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
;
1025 /* return the number of additional 'ints' necessary to store the
1027 static inline int tok_size(const int *p
)
1042 return 1 + ((sizeof(CString
) + ((CString
*)(p
+1))->size
+ 3) >> 2);
1048 return 1 + LDOUBLE_SIZE
/ 4;
1055 /* token string handling */
1056 ST_INLN
void tok_str_new(TokenString
*s
)
1060 s
->allocated_len
= 0;
1061 s
->last_line_num
= -1;
1064 ST_FUNC TokenString
*tok_str_alloc(void)
1066 TokenString
*str
= tal_realloc(tokstr_alloc
, 0, sizeof *str
);
1071 ST_FUNC
int *tok_str_dup(TokenString
*s
)
1075 str
= tal_realloc(tokstr_alloc
, 0, s
->len
* sizeof(int));
1076 memcpy(str
, s
->str
, s
->len
* sizeof(int));
1080 ST_FUNC
void tok_str_free_str(int *str
)
1082 tal_free(tokstr_alloc
, str
);
1085 ST_FUNC
void tok_str_free(TokenString
*str
)
1087 tok_str_free_str(str
->str
);
1088 tal_free(tokstr_alloc
, str
);
1091 ST_FUNC
int *tok_str_realloc(TokenString
*s
, int new_size
)
1095 size
= s
->allocated_len
;
1098 while (size
< new_size
)
1100 if (size
> s
->allocated_len
) {
1101 str
= tal_realloc(tokstr_alloc
, s
->str
, size
* sizeof(int));
1102 s
->allocated_len
= size
;
1108 ST_FUNC
void tok_str_add(TokenString
*s
, int t
)
1114 if (len
>= s
->allocated_len
)
1115 str
= tok_str_realloc(s
, len
+ 1);
1120 ST_FUNC
void begin_macro(TokenString
*str
, int alloc
)
1123 str
->prev
= macro_stack
;
1124 str
->prev_ptr
= macro_ptr
;
1125 macro_ptr
= str
->str
;
1129 ST_FUNC
void end_macro(void)
1131 TokenString
*str
= macro_stack
;
1132 macro_stack
= str
->prev
;
1133 macro_ptr
= str
->prev_ptr
;
1134 if (str
->alloc
== 2) {
1135 str
->alloc
= 3; /* just mark as finished */
1141 static void tok_str_add2(TokenString
*s
, int t
, CValue
*cv
)
1148 /* allocate space for worst case */
1149 if (len
+ TOK_MAX_SIZE
>= s
->allocated_len
)
1150 str
= tok_str_realloc(s
, len
+ TOK_MAX_SIZE
+ 1);
1159 str
[len
++] = cv
->tab
[0];
1166 /* Insert the string into the int array. */
1168 1 + (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1169 if (len
+ nb_words
>= s
->allocated_len
)
1170 str
= tok_str_realloc(s
, len
+ nb_words
+ 1);
1171 str
[len
] = cv
->str
.size
;
1172 memcpy(&str
[len
+ 1], cv
->str
.data
, cv
->str
.size
);
1179 #if LDOUBLE_SIZE == 8
1182 str
[len
++] = cv
->tab
[0];
1183 str
[len
++] = cv
->tab
[1];
1185 #if LDOUBLE_SIZE == 12
1187 str
[len
++] = cv
->tab
[0];
1188 str
[len
++] = cv
->tab
[1];
1189 str
[len
++] = cv
->tab
[2];
1190 #elif LDOUBLE_SIZE == 16
1192 str
[len
++] = cv
->tab
[0];
1193 str
[len
++] = cv
->tab
[1];
1194 str
[len
++] = cv
->tab
[2];
1195 str
[len
++] = cv
->tab
[3];
1196 #elif LDOUBLE_SIZE != 8
1197 #error add long double size support
1206 /* add the current parse token in token string 's' */
1207 ST_FUNC
void tok_str_add_tok(TokenString
*s
)
1211 /* save line number info */
1212 if (file
->line_num
!= s
->last_line_num
) {
1213 s
->last_line_num
= file
->line_num
;
1214 cval
.i
= s
->last_line_num
;
1215 tok_str_add2(s
, TOK_LINENUM
, &cval
);
1217 tok_str_add2(s
, tok
, &tokc
);
1220 /* get a token from an integer array and increment pointer
1221 accordingly. we code it as a macro to avoid pointer aliasing. */
1222 static inline void TOK_GET(int *t
, const int **pp
, CValue
*cv
)
1241 cv
->str
.size
= *p
++;
1243 p
+= (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1251 #if LDOUBLE_SIZE == 16
1253 #elif LDOUBLE_SIZE == 12
1255 #elif LDOUBLE_SIZE == 8
1258 # error add long double size support
1271 /* Calling this function is expensive, but it is not possible
1272 to read a token string backwards. */
1273 static int tok_last(const int *str0
, const int *str1
)
1275 const int *str
= str0
;
1280 TOK_GET(&tok
, &str
, &cval
);
1284 static int macro_is_equal(const int *a
, const int *b
)
1293 /* first time preallocate macro_equal_buf, next time only reset position to start */
1294 cstr_reset(¯o_equal_buf
);
1295 TOK_GET(&t
, &a
, &cv
);
1296 cstr_cat(¯o_equal_buf
, get_tok_str(t
, &cv
), 0);
1297 TOK_GET(&t
, &b
, &cv
);
1298 if (strcmp(macro_equal_buf
.data
, get_tok_str(t
, &cv
)))
1304 /* defines handling */
1305 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1310 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1312 s
->next
= first_arg
;
1313 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1315 if (o
&& !macro_is_equal(o
->d
, s
->d
))
1316 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1319 /* undefined a define symbol. Its name is just set to zero */
1320 ST_FUNC
void define_undef(Sym
*s
)
1323 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1324 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1327 ST_INLN Sym
*define_find(int v
)
1330 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1332 return table_ident
[v
]->sym_define
;
1335 /* free define stack until top reaches 'b' */
1336 ST_FUNC
void free_defines(Sym
*b
)
1338 while (define_stack
!= b
) {
1339 Sym
*top
= define_stack
;
1340 define_stack
= top
->prev
;
1341 tok_str_free_str(top
->d
);
1346 /* restore remaining (-D or predefined) symbols if they were
1347 #undef'd in the file */
1350 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1351 Sym
**d
= &table_ident
[v
- TOK_IDENT
]->sym_define
;
1360 ST_FUNC Sym
*label_find(int v
)
1363 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1365 return table_ident
[v
]->sym_label
;
1368 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1371 s
= sym_push2(ptop
, v
, 0, 0);
1373 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1374 if (ptop
== &global_label_stack
) {
1375 /* modify the top most local identifier, so that
1376 sym_identifier will point to 's' when popped */
1378 ps
= &(*ps
)->prev_tok
;
1385 /* pop labels until element last is reached. Look if any labels are
1386 undefined. Define symbols if '&&label' was used. */
1387 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
)
1390 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1392 if (s
->r
== LABEL_DECLARED
) {
1393 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1394 } else if (s
->r
== LABEL_FORWARD
) {
1395 tcc_error("label '%s' used but not defined",
1396 get_tok_str(s
->v
, NULL
));
1399 /* define corresponding symbol. A size of
