###", macro_str1.str);
3329 return macro_str1
.str
;
3332 /* do macro substitution of macro_str and add result to
3333 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3334 inside to avoid recursing. */
3335 static void macro_subst(
3336 TokenString
*tok_str
,
3338 const int *macro_str
,
3344 int t
, spc
, nosubst
;
3346 int *macro_str1
= NULL
;
3348 /* first scan for '##' operator handling */
3352 /* first scan for '##' operator handling */
3353 if (can_read_stream
) {
3354 macro_str1
= macro_twosharps(ptr
);
3360 TOK_GET(&t
, &ptr
, &cval
);
3364 if (t
>= TOK_IDENT
&& 0 == nosubst
) {
3369 /* if nested substitution, do nothing */
3370 if (sym_find2(*nested_list
, t
)) {
3371 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3372 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3378 str
.str
= (int*)ptr
;
3379 begin_macro(&str
, 2);
3382 macro_subst_tok(tok_str
, nested_list
, s
, can_read_stream
);
3384 if (str
.alloc
== 3) {
3385 /* already finished by reading function macro arguments */
3393 spc
= (tok_str
->len
&&
3394 is_space(tok_last(tok_str
->str
,
3395 tok_str
->str
+ tok_str
->len
)));
3399 if (t
== '\\' && !(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
3400 tcc_error("stray '\\' in program");
3403 if (!check_space(t
, &spc
))
3404 tok_str_add2(tok_str
, t
, &cval
);
3406 if (t
== TOK_NOSUBST
)
3411 tok_str_free_str(macro_str1
);
3415 /* return next token with macro substitution */
3416 ST_FUNC
void next(void)
3419 if (parse_flags
& PARSE_FLAG_SPACES
)
3425 if (tok
== TOK_NOSUBST
|| tok
== TOK_PLCHLDR
) {
3426 /* discard preprocessor markers */
3428 } else if (tok
== 0) {
3429 /* end of macro or unget token string */
3433 } else if (tok
>= TOK_IDENT
&& (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3435 /* if reading from file, try to substitute macros */
3436 s
= define_find(tok
);
3438 Sym
*nested_list
= NULL
;
3440 macro_subst_tok(&tokstr_buf
, &nested_list
, s
, 1);
3441 tok_str_add(&tokstr_buf
, 0);
3442 begin_macro(&tokstr_buf
, 2);
3446 /* convert preprocessor tokens into C tokens */
3447 if (tok
== TOK_PPNUM
) {
3448 if (parse_flags
& PARSE_FLAG_TOK_NUM
)
3449 parse_number((char *)tokc
.str
.data
);
3450 } else if (tok
== TOK_PPSTR
) {
3451 if (parse_flags
& PARSE_FLAG_TOK_STR
)
3452 parse_string((char *)tokc
.str
.data
, tokc
.str
.size
- 1);
3456 /* push back current token and set current token to 'last_tok'. Only
3457 identifier case handled for labels. */
3458 ST_INLN
void unget_tok(int last_tok
)
3461 TokenString
*str
= tok_str_alloc();
3462 tok_str_add2(str
, tok
, &tokc
);
3463 tok_str_add(str
, 0);
3464 begin_macro(str
, 1);
3468 ST_FUNC
void preprocess_start(TCCState
*s1
)
3471 s1
->include_stack_ptr
= s1
->include_stack
;
3472 /* XXX: move that before to avoid having to initialize
3473 file->ifdef_stack_ptr ? */
3474 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3475 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3478 pvtop
= vtop
= vstack
- 1;
3479 s1
->pack_stack
[0] = 0;
3480 s1
->pack_stack_ptr
= s1
