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
24 /* #define to 1 to enable (see parse_pp_string()) */
25 #define ACCEPT_LF_IN_STRINGS 0
27 /********************************************************/
28 /* global variables */
30 ST_DATA
int tok_flags
;
31 ST_DATA
int parse_flags
;
33 ST_DATA
struct BufferedFile
*file
;
36 ST_DATA
const int *macro_ptr
;
37 ST_DATA CString tokcstr
; /* current parsed string, if any */
39 /* display benchmark infos */
40 ST_DATA
int tok_ident
;
41 ST_DATA TokenSym
**table_ident
;
43 /* ------------------------------------------------------------------------- */
45 static TokenSym
*hash_ident
[TOK_HASH_SIZE
];
46 static char token_buf
[STRING_MAX_SIZE
+ 1];
47 static CString cstr_buf
;
48 static CString macro_equal_buf
;
49 static TokenString tokstr_buf
;
50 static unsigned char isidnum_table
[256 - CH_EOF
];
51 static int pp_debug_tok
, pp_debug_symv
;
54 static int pp_counter
;
55 static void tok_print(const char *msg
, const int *str
);
57 static struct TinyAlloc
*toksym_alloc
;
58 static struct TinyAlloc
*tokstr_alloc
;
60 static TokenString
*macro_stack
;
62 static const char tcc_keywords
[] =
63 #define DEF(id, str) str "\0"
68 /* WARNING: the content of this string encodes token numbers */
69 static const unsigned char tok_two_chars
[] =
71 "<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
72 "-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";
94 '#','#', TOK_TWOSHARPS
,
99 static void next_nomacro(void);
101 ST_FUNC
void skip(int c
)
104 tcc_error("'%c' expected (got \"%s\")", c
, get_tok_str(tok
, &tokc
));
108 ST_FUNC
void expect(const char *msg
)
110 tcc_error("%s expected", msg
);
113 /* ------------------------------------------------------------------------- */
114 /* Custom allocator for tiny objects */
119 #define tal_free(al, p) tcc_free(p)
120 #define tal_realloc(al, p, size) tcc_realloc(p, size)
121 #define tal_new(a,b,c)
122 #define tal_delete(a)
124 #if !defined(MEM_DEBUG)
125 #define tal_free(al, p) tal_free_impl(al, p)
126 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size)
127 #define TAL_DEBUG_PARAMS
130 //#define TAL_INFO 1 /* collect and dump allocators stats */
131 #define tal_free(al, p) tal_free_impl(al, p, __FILE__, __LINE__)
132 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size, __FILE__, __LINE__)
133 #define TAL_DEBUG_PARAMS , const char *file, int line
134 #define TAL_DEBUG_FILE_LEN 40
137 #define TOKSYM_TAL_SIZE (768 * 1024) /* allocator for tiny TokenSym in table_ident */
138 #define TOKSTR_TAL_SIZE (768 * 1024) /* allocator for tiny TokenString instances */
139 #define CSTR_TAL_SIZE (256 * 1024) /* allocator for tiny CString instances */
140 #define TOKSYM_TAL_LIMIT 256 /* prefer unique limits to distinguish allocators debug msgs */
141 #define TOKSTR_TAL_LIMIT 128 /* 32 * sizeof(int) */
142 #define CSTR_TAL_LIMIT 1024
144 typedef struct TinyAlloc
{
150 struct TinyAlloc
*next
, *top
;
159 typedef struct tal_header_t
{
162 int line_num
; /* negative line_num used for double free check */
163 char file_name
[TAL_DEBUG_FILE_LEN
+ 1];
167 /* ------------------------------------------------------------------------- */
169 static TinyAlloc
*tal_new(TinyAlloc
**pal
, unsigned limit
, unsigned size
)
171 TinyAlloc
*al
= tcc_mallocz(sizeof(TinyAlloc
));
172 al
->p
= al
->buffer