1401 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1405 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1411 /* eval an expression for #if/#elif */
1412 static int expr_preprocess(void)
1417 str
= tok_str_alloc();
1418 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1419 next(); /* do macro subst */
1420 if (tok
== TOK_DEFINED
) {
1425 c
= define_find(tok
) != 0;
1430 } else if (tok
>= TOK_IDENT
) {
1431 /* if undefined macro */
1435 tok_str_add_tok(str
);
1437 tok_str_add(str
, -1); /* simulate end of file */
1438 tok_str_add(str
, 0);
1439 /* now evaluate C constant expression */
1440 begin_macro(str
, 1);
1448 /* parse after #define */
1449 ST_FUNC
void parse_define(void)
1451 Sym
*s
, *first
, **ps
;
1452 int v
, t
, varg
, is_vaargs
, spc
;
1453 int saved_parse_flags
= parse_flags
;
1457 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1458 /* XXX: should check if same macro (ANSI) */
1461 /* We have to parse the whole define as if not in asm mode, in particular
1462 no line comment with '#' must be ignored. Also for function
1463 macros the argument list must be parsed without '.' being an ID
1465 parse_flags
= ((parse_flags
& ~PARSE_FLAG_ASM_FILE
) | PARSE_FLAG_SPACES
);
1466 /* '(' must be just after macro definition for MACRO_FUNC */
1472 if (tok
!= ')') for (;;) {
1476 if (varg
== TOK_DOTS
) {
1477 varg
= TOK___VA_ARGS__
;
1479 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1483 if (varg
< TOK_IDENT
)
1485 tcc_error("bad macro parameter list");
1486 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1491 if (tok
!= ',' || is_vaargs
)
1501 parse_flags
|= PARSE_FLAG_ACCEPT_STRAYS
| PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
;
1502 /* The body of a macro definition should be parsed such that identifiers
1503 are parsed like the file mode determines (i.e. with '.' being an
1504 ID character in asm mode). But '#' should be retained instead of
1505 regarded as line comment leader, so still don't set ASM_FILE
1507 set_idnum('.', (saved_parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0);
1508 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1509 /* remove spaces around ## and after '#' */
1510 if (TOK_TWOSHARPS
== tok
) {
1517 } else if ('#' == tok
) {
1519 } else if (check_space(tok
, &spc
)) {
1522 tok_str_add2(&tokstr_buf
, tok
, &tokc
);
1527 parse_flags
= saved_parse_flags
;
1529 --tokstr_buf
.len
; /* remove trailing space */
1530 tok_str_add(&tokstr_buf
, 0);
1533 tcc_error("'##' cannot appear at either end of macro");
1534 define_push(v
, t
, tok_str_dup(&tokstr_buf
), first
);
1537 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
, int add
)
1539 const unsigned char *s
;
1545 s
= (unsigned char *) filename
;
1548 h
= TOK_HASH_FUNC(h
, toup(*s
));
1550 h
= TOK_HASH_FUNC(h
, *s
);
1554 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1556 i
= s1
->cached_includes_hash
[h
];
1560 e
= s1
->cached_includes
[i
- 1];
1561 if (0 == PATHCMP(e
->filename
, filename
))
1568 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1569 strcpy(e
->filename
, filename
);
1570 e
->ifndef_macro
= e
->once
= 0;
1571 dynarray_add(&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1572 /* add in hash table */
1573 e
->hash_next
= s1
->cached_includes_hash
[h
];
1574 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1576 printf("adding cached '%s'\n", filename
);
1581 static void pragma_parse(TCCState
*s1
)
1584 if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1588 if (next(), tok
!= '(')
1590 if (next(), tok
!= TOK_STR
)
1592 v
= tok_alloc(tokc
.str
.data
, tokc
.str
.size
- 1)->tok
;
1593 if (next(), tok
!= ')')
1595 if (t
== TOK_push_macro
) {
1596 while (NULL
== (s
= define_find(v
)))
1597 define_push(v
, 0, NULL
, NULL
);
1598 s
->type
.ref
= s
; /* set push boundary */
1600 for (s
= define_stack
; s
; s
= s
->prev
)
1601 if (s
->v
== v
&& s
->type
.ref
== s
) {
1607 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1609 tcc_warning("unbalanced #pragma pop_macro");
1610 pp_debug_tok
= t
, pp_debug_symv
= v
;
1612 } else if (tok
== TOK_once
) {
1613 search_cached_include(s1
, file
->filename
, 1)->once
= pp_once
;
1615 } else if (s1
->ppfp
) {
1616 /* tcc -E: keep pragmas below unchanged */
1618 unget_tok(TOK_PRAGMA
);
1620 unget_tok(TOK_LINEFEED
);
1622 } else if (tok
== TOK_pack
) {
1624 #pragma pack(1) // set
1625 #pragma pack() // reset to default
1626 #pragma pack(push,1) // push & set
1627 #pragma pack(pop) // restore previous */
1630 if (tok
== TOK_ASM_pop
) {
1632 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1634 tcc_error("out of pack stack");
1636 s1
->pack_stack_ptr
--;
1640 if (tok
== TOK_ASM_push
) {
1642 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1644 s1
->pack_stack_ptr
++;
1647 if (tok
!= TOK_CINT
)
1650 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1654 *s1
->pack_stack_ptr
= val
;
1659 } else if (tok
== TOK_comment
) {
1669 file
= tcc_strdup((char *)tokc
.str
.data
);
1670 dynarray_add(&s1
->pragma_libs
, &s1
->nb_pragma_libs
, file
);
1676 if (s1
->warn_unsupported
)
1677 tcc_warning("#pragma %s is ignored", get_tok_str(tok
, &tokc
));
1682 tcc_error("malformed #pragma directive");
1686 /* is_bof is true if first non space token at beginning of file */
1687 ST_FUNC
void preprocess(int is_bof
)
1689 TCCState
*s1
= tcc_state
;
1690 int i
, c
, n
, saved_parse_flags
;
1694 saved_parse_flags
= parse_flags
;
1695 parse_flags
= PARSE_FLAG_PREPROCESS
1696 | PARSE_FLAG_TOK_NUM
1697 | PARSE_FLAG_TOK_STR
1698 | PARSE_FLAG_LINEFEED
1699 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
1708 pp_debug_symv
= tok
;
1714 pp_debug_symv
= tok
;
1715 s
= define_find(tok
);
1716 /* undefine symbol by putting an invalid name */
1721 case TOK_INCLUDE_NEXT
:
1722 ch
= file
->buf_ptr
[0];
1723 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1728 } else if (ch
== '\"') {
1733 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1734 if ((q
- buf
) < sizeof(buf
) - 1)
1737 if (handle_stray_noerror() == 0)
1745 /* eat all spaces and comments after include */
1746 /* XXX: slightly incorrect */
1747 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1752 /* computed #include : concatenate everything up to linefeed,
1753 the result must be one of the two accepted forms.