->pack_stack
;
3482 set_idnum('$', s1
->dollars_in_identifiers
? IS_ID
: 0);
3483 set_idnum('.', (parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0);
3484 buf
= tcc_malloc(3 + strlen(file
->filename
));
3485 sprintf(buf
, "\"%s\"", file
->filename
);
3486 tcc_undefine_symbol(s1
, "__BASE_FILE__");
3487 tcc_define_symbol(s1
, "__BASE_FILE__", buf
);
3489 if (s1
->nb_cmd_include_files
) {
3493 for (i
= 0; i
< s1
->nb_cmd_include_files
; i
++) {
3494 cstr_cat(&cstr
, "#include \"", -1);
3495 cstr_cat(&cstr
, s1
->cmd_include_files
[i
], -1);
3496 cstr_cat(&cstr
, "\"\n", -1);
3498 *s1
->include_stack_ptr
++ = file
;
3499 tcc_open_bf(s1
, "<command line>", cstr
.size
);
3500 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3505 ST_FUNC
void tccpp_new(TCCState
*s
)
3510 /* might be used in error() before preprocess_start() */
3511 s
->include_stack_ptr
= s
->include_stack
;
3513 /* init isid table */
3514 for(i
= CH_EOF
; i
<128; i
++)
3516 is_space(i
) ? IS_SPC
3521 for(i
= 128; i
<256; i
++)
3522 set_idnum(i
, IS_ID
);
3524 /* init allocators */
3525 tal_new(&toksym_alloc
, TOKSYM_TAL_LIMIT
, TOKSYM_TAL_SIZE
);
3526 tal_new(&tokstr_alloc
, TOKSTR_TAL_LIMIT
, TOKSTR_TAL_SIZE
);
3527 tal_new(&cstr_alloc
, CSTR_TAL_LIMIT
, CSTR_TAL_SIZE
);
3529 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3530 cstr_new(&cstr_buf
);
3531 cstr_realloc(&cstr_buf
, STRING_MAX_SIZE
);
3532 tok_str_new(&tokstr_buf
);
3533 tok_str_realloc(&tokstr_buf
, TOKSTR_MAX_SIZE
);
3535 tok_ident
= TOK_IDENT
;
3544 tok_alloc(p
, r
- p
- 1);
3549 ST_FUNC
void tccpp_delete(TCCState
*s
)
3553 /* free -D and compiler defines */
3556 /* cleanup from error/setjmp */
3562 n
= tok_ident
- TOK_IDENT
;
3563 for(i
= 0; i
< n
; i
++)
3564 tal_free(toksym_alloc
, table_ident
[i
]);
3565 tcc_free(table_ident
);
3568 /* free static buffers */
3569 cstr_free(&tokcstr
);
3570 cstr_free(&cstr_buf
);
3571 cstr_free(¯o_equal_buf
);
3572 tok_str_free_str(tokstr_buf
.str
);
3574 /* free allocators */
3575 tal_delete(toksym_alloc
);
3576 toksym_alloc
= NULL
;
3577 tal_delete(tokstr_alloc
);
3578 tokstr_alloc
= NULL
;
3579 tal_delete(cstr_alloc
);
3583 /* ------------------------------------------------------------------------- */
3584 /* tcc -E [-P[1]] [-dD} support */
3586 static void tok_print(const char *msg
, const int *str
)
3592 fp
= tcc_state
->ppfp
;
3593 if (!fp
|| !tcc_state
->dflag
)
3596 fprintf(fp
, "%s ", msg
);
3598 TOK_GET(&t
, &str
, &cval
);
3601 fprintf(fp
,"%s", get_tok_str(t
, &cval
));
3606 static void pp_line(TCCState
*s1
, BufferedFile
*f
, int level
)
3608 int d
= f
->line_num
- f
->line_ref
;
3613 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_NONE
) {
3615 } else if (level
== 0 && f
->line_ref
&& d
< 8) {
3617 fputs("\n", s1
->ppfp
), --d
;
3618 } else if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_STD
) {
3619 fprintf(s1
->ppfp
, "#line %d \"%s\"\n", f
->line_num
, f
->filename
);
3621 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
,
3622 level
> 0 ? " 1" : level