= tcc_malloc(size
);
179 static void tal_delete(TinyAlloc
*al
)
187 fprintf(stderr
, "limit=%5d, size=%5g MB, nb_peak=%6d, nb_total=%8d, nb_missed=%6d, usage=%5.1f%%\n",
188 al
->limit
, al
->size
/ 1024.0 / 1024.0, al
->nb_peak
, al
->nb_total
, al
->nb_missed
,
189 (al
->peak_p
- al
->buffer
) * 100.0 / al
->size
);
192 if (al
->nb_allocs
> 0) {
194 fprintf(stderr
, "TAL_DEBUG: memory leak %d chunk(s) (limit= %d)\n",
195 al
->nb_allocs
, al
->limit
);
198 tal_header_t
*header
= (tal_header_t
*)p
;
199 if (header
->line_num
> 0) {
200 fprintf(stderr
, "%s:%d: chunk of %d bytes leaked\n",
201 header
->file_name
, header
->line_num
, header
->size
);
203 p
+= header
->size
+ sizeof(tal_header_t
);
211 tcc_free(al
->buffer
);
217 static void tal_free_impl(TinyAlloc
*al
, void *p TAL_DEBUG_PARAMS
)
222 if (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
) {
224 tal_header_t
*header
= (((tal_header_t
*)p
) - 1);
225 if (header
->line_num
< 0) {
226 fprintf(stderr
, "%s:%d: TAL_DEBUG: double frees chunk from\n",
228 fprintf(stderr
, "%s:%d: %d bytes\n",
229 header
->file_name
, (int)-header
->line_num
, (int)header
->size
);
231 header
->line_num
= -header
->line_num
;
236 } else if (al
->next
) {
244 static void *tal_realloc_impl(TinyAlloc
**pal
, void *p
, unsigned size TAL_DEBUG_PARAMS
)
246 tal_header_t
*header
;
249 unsigned adj_size
= (size
+ 3) & -4;
250 TinyAlloc
*al
= *pal
;
253 is_own
= (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
);
254 if ((!p
|| is_own
) && size
<= al
->limit
) {
255 if (al
->p
- al
->buffer
+ adj_size
+ sizeof(tal_header_t
) < al
->size
) {
256 header
= (tal_header_t
*)al
->p
;
257 header
->size
= adj_size
;
259 { int ofs
= strlen(file
) - TAL_DEBUG_FILE_LEN
;
260 strncpy(header
->file_name
, file
+ (ofs
> 0 ? ofs
: 0), TAL_DEBUG_FILE_LEN
);
261 header
->file_name
[TAL_DEBUG_FILE_LEN
] = 0;
262 header
->line_num
= line
; }
264 ret
= al
->p
+ sizeof(tal_header_t
);
265 al
->p
+= adj_size
+ sizeof(tal_header_t
);
267 header
= (((tal_header_t
*)p
) - 1);
268 if (p
) memcpy(ret
, p
, header
->size
);
270 header
->line_num
= -header
->line_num
;
276 if (al
->nb_peak
< al
->nb_allocs
)
277 al
->nb_peak
= al
->nb_allocs
;
278 if (al
->peak_p
< al
->p
)
285 ret
= tal_realloc(*pal
, 0, size
);
286 header
= (((tal_header_t
*)p
) - 1);
287 if (p
) memcpy(ret
, p
, header
->size
);
289 header
->line_num
= -header
->line_num
;
296 TinyAlloc
*bottom
= al
, *next
= al
->top
? al
->top
: al
;
298 al
= tal_new(pal
, next
->limit
, next
->size
* 2);
306 ret
= tcc_malloc(size
);
307 header
= (((tal_header_t
*)p
) - 1);
308 if (p
) memcpy(ret
, p
, header
->size
);
310 header
->line_num
= -header
->line_num
;
312 } else if (al
->next
) {
316 ret
= tcc_realloc(p
, size
);
325 /* ------------------------------------------------------------------------- */
326 /* CString handling */
327 static void cstr_realloc(CString
*cstr
, int new_size
)
331 size
= cstr
->size_allocated
;
333 size
= 8; /* no need to allocate a too small first string */
334 while (size
< new_size
)
336 cstr
->data
= tcc_realloc(cstr
->data
, size
);
337 cstr
->size_allocated
= size
;
341 ST_INLN
void cstr_ccat(CString
*cstr
, int ch
)
344 size
= cstr
->size
+ 1;
345 if (size
> cstr
->size_allocated
)
346 cstr_realloc(cstr
, size
);
347 ((unsigned char *)cstr
->data
)[size