1754 Don't convert pp-tokens to tokens here. */
1755 parse_flags
= (PARSE_FLAG_PREPROCESS
1756 | PARSE_FLAG_LINEFEED
1757 | (parse_flags
& PARSE_FLAG_ASM_FILE
));
1760 while (tok
!= TOK_LINEFEED
) {
1761 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1765 /* check syntax and remove '<>|""' */
1766 if ((len
< 2 || ((buf
[0] != '"' || buf
[len
-1] != '"') &&
1767 (buf
[0] != '<' || buf
[len
-1] != '>'))))
1768 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1770 memmove(buf
, buf
+ 1, len
- 2);
1771 buf
[len
- 2] = '\0';
1774 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1775 tcc_error("#include recursion too deep");
1776 /* store current file in stack, but increment stack later below */
1777 *s1
->include_stack_ptr
= file
;
1778 i
= tok
== TOK_INCLUDE_NEXT
? file
->include_next_index
: 0;
1779 n
= 2 + s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1780 for (; i
< n
; ++i
) {
1781 char buf1
[sizeof file
->filename
];
1786 /* check absolute include path */
1787 if (!IS_ABSPATH(buf
))
1791 } else if (i
== 1) {
1792 /* search in file's dir if "header.h" */
1795 /* https://savannah.nongnu.org/bugs/index.php?50847 */
1796 path
= file
->filename2
;
1797 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1800 /* search in all the include paths */
1801 int j
= i
- 2, k
= j
- s1
->nb_include_paths
;
1802 path
= k
< 0 ? s1
->include_paths
[j
] : s1
->sysinclude_paths
[k
];
1803 pstrcpy(buf1
, sizeof(buf1
), path
);
1804 pstrcat(buf1
, sizeof(buf1
), "/");
1807 pstrcat(buf1
, sizeof(buf1
), buf
);
1808 e
= search_cached_include(s1
, buf1
, 0);
1809 if (e
&& (define_find(e
->ifndef_macro
) || e
->once
== pp_once
)) {
1810 /* no need to parse the include because the 'ifndef macro'
1811 is defined (or had #pragma once) */
1813 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1818 if (tcc_open(s1
, buf1
) < 0)
1821 file
->include_next_index
= i
+ 1;
1823 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1825 /* update target deps */
1826 dynarray_add(&s1
->target_deps
, &s1
->nb_target_deps
,
1828 /* push current file in stack */
1829 ++s1
->include_stack_ptr
;
1830 /* add include file debug info */
1832 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1833 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1834 ch
= file
->buf_ptr
[0];
1837 tcc_error("include file '%s' not found", buf
);
1844 c
= expr_preprocess();
1850 if (tok
< TOK_IDENT
)
1851 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1855 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1857 file
->ifndef_macro
= tok
;
1860 c
= (define_find(tok
) != 0) ^ c
;
1862 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1863 tcc_error("memory full (ifdef)");
1864 *s1
->ifdef_stack_ptr
++ = c
;
1867 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1868 tcc_error("#else without matching #if");
1869 if (s1
->ifdef_stack_ptr
[-1] & 2)
1870 tcc_error("#else after #else");
1871 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1874 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1875 tcc_error("#elif without matching #if");
1876 c
= s1
->ifdef_stack_ptr
[-1];
1878 tcc_error("#elif after #else");
1879 /* last #if/#elif expression was true: we skip */
1883 c
= expr_preprocess();
1884 s1
->ifdef_stack_ptr
[-1] = c
;
1887 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1888 file
->ifndef_macro
= 0;
1897 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1898 tcc_error("#endif without matching #if");
1899 s1
->ifdef_stack_ptr
--;
1900 /* '#ifndef macro' was at the start of file. Now we check if
1901 an '#endif' is exactly at the end of file */
1902 if (file
->ifndef_macro
&&
1903 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1904 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1905 /* need to set to zero to avoid false matches if another
1906 #ifndef at middle of file */
1907 file
->ifndef_macro
= 0;
1908 while (tok
!= TOK_LINEFEED
)
1910 tok_flags
|= TOK_FLAG_ENDIF
;
1915 n
= strtoul((char*)tokc
.str
.data
, &q
, 10);
1919 if (tok
!= TOK_CINT
)
1921 tcc_error("wrong #line format");
1925 if (tok
!= TOK_LINEFEED
) {
1927 pstrcpy(file
->filename
, sizeof(file
->filename
), (char *)tokc
.str
.data
);
1928 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
1935 total_lines
+= file
->line_num
- n
;
1938 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1943 ch
= file
->buf_ptr
[0];
1946 while (ch
!= '\n' && ch
!= CH_EOF
) {
1947 if ((q
- buf
) < sizeof(buf
) - 1)
1950 if (handle_stray_noerror() == 0)
1957 tcc_error("#error %s", buf
);
1959 tcc_warning("#warning %s", buf
);
1967 /* ignore gas line comment in an 'S' file. */
1968 if (saved_parse_flags
& PARSE_FLAG_ASM_FILE
)
1970 if (tok
== '!' && is_bof
)
1971 /* '!' is ignored at beginning to allow C scripts. */
1973 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
1975 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
- 1);
1978 /* ignore other preprocess commands or #! for C scripts */
1979 while (tok
!= TOK_LINEFEED
)
1982 parse_flags
= saved_parse_flags
;
1985 /* evaluate escape codes in a string. */
1986 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
2001 case '0': case '1': case '2': case '3':
2002 case '4': case '5': case '6': case '7':
2003 /* at most three octal digits */
2008 n
= n
* 8 + c
- '0';
2012 n
= n
* 8 + c
- '0';
2017 goto add_char_nonext
;
2025 if (c
>= 'a' && c
<= 'f')
2027 else if (c
>= 'A' && c
<= 'F')
2037 goto add_char_nonext
;
2061 goto invalid_escape
;
2071 if (c
>= '!' && c
<= '~')
2072 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
2074 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