< 0 ? " 2" : "");
3624 f
->line_ref
= f
->line_num
;
3627 static void define_print(TCCState
*s1
, int v
)
3633 if (NULL
== s
|| NULL
== s
->d
)
3637 fprintf(fp
, "#define %s", get_tok_str(v
, NULL
));
3638 if (s
->type
.t
== MACRO_FUNC
) {
3643 fprintf(fp
,"%s", get_tok_str(a
->v
& ~SYM_FIELD
, NULL
));
3650 tok_print("", s
->d
);
3653 static void pp_debug_defines(TCCState
*s1
)
3664 pp_line(s1
, file
, 0);
3665 file
->line_ref
= ++file
->line_num
;
3669 vs
= get_tok_str(v
, NULL
);
3670 if (t
== TOK_DEFINE
) {
3671 define_print(s1
, v
);
3672 } else if (t
== TOK_UNDEF
) {
3673 fprintf(fp
, "#undef %s\n", vs
);
3674 } else if (t
== TOK_push_macro
) {
3675 fprintf(fp
, "#pragma push_macro(\"%s\")\n", vs
);
3676 } else if (t
== TOK_pop_macro
) {
3677 fprintf(fp
, "#pragma pop_macro(\"%s\")\n", vs
);
3682 static void pp_debug_builtins(TCCState
*s1
)
3685 for (v
= TOK_IDENT
; v
< tok_ident
; ++v
)
3686 define_print(s1
, v
);
3689 /* Add a space between tokens a and b to avoid unwanted textual pasting */
3690 static int pp_need_space(int a
, int b
)
3692 return 'E' == a
? '+' == b
|| '-' == b
3693 : '+' == a
? TOK_INC
== b
|| '+' == b
3694 : '-' == a
? TOK_DEC
== b
|| '-' == b
3695 : a
>= TOK_IDENT
? b
>= TOK_IDENT
3699 /* maybe hex like 0x1e */
3700 static int pp_check_he0xE(int t
, const char *p
)
3702 if (t
== TOK_PPNUM
&& toup(strchr(p
, 0)[-1]) == 'E')
3707 /* Preprocess the current file */
3708 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3710 BufferedFile
**iptr
;
3711 int token_seen
, spcs
, level
;
3715 preprocess_start(s1
);
3716 ch
= file
->buf_ptr
[0];
3717 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3718 parse_flags
= PARSE_FLAG_PREPROCESS
3719 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
3720 | PARSE_FLAG_LINEFEED
3722 | PARSE_FLAG_ACCEPT_STRAYS
3724 define_start
= define_stack
;
3726 /* Credits to Fabrice Bellard's initial revision to demonstrate its
3727 capability to compile and run itself, provided all numbers are
3728 given as decimals. tcc -E -P10 will do. */
3729 if (s1
->Pflag
== 1 + 10)
3730 parse_flags
|= PARSE_FLAG_TOK_NUM
, s1
->Pflag
= 1;
3733 /* for PP benchmarks */
3734 do next(); while (tok
!= TOK_EOF
);
3735 free_defines(define_start
);
3739 if (s1
->dflag
& 1) {
3740 pp_debug_builtins(s1
);
3744 token_seen
= TOK_LINEFEED
, spcs
= 0;
3745 pp_line(s1
, file
, 0);
3748 iptr
= s1
->include_stack_ptr
;
3752 level
= s1
->include_stack_ptr
- iptr
;
3755 pp_line(s1
, *iptr
, 0);
3756 pp_line(s1
, file
, level
);
3760 pp_debug_defines(s1
);
3765 if (token_seen
== TOK_LINEFEED
) {
3770 if (tok
== TOK_LINEFEED
) {
3774 pp_line(s1
, file
, 0);
3775 } else if (tok
== TOK_LINEFEED
) {
3778 spcs
= pp_need_space(token_seen
, tok
);
3782 fputs(" ", s1
->ppfp
), --spcs
;
3783 fputs(p
= get_tok_str(tok
, &tokc
), s1
->ppfp
);
3784 token_seen
= pp_check_he0xE(tok
, p
);;
3786 /* reset define stack, but keep -D and built-ins */
3787 free_defines(define_start
);
3791 /* ------------------------------------------------------------------------- */