- 1] = ch
;
351 ST_INLN
char *unicode_to_utf8 (char *b
, uint32_t Uc
)
353 if (Uc
<0x80) *b
++=Uc
;
354 else if (Uc
<0x800) *b
++=192+Uc
/64, *b
++=128+Uc
%64;
355 else if (Uc
-0xd800u
<0x800) goto error
;
356 else if (Uc
<0x10000) *b
++=224+Uc
/4096, *b
++=128+Uc
/64%64, *b
++=128+Uc
%64;
357 else if (Uc
<0x110000) *b
++=240+Uc
/262144, *b
++=128+Uc
/4096%64, *b
++=128+Uc
/64%64, *b
++=128+Uc
%64;
358 else error
: tcc_error("0x%x is not a valid universal character", Uc
);
362 /* add a unicode character expanded into utf8 */
363 ST_INLN
void cstr_u8cat(CString
*cstr
, int ch
)
366 e
= unicode_to_utf8(buf
, (uint32_t)ch
);
367 cstr_cat(cstr
, buf
, e
- buf
);
370 ST_FUNC
void cstr_cat(CString
*cstr
, const char *str
, int len
)
374 len
= strlen(str
) + 1 + len
;
375 size
= cstr
->size
+ len
;
376 if (size
> cstr
->size_allocated
)
377 cstr_realloc(cstr
, size
);
378 memmove(((unsigned char *)cstr
->data
) + cstr
->size
, str
, len
);
382 /* add a wide char */
383 ST_FUNC
void cstr_wccat(CString
*cstr
, int ch
)
386 size
= cstr
->size
+ sizeof(nwchar_t
);
387 if (size
> cstr
->size_allocated
)
388 cstr_realloc(cstr
, size
);
389 *(nwchar_t
*)(((unsigned char *)cstr
->data
) + size
- sizeof(nwchar_t
)) = ch
;
393 ST_FUNC
void cstr_new(CString
*cstr
)
395 memset(cstr
, 0, sizeof(CString
));
398 /* free string and reset it to NULL */
399 ST_FUNC
void cstr_free(CString
*cstr
)
401 tcc_free(cstr
->data
);
405 /* reset string to empty */
406 ST_FUNC
void cstr_reset(CString
*cstr
)
411 ST_FUNC
int cstr_vprintf(CString
*cstr
, const char *fmt
, va_list ap
)
417 if (size
> cstr
->size_allocated
)
418 cstr_realloc(cstr
, size
);
419 size
= cstr
->size_allocated
- cstr
->size
;
421 len
= vsnprintf((char*)cstr
->data
+ cstr
->size
, size
, fmt
, v
);
423 if (len
>= 0 && len
< size
)
431 ST_FUNC
int cstr_printf(CString
*cstr
, const char *fmt
, ...)
435 len
= cstr_vprintf(cstr
, fmt
, ap
);
441 static void add_char(CString
*cstr
, int c
)
443 if (c
== '\'' || c
== '\"' || c
== '\\') {
444 /* XXX: could be more precise if char or string */
445 cstr_ccat(cstr
, '\\');
447 if (c
>= 32 && c
<= 126) {
450 cstr_ccat(cstr
, '\\');
452 cstr_ccat(cstr
, 'n');
454 cstr_ccat(cstr
, '0' + ((c
>> 6) & 7));
455 cstr_ccat(cstr
, '0' + ((c
>> 3) & 7));
456 cstr_ccat(cstr
, '0' + (c
& 7));
461 /* ------------------------------------------------------------------------- */
462 /* allocate a new token */
463 static TokenSym
*tok_alloc_new(TokenSym
**pts
, const char *str
, int len
)
465 TokenSym
*ts
, **ptable
;
468 if (tok_ident
>= SYM_FIRST_ANOM
)
469 tcc_error("memory full (symbols)");
471 /* expand token table if needed */
472 i
= tok_ident
- TOK_IDENT
;
473 if ((i
% TOK_ALLOC_INCR
) == 0) {
474 ptable
= tcc_realloc(table_ident
, (i
+ TOK_ALLOC_INCR
) * sizeof(TokenSym
*));
475 table_ident
= ptable
;
478 ts
= tal_realloc(toksym_alloc
, 0, sizeof(TokenSym
) + len
);
480 ts
->tok
= tok_ident
++;
481 ts
->sym_define
= NULL
;
482 ts
->sym_label
= NULL
;
483 ts
->sym_struct
= NULL
;
484 ts
->sym_identifier
= NULL
;
486 ts
->hash_next
= NULL
;
487 memcpy(ts
->str
, str
, len
);
493 #define TOK_HASH_INIT 1
494 #define TOK_HASH_FUNC(h, c) ((h) + ((h) << 5) + ((h) >> 27) + (c))
497 /* find a token and add it if not found */
498 ST_FUNC TokenSym
*tok_alloc(const char *str
, int len
)
506 h
= TOK_HASH_FUNC(h