2081 cstr_ccat(outstr
, c
);
2083 cstr_wccat(outstr
, c
);
2085 /* add a trailing '\0' */
2087 cstr_ccat(outstr
, '\0');
2089 cstr_wccat(outstr
, '\0');
2092 static void parse_string(const char *s
, int len
)
2094 uint8_t buf
[1000], *p
= buf
;
2097 if ((is_long
= *s
== 'L'))
2101 if (len
>= sizeof buf
)
2102 p
= tcc_malloc(len
+ 1);
2106 cstr_reset(&tokcstr
);
2107 parse_escape_string(&tokcstr
, p
, is_long
);
2113 /* XXX: make it portable */
2117 char_size
= sizeof(nwchar_t
);
2118 if (tokcstr
.size
<= char_size
)
2119 tcc_error("empty character constant");
2120 if (tokcstr
.size
> 2 * char_size
)
2121 tcc_warning("multi-character character constant");
2123 tokc
.i
= *(int8_t *)tokcstr
.data
;
2126 tokc
.i
= *(nwchar_t
*)tokcstr
.data
;
2130 tokc
.str
.size
= tokcstr
.size
;
2131 tokc
.str
.data
= tokcstr
.data
;
2139 /* we use 64 bit numbers */
2142 /* bn = (bn << shift) | or_val */
2143 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
2147 for(i
=0;i
<BN_SIZE
;i
++) {
2149 bn
[i
] = (v
<< shift
) | or_val
;
2150 or_val
= v
>> (32 - shift
);
2154 static void bn_zero(unsigned int *bn
)
2157 for(i
=0;i
<BN_SIZE
;i
++) {
2162 /* parse number in null terminated string 'p' and return it in the
2164 static void parse_number(const char *p
)
2166 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
2168 unsigned int bn
[BN_SIZE
];
2179 goto float_frac_parse
;
2180 } else if (t
== '0') {
2181 if (ch
== 'x' || ch
== 'X') {
2185 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
2191 /* parse all digits. cannot check octal numbers at this stage
2192 because of floating point constants */
2194 if (ch
>= 'a' && ch
<= 'f')
2196 else if (ch
>= 'A' && ch
<= 'F')
2204 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
2206 tcc_error("number too long");
2212 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
2213 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
2215 /* NOTE: strtox should support that for hexa numbers, but
2216 non ISOC99 libcs do not support it, so we prefer to do
2218 /* hexadecimal or binary floats */
2219 /* XXX: handle overflows */
2231 } else if (t
>= 'a') {
2233 } else if (t
>= 'A') {
2238 bn_lshift(bn
, shift
, t
);
2245 if (t
>= 'a' && t
<= 'f') {
2247 } else if (t
>= 'A' && t
<= 'F') {
2249 } else if (t
>= '0' && t
<= '9') {
2255 tcc_error("invalid digit");
2256 bn_lshift(bn
, shift
, t
);
2261 if (ch
!= 'p' && ch
!= 'P')
2268 } else if (ch
== '-') {
2272 if (ch
< '0' || ch
> '9')
2273 expect("exponent digits");
2274 while (ch
>= '0' && ch
<= '9') {
2275 exp_val
= exp_val
* 10 + ch
- '0';
2278 exp_val
= exp_val
* s
;
2280 /* now we can generate the number */
2281 /* XXX: should patch directly float number */
2282 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2283 d
= ldexp(d
, exp_val
- frac_bits
);
2288 /* float : should handle overflow */
2290 } else if (t
== 'L') {
2292 #ifdef TCC_TARGET_PE
2297 /* XXX: not large enough */
2298 tokc
.ld
= (long double)d
;
2305 /* decimal floats */
2307 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2312 while (ch
>= '0' && ch
<= '9') {
2313 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2319 if (ch
== 'e' || ch
== 'E') {
2320 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2324 if (ch
== '-' || ch
== '+') {
2325 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2330 if (ch
< '0' || ch
> '9')
2331 expect("exponent digits");
2332 while (ch
>= '0' && ch
<= '9') {
2333 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2345 tokc
.f
= strtof(token_buf
, NULL
);
2346 } else if (t
== 'L') {
2348 #ifdef TCC_TARGET_PE
2350 tokc
.d
= strtod(token_buf
, NULL
);
2353 tokc
.ld
= strtold(token_buf
, NULL
);
2357 tokc
.d
= strtod(token_buf
, NULL
);
2361 unsigned long long n
, n1
;
2362 int lcount
, ucount
, must_64bit
;
2365 /* integer number */
2368 if (b
== 10 && *q
== '0') {
2375 /* no need for checks except for base 10 / 8 errors */
2385 tcc_error("invalid digit");
2388 /* detect overflow */
2389 /* XXX: this test is not reliable */
2391 tcc_error("integer constant overflow");
2394 /* Determine the characteristics (unsigned and/or 64bit) the type of
2395 the constant must have according to the constant suffix(es) */
2396 lcount
= ucount
= must_64bit
= 0;
2402 tcc_error("three 'l's in integer constant");
2403 if (lcount
&& *(p
- 1) != ch
)
2404 tcc_error("incorrect integer suffix: %s", p1
);
2406 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
2411 } else if (t
== 'U') {
2413 tcc_error("two 'u's in integer constant");
2421 /* Whether 64 bits are needed to hold the constant's value */
2422 if (n
& 0xffffffff00000000LL
|| must_64bit
) {
2430 /* Whether type must be unsigned to hold the constant's value */
2431 if (ucount
|| ((n1
>> 31) && (b
!= 10))) {
2432 if (tok
== TOK_CLLONG
)
2436 /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
2437 } else if (n1
>> 31) {
2438 if (tok
== TOK_CINT
)
2441 tcc_error("integer constant overflow");
2447 tcc_error("invalid number\n");
2451 #define PARSE2(c1, tok1, c2, tok2) \
2462 /* return next token without macro substitution */
2463 static inline void next_nomacro1(void)
2465 int t
, c
, is_long
, len
;
2478 if (parse_flags
& PARSE_FLAG_SPACES
)
2479 goto keep_tok_flags
;
2480 while (isidnum_table
[*p
- CH_EOF
] & IS_SPC
)
2489 /* first look if it is in fact an end of buffer */
2490 c
= handle_stray1(p
);
2497 TCCState
*s1
= tcc_state
;
2498 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2499 && !(tok_flags
& TOK_FLAG_EOF
)) {
2500 tok_flags
|= TOK_FLAG_EOF
;
2502 goto keep_tok_flags
;
2503 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2505 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2506 tcc_error("missing #endif");
2507 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2508 /* no include left : end of file. */
2511 tok_flags
&= ~TOK_FLAG_EOF
;
2512 /* pop include file */
2514 /* test if previous '#endif' was after a #ifdef at
2516 if (tok_flags
& TOK_FLAG_ENDIF
) {
2518 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2520 search_cached_include(s1
, file
->filename
, 1)
2521 ->ifndef_macro
= file
->ifndef_macro_saved
;
2522 tok_flags
&= ~TOK_FLAG_ENDIF
;
2525 /* add end of include file debug info */
2526 if (tcc_state
->do_debug
) {
2527 put_stabd(N_EINCL
, 0, 0);
2529 /* pop include stack */
2531 s1
->include_stack_ptr
--;
2540 tok_flags
|= TOK_FLAG_BOL
;
2543 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2546 goto keep_tok_flags
;
2551 if ((tok_flags
& TOK_FLAG_BOL
) &&
2552 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2554 preprocess(tok_flags
& TOK_FLAG_BOF
);
2560 tok
= TOK_TWOSHARPS
;
2562 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2563 p
= parse_line_comment(p
- 1);
2572 /* dollar is allowed to start identifiers when not parsing asm */
2574 if (!(isidnum_table
[c
- CH_EOF
] & IS_ID
)
2575 || (parse_flags
& PARSE_FLAG_ASM_FILE
))
2578 case 'a': case 'b': case 'c': case 'd':
2579 case 'e': case 'f': case 'g': case 'h':
2580 case 'i': case 'j': case 'k': case 'l':
2581 case 'm': case 'n': case 'o': case 'p':
2582 case 'q': case 'r': case 's': case 't':
2583 case 'u': case 'v': case 'w': case 'x':
2585 case 'A': case 'B': case 'C': case 'D':
2586 case 'E': case 'F': case 'G': case 'H':
2587 case 'I': case 'J': case 'K':
2588 case 'M': case 'N': case 'O': case 'P':
2589 case 'Q': case 'R': case 'S': case 'T':
2590 case 'U': case 'V': case 'W': case 'X':
2596 h
= TOK_HASH_FUNC(h
, c
);
2597 while (c
= *++p
, isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2598 h
= TOK_HASH_FUNC(h
, c
);
2603 /* fast case : no stray found, so we have the full token
2604 and we have already hashed it */
2605 h
&= (TOK_HASH_SIZE
- 1);
2606 pts
= &hash_ident
[h
];
2611 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2613 pts
= &(ts
->hash_next
);
2615 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2619 cstr_reset(&tokcstr
);
2620 cstr_cat(&tokcstr
, (char *) p1
, len
);
2624 while (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2626 cstr_ccat(&tokcstr
, c
);
2629 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2635 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2637 goto parse_ident_fast
;
2640 if (c
== '\'' || c
== '\"') {
2644 cstr_reset(&tokcstr
);
2645 cstr_ccat(&tokcstr
, 'L');
2646 goto parse_ident_slow
;
2651 case '0': case '1': case '2': case '3':
2652 case '4': case '5': case '6': case '7':
2656 /* after the first digit, accept digits, alpha, '.' or sign if
2657 prefixed by 'eEpP' */
2659 cstr_reset(&tokcstr
);
2661 cstr_ccat(&tokcstr
, t
);
2662 if (!((isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2664 || ((c
== '+' || c
== '-')
2665 && (((t
== 'e' || t
== 'E')
2666 && !(parse_flags
& PARSE_FLAG_ASM_FILE
2667 /* 0xe+1 is 3 tokens in asm */
2668 && ((char*)tokcstr
.data
)[0] == '0'
2669 && toup(((char*)tokcstr
.data
)[1]) == 'X'))
2670 || t
== 'p' || t
== 'P'))))
2675 /* We add a trailing '\0' to ease parsing */
2676 cstr_ccat(&tokcstr
, '\0');
2677 tokc
.str
.size
= tokcstr
.size
;
2678 tokc
.str
.data
= tokcstr
.data
;
2683 /* special dot handling because it can also start a number */
2688 } else if ((isidnum_table
['.' - CH_EOF
] & IS_ID
)
2689 && (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))) {
2691 goto parse_ident_fast
;
2692 } else if (c
== '.') {
2698 *--p
= '.'; /* may underflow into file->unget[] */
2709 cstr_reset(&tokcstr
);
2711 cstr_ccat(&tokcstr
, 'L');
2712 cstr_ccat(&tokcstr
, c
);
2713 p
= parse_pp_string(p
, c
, &tokcstr
);
2714 cstr_ccat(&tokcstr
, c
);
2715 cstr_ccat(&tokcstr
, '\0');
2716 tokc
.str
.size
= tokcstr
.size
;
2717 tokc
.str
.data
= tokcstr
.data
;
2726 } else if (c
== '<') {
2743 } else if (c
== '>') {
2761 } else if (c
== '=') {
2774 } else if (c
== '=') {
2787 } else if (c
== '=') {
2800 } else if (c
== '=') {
2803 } else if (c
== '>') {
2811 PARSE2('!', '!', '=', TOK_NE
)
2812 PARSE2('=', '=', '=', TOK_EQ
)
2813 PARSE2('*', '*', '=', TOK_A_MUL
)
2814 PARSE2('%', '%', '=', TOK_A_MOD
)
2815 PARSE2('^', '^', '=', TOK_A_XOR
)
2817 /* comments or operator */
2821 p
= parse_comment(p
);
2822 /* comments replaced by a blank */
2824 goto keep_tok_flags
;
2825 } else if (c
== '/') {
2826 p
= parse_line_comment(p
);
2828 goto keep_tok_flags
;
2829 } else if (c
== '=') {
2849 case '@': /* only used in assembler */
2855 if (c
>= 0x80 && c
<= 0xFF) /* utf8 identifiers */
2856 goto parse_ident_fast
;
2857 if (parse_flags
& PARSE_FLAG_ASM_FILE
)
2859 tcc_error("unrecognized character \\x%02x", c
);
2865 #if defined(PARSE_DEBUG)
2866 printf("token = %s\n", get_tok_str(tok
, &tokc
));
2870 /* return next token without macro substitution. Can read input from
2872 static void next_nomacro_spc(void)
2878 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2879 if (tok
== TOK_LINENUM
) {
2880 file
->line_num
= tokc
.i
;
2889 ST_FUNC
void next_nomacro(void)
2893 } while (tok
< 256 && (isidnum_table
[tok
- CH_EOF
] & IS_SPC
));
2897 static void macro_subst(
2898 TokenString
*tok_str
,
2900 const int *macro_str
,
2904 /* substitute arguments in replacement lists in macro_str by the values in
2905 args (field d) and return allocated string */
2906 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