, ((unsigned char *)str
)[i
]);
507 h
&= (TOK_HASH_SIZE
- 1);
509 pts
= &hash_ident
[h
];
514 if (ts
->len
== len
&& !memcmp(ts
->str
, str
, len
))
516 pts
= &(ts
->hash_next
);
518 return tok_alloc_new(pts
, str
, len
);
521 ST_FUNC
int tok_alloc_const(const char *str
)
523 return tok_alloc(str
, strlen(str
))->tok
;
527 /* XXX: buffer overflow */
528 /* XXX: float tokens */
529 ST_FUNC
const char *get_tok_str(int v
, CValue
*cv
)
534 cstr_reset(&cstr_buf
);
544 /* XXX: not quite exact, but only useful for testing */
546 sprintf(p
, "%u", (unsigned)cv
->i
);
548 sprintf(p
, "%llu", (unsigned long long)cv
->i
);
552 cstr_ccat(&cstr_buf
, 'L');
554 cstr_ccat(&cstr_buf
, '\'');
555 add_char(&cstr_buf
, cv
->i
);
556 cstr_ccat(&cstr_buf
, '\'');
557 cstr_ccat(&cstr_buf
, '\0');
561 return (char*)cv
->str
.data
;
563 cstr_ccat(&cstr_buf
, 'L');
565 cstr_ccat(&cstr_buf
, '\"');
567 len
= cv
->str
.size
- 1;
569 add_char(&cstr_buf
, ((unsigned char *)cv
->str
.data
)[i
]);
571 len
= (cv
->str
.size
/ sizeof(nwchar_t
)) - 1;
573 add_char(&cstr_buf
, ((nwchar_t
*)cv
->str
.data
)[i
]);
575 cstr_ccat(&cstr_buf
, '\"');
576 cstr_ccat(&cstr_buf
, '\0');
580 cstr_cat(&cstr_buf
, "<float>", 0);
583 cstr_cat(&cstr_buf
, "<double>", 0);
586 cstr_cat(&cstr_buf
, "<long double>", 0);
589 cstr_cat(&cstr_buf
, "<linenumber>", 0);
592 /* above tokens have value, the ones below don't */
600 return strcpy(p
, "...");
602 return strcpy(p
, "<<=");
604 return strcpy(p
, ">>=");
606 return strcpy(p
, "<eof>");
609 /* search in two bytes table */
610 const unsigned char *q
= tok_two_chars
;
616 return cstr_buf
.data
;
621 sprintf(cstr_buf
.data
, "<%02x>", v
);
622 return cstr_buf
.data
;
626 case 0: /* nameless anonymous symbol */
628 } else if (v
< tok_ident
) {
629 return table_ident
[v
- TOK_IDENT
]->str
;
630 } else if (v
>= SYM_FIRST_ANOM
) {
631 /* special name for anonymous symbol */
632 sprintf(p
, "L.%u", v
- SYM_FIRST_ANOM
);
634 /* should never happen */
639 return cstr_buf
.data
;
642 static inline int check_space(int t
, int *spc
)
644 if (t
< 256 && (isidnum_table
[t
- CH_EOF
] & IS_SPC
)) {
653 /* return the current character, handling end of block if necessary
655 static int handle_eob(void)
657 BufferedFile
*bf
= file
;
660 /* only tries to read if really end of buffer */
661 if (bf
->buf_ptr
>= bf
->buf_end
) {
663 #if defined(PARSE_DEBUG)
668 len
= read(bf
->fd
, bf
->buffer
, len
);
675 bf
->buf_ptr
= bf
->buffer
;
676 bf
->buf_end
= bf
->buffer
+ len
;
677 *bf
->buf_end
= CH_EOB
;
679 if (bf
->buf_ptr
< bf
->buf_end
) {
680 return bf
->buf_ptr
[0];
682 bf
->buf_ptr
= bf
->buf_end
;
687 /* read next char from current input file and handle end of input buffer */
688 static int next_c(void)
690 int ch
= *++file
->buf_ptr
;
691 /* end of buffer/file handling */
692 if (ch
== CH_EOB
&& file
->buf_ptr
>= file
->buf_end
)
697 /* input with '\[\r]\n' handling. */
698 static int handle_stray_noerror(int err
)
701 while ((ch
= next_c()) == '\\') {
711 *--file
->buf_ptr
= '\r';
714 tcc_error("stray '\\' in program");
715 /* may take advantage of 'BufferedFile.unget[4}' */
716 return *--file
->buf_ptr
= '\\';
722 #define ninp() handle_stray_noerror(0)
724 /* handle '\\' in strings, comments and skipped regions */
725 static int handle_bs(uint8_t **p
)
728 file
->buf_ptr
= *p
- 1;