2918 TOK_GET(&t
, ¯o_str
, &cval
);
2923 TOK_GET(&t
, ¯o_str
, &cval
);
2926 s
= sym_find2(args
, t
);
2929 cstr_ccat(&cstr
, '\"');
2933 TOK_GET(&t
, &st
, &cval
);
2934 if (t
!= TOK_PLCHLDR
2936 && 0 == check_space(t
, &spc
)) {
2937 const char *s
= get_tok_str(t
, &cval
);
2939 if (t
== TOK_PPSTR
&& *s
!= '\'')
2940 add_char(&cstr
, *s
);
2942 cstr_ccat(&cstr
, *s
);
2948 cstr_ccat(&cstr
, '\"');
2949 cstr_ccat(&cstr
, '\0');
2951 printf("\nstringize: <%s>\n", (char *)cstr
.data
);
2954 cval
.str
.size
= cstr
.size
;
2955 cval
.str
.data
= cstr
.data
;
2956 tok_str_add2(&str
, TOK_PPSTR
, &cval
);
2960 expect("macro parameter after '#'");
2962 } else if (t
>= TOK_IDENT
) {
2963 s
= sym_find2(args
, t
);
2967 /* if '##' is present before or after, no arg substitution */
2968 if (*macro_str
== TOK_PPJOIN
|| t1
== TOK_PPJOIN
) {
2969 /* special case for var arg macros : ## eats the ','
2970 if empty VA_ARGS variable. */
2971 if (t1
== TOK_PPJOIN
&& t0
== ',' && gnu_ext
&& s
->type
.t
) {
2973 /* suppress ',' '##' */
2976 /* suppress '##' and add variable */
2983 TOK_GET(&t1
, &st
, &cval
);
2986 tok_str_add2(&str
, t1
, &cval
);
2992 /* NOTE: the stream cannot be read when macro
2993 substituing an argument */
2994 macro_subst(&str
, nested_list
, st
, 0);
2996 if (str
.len
== l0
) /* exanded to empty string */
2997 tok_str_add(&str
, TOK_PLCHLDR
);
2999 tok_str_add(&str
, t
);
3002 tok_str_add2(&str
, t
, &cval
);
3006 tok_str_add(&str
, 0);
3010 static char const ab_month_name
[12][4] =
3012 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
3013 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
3016 /* peek or read [ws_str == NULL] next token from function macro call,
3017 walking up macro levels up to the file if necessary */
3018 static int next_argstream(Sym
**nested_list
, int can_read_stream
, TokenString
*ws_str
)
3026 p
= macro_ptr
, t
= *p
;
3028 while (is_space(t
) || TOK_LINEFEED
== t
|| TOK_PLCHLDR
== t
)
3029 tok_str_add(ws_str
, t
), t
= *++p
;
3031 if (t
== 0 && can_read_stream
) {
3033 /* also, end of scope for nested defined symbol */
3035 while (sa
&& sa
->v
== 0)
3044 while (is_space(ch
) || ch
== '\n' || ch
== '/') {
3047 uint8_t *p
= file
->buf_ptr
;
3050 p
= parse_comment(p
);
3051 file
->buf_ptr
= p
- 1;
3052 } else if (c
== '/') {
3053 p
= parse_line_comment(p
);
3054 file
->buf_ptr
= p
- 1;
3059 if (!(ch
== '\f' || ch
== '\v' || ch
== '\r'))
3060 tok_str_add(ws_str
, ch
);
3074 /* do macro substitution of current token with macro 's' and add
3075 result to (tok_str,tok_len). 'nested_list' is the list of all
3076 macros we got inside to avoid recursing. Return non zero if no
3077 substitution needs to be done */
3078 static int macro_subst_tok(
3079 TokenString
*tok_str
,
3082 int can_read_stream
)
3084 Sym
*args
, *sa
, *sa1
;
3085 int parlevel
, *mstr
, t
, t1
, spc
;
3092 /* if symbol is a macro, prepare substitution */
3093 /* special macros */
3094 if (tok
== TOK___LINE__
) {
3095 snprintf(buf
, sizeof(buf
), "%d", file
->line_num
);
3099 } else if (tok
== TOK___FILE__
) {
3100 cstrval
= file
->filename
;
3102 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
3107 tm
= localtime(&ti
);
3108 if (tok
== TOK___DATE__
) {
3109 snprintf(buf
, sizeof(buf
), "%s %2d %d",
3110 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
3112 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
3113 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
3120 cstr_cat(&cstr
, cstrval
, 0);
3121 cval
.str
.size
= cstr
.size
;
3122 cval
.str
.data
= cstr
.data
;
3123 tok_str_add2(tok_str
, t1
, &cval
);
3126 int saved_parse_flags
= parse_flags
;
3129 if (s
->type
.t
== MACRO_FUNC
) {
3130 /* whitespace between macro name and argument list */
3132 tok_str_new(&ws_str
);
3135 parse_flags
|= PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
3136 | PARSE_FLAG_ACCEPT_STRAYS
;
3138 /* get next token from argument stream */
3139 t
= next_argstream(nested_list
, can_read_stream
, &ws_str
);
3141 /* not a macro substitution after all, restore the
3142 * macro token plus all whitespace we've read.
3143 * whitespace is intentionally not merged to preserve
3145 parse_flags
= saved_parse_flags
;
3146 tok_str_add(tok_str
, tok
);
3147 if (parse_flags
& PARSE_FLAG_SPACES
) {
3149 for (i
= 0; i
< ws_str
.len
; i
++)
3150 tok_str_add(tok_str
, ws_str
.str
[i
]);
3152 tok_str_free_str(ws_str
.str
);
3155 tok_str_free_str(ws_str
.str
);
3158 next_nomacro(); /* eat '(' */
3159 } while (tok
== TOK_PLCHLDR
);
3161 /* argument macro */
3164 /* NOTE: empty args are allowed, except if no args */
3167 next_argstream(nested_list
, can_read_stream
, NULL
);
3168 } while (is_space(tok
) || TOK_LINEFEED
== tok
);
3170 /* handle '()' case */
3171 if (!args
&& !sa
&& tok
== ')')
3174 tcc_error("macro '%s' used with too many args",
3175 get_tok_str(s
->v
, 0));
3178 /* NOTE: non zero sa->t indicates VA_ARGS */
3179 while ((parlevel
> 0 ||
3181 (tok
!= ',' || sa
->type
.t
)))) {
3182 if (tok
== TOK_EOF
|| tok
== 0)
3186 else if (tok
== ')')
3188 if (tok
== TOK_LINEFEED
)
3190 if (!check_space(tok
, &spc
))
3191 tok_str_add2(&str
, tok
, &tokc
);
3192 next_argstream(nested_list
, can_read_stream
, NULL
);
3197 tok_str_add(&str
, 0);
3198 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
3202 /* special case for gcc var args: add an empty
3203 var arg argument if it is omitted */
3204 if (sa
&& sa
->type
.t
&& gnu_ext
)
3212 tcc_error("macro '%s' used with too few args",
3213 get_tok_str(s
->v
, 0));
3216 parse_flags
= saved_parse_flags
;
3218 /* now subst each arg */
3219 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
3224 tok_str_free_str(sa
->d
);
3230 sym_push2(nested_list
, s
->v
, 0, 0);
3231 parse_flags
= saved_parse_flags
;
3232 macro_subst(tok_str
, nested_list
, mstr
, can_read_stream
| 2);
3234 /* pop nested defined symbol */
3236 *nested_list
= sa1
->prev
;
3239 tok_str_free_str(mstr
);
3244 static int paste_tokens(int t1
, CValue
*v1
, int t2
, CValue
*v2
)
3250 if (t1
!= TOK_PLCHLDR
)
3251 cstr_cat(&cstr
, get_tok_str(t1
, v1
), -1);
3253 if (t2
!= TOK_PLCHLDR
)
3254 cstr_cat(&cstr
, get_tok_str(t2
, v2
), -1);
3255 cstr_ccat(&cstr
, '\0');
3257 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3258 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3261 if (0 == *file
->buf_ptr
)
3265 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid"
3266 " preprocessing token", n
, cstr
.data
, (char*)cstr
.data
+ n
);
3271 //printf("paste <%s>\n", (char*)cstr.data);
3276 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
3277 return the resulting string (which must be freed). */
3278 static inline int *macro_twosharps(const int *ptr0
)
3282 TokenString macro_str1
;
3283 int start_of_nosubsts
= -1;
3286 /* we search the first '##' */
3287 for (ptr
= ptr0
;;) {
3288 TOK_GET(&t
, &ptr
, &cval
);
3289 if (t
== TOK_PPJOIN
)
3295 tok_str_new(¯o_str1
);
3297 //tok_print(" $$$", ptr0);
3298 for (ptr
= ptr0
;;) {
3299 TOK_GET(&t
, &ptr
, &cval
);
3302 if (t
== TOK_PPJOIN
)
3304 while (*ptr
== TOK_PPJOIN
) {
3306 /* given 'a##b', remove nosubsts preceding 'a' */
3307 if (start_of_nosubsts
>= 0)
3308 macro_str1
.len
= start_of_nosubsts
;
3309 /* given 'a##b', remove nosubsts preceding 'b' */
3310 while ((t1
= *++ptr
) == TOK_NOSUBST
)
3312 if (t1
&& t1
!= TOK_PPJOIN
) {
3313 TOK_GET(&t1
, &ptr
, &cv1
);
3314 if (t
!= TOK_PLCHLDR
|| t1
!= TOK_PLCHLDR
) {
3315 if (paste_tokens(t
, &cval
, t1
, &cv1
)) {
3316 t
= tok
, cval
= tokc
;
3318 tok_str_add2(¯o_str1
, t
, &cval
);
3324 if (t
== TOK_NOSUBST
) {
3325 if (start_of_nosubsts
< 0)
3326 start_of_nosubsts
= macro_str1
.len
;
3328 start_of_nosubsts
= -1;
3330 tok_str_add2(¯o_str1
, t
, &cval
);
3332 tok_str_add(¯o_str1
, 0);
3333 //tok_print(" ###", macro_str1.str);
3334 return macro_str1
.str
;
3337 /* do macro substitution of macro_str and add result to
3338 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3339 inside to avoid recursing. */
3340 static void macro_subst(
3341 TokenString
*tok_str
,
3343 const int *macro_str
,
3349 int t
, spc
, nosubst
;
3351 int *macro_str1
= NULL
;
3353 /* first scan for '##' operator handling */
3357 /* first scan for '##' operator handling */
3358 if (can_read_stream
) {
3359 macro_str1
= macro_twosharps(ptr
);
3365 TOK_GET(&t
, &ptr
, &cval
);
3369 if (t
>= TOK_IDENT
&& 0 == nosubst
) {
3374 /* if nested substitution, do nothing */
3375 if (sym_find2(*nested_list
, t
)) {
3376 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3377 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3383 str
.str
= (int*)ptr
;
3384 begin_macro(&str
, 2);
3387 macro_subst_tok(tok_str
, nested_list
, s
, can_read_stream
);
3389 if (str
.alloc
== 3) {
3390 /* already finished by reading function macro arguments */
3398 spc
= (tok_str
->len
&&
3399 is_space(tok_last(tok_str
->str
,
3400 tok_str
->str
+ tok_str
->len
)));
3404 if (t
== '\\' && !(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
3405 tcc_error("stray '\\' in program");
3408 if (!check_space(t
, &spc
))
3409 tok_str_add2(tok_str
, t
, &cval
);
3411 if (t
== TOK_NOSUBST
)
3416 tok_str_free_str(macro_str1
);
3420 /* return next token with macro substitution */
3421 ST_FUNC
void next(void)
3424 if (parse_flags
& PARSE_FLAG_SPACES
)
3430 if (tok
== TOK_NOSUBST
|| tok
== TOK_PLCHLDR
) {
3431 /* discard preprocessor markers */
3433 } else if (tok
== 0) {
3434 /* end of macro or unget token string */
3438 } else if (tok
>= TOK_IDENT
&& (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3440 /* if reading from file, try to substitute macros */
3441 s
= define_find(tok
);
3443 Sym
*nested_list
= NULL
;
3445 macro_subst_tok(&tokstr_buf
, &nested_list
, s
, 1);
3446 tok_str_add(&tokstr_buf
, 0);
3447 begin_macro(&tokstr_buf
, 2);
3451 /* convert preprocessor tokens into C tokens */
3452 if (tok
== TOK_PPNUM
) {
3453 if (parse_flags
& PARSE_FLAG_TOK_NUM
)
3454 parse_number((char *)tokc
.str
.data
);
3455 } else if (tok
== TOK_PPSTR
) {
3456 if (parse_flags
& PARSE_FLAG_TOK_STR
)
3457 parse_string((char *)tokc
.str
.data
, tokc
.str
.size
- 1);
3461 /* push back current token and set current token to 'last_tok'. Only
3462 identifier case handled for labels. */
3463 ST_INLN
void unget_tok(int last_tok
)
3466 TokenString
*str
= tok_str_alloc();
3467 tok_str_add2(str
, tok
, &tokc
);
3468 tok_str_add(str
, 0);
3469 begin_macro(str
, 1);
3473 ST_FUNC
void preprocess_start(TCCState
*s1
)
3477 s1
->include_stack_ptr
= s1
->include_stack
;
3478 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3479 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3481 pvtop
= vtop
= vstack
- 1;
3482 s1
->pack_stack
[0] = 0;
3483 s1
->pack_stack_ptr
= s1
->pack_stack
;
3485 set_idnum('$', s1
->dollars_in_identifiers
? IS_ID
: 0);
3486 set_idnum('.', (parse_flags
& PARSE_FLAG_ASM_FILE
) ? IS_ID
: 0);
3488 buf
= tcc_malloc(3 + strlen(file
->filename
));
3489 sprintf(buf
, "\"%s\"", file
->filename
);
3490 tcc_define_symbol(s1
, "__BASE_FILE__", buf
);
3493 if (s1
->nb_cmd_include_files
) {
3497 for (i
= 0; i
< s1
->nb_cmd_include_files
; i
++) {
3498 cstr_cat(&cstr
, "#include \"", -1);
3499 cstr_cat(&cstr