734 /* skip the stray and handle the \\n case. Output an error if
735 incorrect char after the stray */
736 static int handle_stray(uint8_t **p
)
739 file
->buf_ptr
= *p
- 1;
740 c
= handle_stray_noerror(!(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
));
745 /* handle the complicated stray case */
750 c = handle_stray(&p); \
753 static int skip_spaces(void)
759 } while (isidnum_table
[ch
- CH_EOF
] & IS_SPC
);
763 /* single line C++ comments */
764 static uint8_t *parse_line_comment(uint8_t *p
)
771 if (c
== '\n' || c
== '\\')
774 if (c
== '\n' || c
== '\\')
789 static uint8_t *parse_comment(uint8_t *p
)
797 if (c
== '\n' || c
== '*' || c
== '\\')
800 if (c
== '\n' || c
== '*' || c
== '\\')
803 /* now we can handle all the cases */
806 } else if (c
== '*') {
819 tcc_error("unexpected end of file in comment");
827 /* parse a string without interpreting escapes */
828 static uint8_t *parse_pp_string(uint8_t *p
, int sep
, CString
*str
)
836 } else if (c
== '\\') {
840 /* XXX: indicate line number of start of string */
841 tok_flags
&= ~TOK_FLAG_BOL
;
842 tcc_error("missing terminating %c character", sep
);
843 } else if (c
== '\\') {
847 /* add char after '\\' unconditionally */
851 goto unterminated_string
;
857 } else if (c
== '\n') {
859 if (ACCEPT_LF_IN_STRINGS
) {
862 } else if (str
) { /* not skipping */
863 goto unterminated_string
;
865 //tcc_warning("missing terminating %c character", sep);
868 } else if (c
== '\r') {
875 goto unterminated_string
;
877 cstr_ccat(str
, '\r');
889 /* skip block of text until #else, #elif or #endif. skip also pairs of
891 static void preprocess_skip(void)
893 int a
, start_of_line
, c
, in_warn_or_error
;
900 in_warn_or_error
= 0;
926 if (in_warn_or_error
)
928 tok_flags
&= ~TOK_FLAG_BOL
;
929 p
= parse_pp_string(p
, c
, NULL
);
933 if (in_warn_or_error
)
938 p
= parse_comment(p
);
939 } else if (c
== '/') {
940 p
= parse_line_comment(p
);
950 (tok
== TOK_ELSE
|| tok
== TOK_ELIF
|| tok
== TOK_ENDIF
))
952 if (tok
== TOK_IF
|| tok
== TOK_IFDEF
|| tok
== TOK_IFNDEF
)
954 else if (tok
== TOK_ENDIF
)
956 else if( tok
== TOK_ERROR
|| tok
== TOK_WARNING
)
957 in_warn_or_error
= 1;
958 else if (tok
== TOK_LINEFEED
)
960 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
961 p
= parse_line_comment(p
- 1);
963 #if !defined(TCC_TARGET_ARM)
964 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
965 p
= parse_line_comment(p
- 1);
967 /* ARM assembly uses '#' for constants */
982 /* return the number of additional 'ints' necessary to store the
984 static inline int tok_size(const int *p
)
999 return 1 + ((sizeof(CString
) + ((CString
*)(p
+1))->size
+ 3) >> 2);
1002 return 1 + LONG_SIZE
/ 4;
1008 return 1 + LDOUBLE_SIZE
/ 4;
1015 /* token string handling */
1016 ST_INLN
void tok_str_new(TokenString
*s
)
1019 s
->len
= s
->lastlen
= 0;
1020 s
->allocated_len
= 0;
1021 s
->last_line_num
= -1;
1024 ST_FUNC TokenString
*tok_str_alloc(void)
1026 TokenString
*str
= tal_realloc(tokstr_alloc
, 0, sizeof *str
);
1031 ST_FUNC
int *tok_str_dup(TokenString
*s
)
1035 str
= tal_realloc(tokstr_alloc
, 0, s
->len
* sizeof(int));
1036 memcpy(str
, s
->str
, s
->len
* sizeof(int));
1040 ST_FUNC
void tok_str_free_str(int *str
)
1042 tal_free(tokstr_alloc
, str
);
1045 ST_FUNC
void tok_str_free(TokenString
*str
)
1047 tok_str_free_str(str
->str
);
1048 tal_free(tokstr_alloc
, str
);
1051 ST_FUNC
int *tok_str_realloc(TokenString
*s
, int new_size
)
1055 size
= s
->allocated_len
;
1058 while (size
< new_size
)
1060 if (size
> s
->allocated_len
) {
1061 str
= tal_realloc(tokstr_alloc
, s
->str
, size
* sizeof(int));
1062 s
->allocated_len
= size
;
1068 ST_FUNC
void tok_str_add(TokenString
*s
, int t
)
1074 if (len
>= s
->allocated_len
)
1075 str
= tok_str_realloc(s
, len
+ 1);
1080 ST_FUNC
void begin_macro(TokenString
*str
, int alloc
)
1083 str
->prev
= macro_stack
;
1084 str
->prev_ptr
= macro_ptr
;
1085 str
->save_line_num
= file
->line_num
;
1086 macro_ptr
= str
->str
;
1090 ST_FUNC
void end_macro(void)
1092 TokenString
*str
= macro_stack
;
1093 macro_stack
= str
->prev
;
1094 macro_ptr
= str
->prev_ptr
;
1095 file
->line_num
= str
->save_line_num
;
1096 str
->len
= 0; /* matters if str not alloced, may be tokstr_buf */
1097 if (str
->alloc
!= 0) {
1098 if (str
->alloc
== 2)
1099 str
->str
= NULL
; /* don't free */
1104 static void tok_str_add2(TokenString
*s
, int t
, CValue
*cv
)
1108 len
= s
->lastlen
= s
->len
;
1111 /* allocate space for worst case */
1112 if (len
+ TOK_MAX_SIZE
>= s
->allocated_len
)
1113 str
= tok_str_realloc(s
, len
+ TOK_MAX_SIZE
+ 1);
1126 str
[len
++] = cv
->tab
[0];
1133 /* Insert the string into the int array. */
1135 1 + (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1136 if (len
+ nb_words
>= s
->allocated_len
)
1137 str
= tok_str_realloc(s
, len
+ nb_words
+ 1);
1138 str
[len
] = cv
->str
.size
;
1139 memcpy(&str
[len
+ 1], cv
->str
.data
, cv
->str
.size
);
1150 #if LDOUBLE_SIZE == 8
1153 str
[len
++] = cv
->tab
[0];
1154 str
[len
++] = cv
->tab
[1];
1156 #if LDOUBLE_SIZE == 12
1158 str
[len
++] = cv
->tab
[0];
1159 str
[len
++] = cv
->tab
[1];
1160 str
[len
++] = cv
->tab
[2];
1161 #elif LDOUBLE_SIZE == 16
1163 str
[len
++] = cv
->tab
[0];
1164 str
[len
++] = cv
->tab
[1];
1165 str
[len
++] = cv
->tab
[2];
1166 str
[len
++] = cv
->tab
[3];
1167 #elif LDOUBLE_SIZE != 8
1168 #error add long double size support
1177 /* add the current parse token in token string 's' */
1178 ST_FUNC
void tok_str_add_tok(TokenString
*s
)
1182 /* save line number info */
1183 if (file
->line_num
!= s
->last_line_num
) {
1184 s
->last_line_num
= file
->line_num
;
1185 cval
.i
= s
->last_line_num
;
1186 tok_str_add2(s
, TOK_LINENUM
, &cval
);
1188 tok_str_add2(s
, tok
, &tokc
);
1191 /* get a token from an integer array and increment pointer. */
1192 static inline void tok_get(int *t
, const int **pp
, CValue
*cv
)
1212 cv
->i
= (unsigned)*p
++;
1221 cv
->str
.size
= *p
++;
1223 p
+= (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1235 #if LDOUBLE_SIZE == 16
1237 #elif LDOUBLE_SIZE == 12
1239 #elif LDOUBLE_SIZE == 8
1242 # error add long double size support
1256 # define TOK_GET(t,p,c) tok_get(t,p,c)
1258 # define TOK_GET(t,p,c) do { \
1260 if (TOK_HAS_VALUE(_t)) \
1263 *(t) = _t, ++*(p); \
1267 static int macro_is_equal(const int *a
, const int *b
)
1276 /* first time preallocate macro_equal_buf, next time only reset position to start */
1277 cstr_reset(¯o_equal_buf
);
1278 TOK_GET(&t
, &a
, &cv
);
1279 cstr_cat(¯o_equal_buf
, get_tok_str(t
, &cv
), 0);
1280 TOK_GET(&t
, &b
, &cv
);
1281 if (strcmp(macro_equal_buf
.data
, get_tok_str(t
, &cv
)))
1287 /* defines handling */