, s1
->cmd_include_files
[i
], -1);
3500 cstr_cat(&cstr
, "\"\n", -1);
3502 *s1
->include_stack_ptr
++ = file
;
3503 tcc_open_bf(s1
, "<command line>", cstr
.size
);
3504 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3509 ST_FUNC
void tccpp_new(TCCState
*s
)
3514 /* might be used in error() before preprocess_start() */
3515 s
->include_stack_ptr
= s
->include_stack
;
3517 /* init isid table */
3518 for(i
= CH_EOF
; i
<128; i
++)
3520 is_space(i
) ? IS_SPC
3525 for(i
= 128; i
<256; i
++)
3526 set_idnum(i
, IS_ID
);
3528 /* init allocators */
3529 tal_new(&toksym_alloc
, TOKSYM_TAL_LIMIT
, TOKSYM_TAL_SIZE
);
3530 tal_new(&tokstr_alloc
, TOKSTR_TAL_LIMIT
, TOKSTR_TAL_SIZE
);
3531 tal_new(&cstr_alloc
, CSTR_TAL_LIMIT
, CSTR_TAL_SIZE
);
3533 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3534 cstr_new(&cstr_buf
);
3535 cstr_realloc(&cstr_buf
, STRING_MAX_SIZE
);
3536 tok_str_new(&tokstr_buf
);
3537 tok_str_realloc(&tokstr_buf
, TOKSTR_MAX_SIZE
);
3539 tok_ident
= TOK_IDENT
;
3548 tok_alloc(p
, r
- p
- 1);
3553 ST_FUNC
void tccpp_delete(TCCState
*s
)
3557 /* free -D and compiler defines */
3560 /* cleanup from error/setjmp */
3569 n
= tok_ident
- TOK_IDENT
;
3570 for(i
= 0; i
< n
; i
++)
3571 tal_free(toksym_alloc
, table_ident
[i
]);
3572 tcc_free(table_ident
);
3575 /* free static buffers */
3576 cstr_free(&tokcstr
);
3577 cstr_free(&cstr_buf
);
3578 cstr_free(¯o_equal_buf
);
3579 tok_str_free_str(tokstr_buf
.str
);
3581 /* free allocators */
3582 tal_delete(toksym_alloc
);
3583 toksym_alloc
= NULL
;
3584 tal_delete(tokstr_alloc
);
3585 tokstr_alloc
= NULL
;
3586 tal_delete(cstr_alloc
);
3590 /* ------------------------------------------------------------------------- */
3591 /* tcc -E [-P[1]] [-dD} support */
3593 static void tok_print(const char *msg
, const int *str
)
3599 fp
= tcc_state
->ppfp
;
3600 if (!fp
|| !tcc_state
->dflag
)
3603 fprintf(fp
, "%s ", msg
);
3605 TOK_GET(&t
, &str
, &cval
);
3608 fprintf(fp
,"%s", get_tok_str(t
, &cval
));
3613 static void pp_line(TCCState
*s1
, BufferedFile
*f
, int level
)
3615 int d
= f
->line_num
- f
->line_ref
;
3620 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_NONE
) {
3622 } else if (level
== 0 && f
->line_ref
&& d
< 8) {
3624 fputs("\n", s1
->ppfp
), --d
;
3625 } else if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_STD
) {
3626 fprintf(s1
->ppfp
, "#line %d \"%s\"\n", f
->line_num
, f
->filename
);
3628 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
,
3629 level
> 0 ? " 1" : level
< 0 ? " 2" : "");
3631 f
->line_ref
= f
->line_num
;
3634 static void define_print(TCCState
*s1
, int v
)
3640 if (NULL
== s
|| NULL
== s
->d
)
3644 fprintf(fp
, "#define %s", get_tok_str(v
, NULL
));
3645 if (s
->type
.t
== MACRO_FUNC
) {
3650 fprintf(fp
,"%s", get_tok_str(a
->v
& ~SYM_FIELD
, NULL
));
3657 tok_print("", s
->d
);
3660 static void pp_debug_defines(TCCState
*s1
)
3671 pp_line(s1
, file
, 0);
3672 file
->line_ref
= ++file
->line_num
;
3676 vs
= get_tok_str(v
, NULL
);
3677 if (t
== TOK_DEFINE
) {
3678 define_print(s1
, v
);
3679 } else if (t
== TOK_UNDEF
) {
3680 fprintf(fp
, "#undef %s\n", vs
);
3681 } else if (t
== TOK_push_macro
) {
3682 fprintf(fp
, "#pragma push_macro(\"%s\")\n", vs
);
3683 } else if (t
== TOK_pop_macro
) {
3684 fprintf(fp
, "#pragma pop_macro(\"%s\")\n", vs
);
3689 static void pp_debug_builtins(TCCState
*s1
)
3692 for (v
= TOK_IDENT
; v
< tok_ident
; ++v
)
3693 define_print(s1
, v
);
3696 /* Add a space between tokens a and b to avoid unwanted textual pasting */
3697 static int pp_need_space(int a
, int b
)
3699 return 'E' == a
? '+' == b
|| '-' == b
3700 : '+' == a
? TOK_INC
== b
|| '+' == b
3701 : '-' == a
? TOK_DEC
== b
|| '-' == b
3702 : a
>= TOK_IDENT
? b
>= TOK_IDENT
3706 /* maybe hex like 0x1e */
3707 static int pp_check_he0xE(int t
, const char *p
)
3709 if (t
== TOK_PPNUM
&& toup(strchr(p
, 0)[-1]) == 'E')
3714 /* Preprocess the current file */
3715 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3717 BufferedFile
**iptr
;
3718 int token_seen
, spcs
, level
;
3722 define_start
= define_stack
;
3723 preprocess_start(s1
);
3725 ch
= file
->buf_ptr
[0];
3726 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3727 parse_flags
= PARSE_FLAG_PREPROCESS
3728 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
3729 | PARSE_FLAG_LINEFEED
3731 | PARSE_FLAG_ACCEPT_STRAYS
3734 /* Credits to Fabrice Bellard's initial revision to demonstrate its
3735 capability to compile and run itself, provided all numbers are
3736 given as decimals. tcc -E -P10 will do. */
3737 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_P10
)
3738 parse_flags
|= PARSE_FLAG_TOK_NUM
, s1
->Pflag
= 1;
3741 /* for PP benchmarks */
3742 do next(); while (tok
!= TOK_EOF
);
3743 free_defines(define_start
);
3747 if (s1
->dflag
& 1) {
3748 pp_debug_builtins(s1
);
3752 token_seen
= TOK_LINEFEED
, spcs
= 0;
3753 pp_line(s1
, file
, 0);
3756 iptr
= s1
->include_stack_ptr
;
3760 level
= s1
->include_stack_ptr
- iptr
;
3763 pp_line(s1
, *iptr
, 0);
3764 pp_line(s1
, file
, level
);
3768 pp_debug_defines(s1
);
3773 if (token_seen
== TOK_LINEFEED
) {
3778 if (tok
== TOK_LINEFEED
) {
3782 pp_line(s1
, file
, 0);
3783 } else if (tok
== TOK_LINEFEED
) {
3786 spcs
= pp_need_space(token_seen
, tok
);
3790 fputs(" ", s1
->ppfp
), --spcs
;
3791 fputs(p
= get_tok_str(tok
, &tokc
), s1
->ppfp
);
3792 token_seen
= pp_check_he0xE(tok
, p
);;
3794 /* reset define stack, but keep -D and built-ins */
3795 free_defines(define_start
);
3799 /* ------------------------------------------------------------------------- */