1288 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1293 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1295 s
->next
= first_arg
;
1296 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1298 if (o
&& !macro_is_equal(o
->d
, s
->d
))
1299 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1302 /* undefined a define symbol. Its name is just set to zero */
1303 ST_FUNC
void define_undef(Sym
*s
)
1306 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1307 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1310 ST_INLN Sym
*define_find(int v
)
1313 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1315 return table_ident
[v
]->sym_define
;
1318 /* free define stack until top reaches 'b' */
1319 ST_FUNC
void free_defines(Sym
*b
)
1321 while (define_stack
!= b
) {
1322 Sym
*top
= define_stack
;
1323 define_stack
= top
->prev
;
1324 tok_str_free_str(top
->d
);
1331 ST_FUNC Sym
*label_find(int v
)
1334 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1336 return table_ident
[v
]->sym_label
;
1339 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1342 s
= sym_push2(ptop
, v
, VT_STATIC
, 0);
1344 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1345 if (ptop
== &global_label_stack
) {
1346 /* modify the top most local identifier, so that
1347 sym_identifier will point to 's' when popped */
1349 ps
= &(*ps
)->prev_tok
;
1356 /* pop labels until element last is reached. Look if any labels are
1357 undefined. Define symbols if '&&label' was used. */
1358 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
, int keep
)
1361 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1363 if (s
->r
== LABEL_DECLARED
) {
1364 tcc_warning_c(warn_all
)("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1365 } else if (s
->r
== LABEL_FORWARD
) {
1366 tcc_error("label '%s' used but not defined",
1367 get_tok_str(s
->v
, NULL
));
1370 /* define corresponding symbol. A size of
1372 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1376 if (s
->r
!= LABEL_GONE
)
1377 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1387 /* fake the nth "#if defined test_..." for tcc -dt -run */
1388 static void maybe_run_test(TCCState
*s
)
1391 if (s
->include_stack_ptr
!= s
->include_stack
)
1393 p
= get_tok_str(tok
, NULL
);
1394 if (0 != memcmp(p
, "test_", 5))
1396 if (0 != --s
->run_test
)
1398 fprintf(s
->ppfp
, &"\n[%s]\n"[!(s
->dflag
& 32)], p
), fflush(s
->ppfp
);
1399 define_push(tok
, MACRO_OBJ
, NULL
, NULL
);
1402 static CachedInclude
*
1403 search_cached_include(TCCState
*s1
, const char *filename
, int add
);
1405 static int parse_include(TCCState
*s1
, int do_next
, int test
)
1409 char name
[1024], buf
[1024], *p
;
1414 if (c
== '<' || c
== '\"') {
1415 file
->buf_ptr
= parse_pp_string(file
->buf_ptr
, c
== '<' ? '>' : c
, &cs
);
1418 /* computed #include : concatenate tokens until result is one of
1419 the two accepted forms. Don't convert pp-tokens to tokens here. */
1420 parse_flags
= PARSE_FLAG_PREPROCESS
1421 | PARSE_FLAG_LINEFEED
1422 | (parse_flags
& PARSE_FLAG_ASM_FILE
);
1425 p
= cs
.data
, i
= cs
.size
- 1;
1427 && ((p
[0] == '"' && p
[i
] == '"')
1428 || (p
[0] == '<' && p
[i
] == '>')))
1430 if (tok
== TOK_LINEFEED
)
1431 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1432 cstr_cat(&cs
, get_tok_str(tok
, &tokc
), -1);
1435 /* remove '<>|""' */
1436 memmove(p
, p
+ 1, cs
.size
-= 2);
1438 cstr_ccat(&cs
, '\0');
1439 pstrcpy(name
, sizeof name
, cs
.data
);
1442 i
= do_next
? file
->include_next_index
: -1;
1446 /* check absolute include path */
1447 if (!IS_ABSPATH(name
))
1450 } else if (i
== 1) {
1451 /* search in file's dir if "header.h" */
1454 p
= file
->true_filename
;
1455 pstrncpy(buf
, p
, tcc_basename(p
) - p
);
1457 int j
= i
- 2, k
= j
- s1
->nb_include_paths
;
1459 p
= s1
->include_paths
[j
];
1460 else if (k
< s1
->nb_sysinclude_paths
)
1461 p
= s1
->sysinclude_paths
[k
];
1465 tcc_error("include file '%s' not found", name
);
1466 pstrcpy(buf
, sizeof buf
, p
);
1467 pstrcat(buf
, sizeof buf
, "/");
1469 pstrcat(buf
, sizeof buf
, name
);
1470 e
= search_cached_include(s1
, buf
, 0);
1471 if (e
&& (define_find(e
->ifndef_macro
) || e
->once
== pp_once
)) {
1472 /* no need to parse the include because the 'ifndef macro'
1473 is defined (or had #pragma once) */
1475 printf("%s: skipping cached %s\n", file
->filename
, buf
);
1479 if (tcc_open(s1
, buf
) >= 0)
1486 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1487 tcc_error("#include recursion too deep");
1488 /* push previous file on stack */
1489 *s1
->include_stack_ptr
++ = file
->prev
;
1490 file
->include_next_index
= i
;
1492 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1494 /* update target deps */
1496 BufferedFile
*bf
= file
;
1497 while (i
== 1 && (bf
= bf
->prev
))
1498 i
= bf
->include_next_index
;
1499 /* skip system include files */
1500 if (s1
->include_sys_deps
|| i
- 2 < s1
->nb_include_paths
)
1501 dynarray_add(&s1
->target_deps
, &s1
->nb_target_deps
,
1504 /* add include file debug info */
1505 tcc_debug_bincl(s1
);
1506 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1511 /* eval an expression for #if/#elif */
1512 static int expr_preprocess(TCCState
*s1
)
1517 str
= tok_str_alloc();
1519 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1520 next(); /* do macro subst */
1522 if (tok
== TOK_DEFINED
) {
1527 if (tok
< TOK_IDENT
)
1528 expect("identifier");
1532 if (define_find(tok
)
1533 || tok
== TOK___HAS_INCLUDE
1534 || tok
== TOK___HAS_INCLUDE_NEXT
)
1543 } else if (tok
== TOK___HAS_INCLUDE
||
1544 tok
== TOK___HAS_INCLUDE_NEXT
) {
1549 c
= parse_include(s1
, t
- TOK___HAS_INCLUDE
, 1);
1554 } else if (tok
>= TOK_IDENT
) {
1555 /* if undefined macro, replace with zero, check for func-like */
1559 tok_str_add_tok(str
);
1562 tcc_error("function-like macro '%s' is not defined",
1563 get_tok_str(t
, NULL
));
1566 tok_str_add_tok(str
);
1569 tok_str_add(str
, -1); /* simulate end of file */
1570 tok_str_add(str
, 0);
1571 /* now evaluate C constant expression */
1572 begin_macro(str
, 1);
1580 /* parse after #define */
1581 ST_FUNC
void parse_define(void)
1583 Sym
*s
, *first
, **ps
;
1584 int v
, t
, varg
, is_vaargs
, spc
;
1585 int saved_parse_flags
= parse_flags
;
1588 if (v
< TOK_IDENT
|| v
== TOK_DEFINED
)
1589 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1590 /* XXX: should check if same macro (ANSI) */
1593 /* We have to parse the whole define as if not in asm mode, in particular
1594 no line comment with '#' must be ignored. Also for function
1595 macros the argument list must be parsed without '.' being an ID
1597 parse_flags
= ((parse_flags
& ~PARSE_FLAG_ASM_FILE
) | PARSE_FLAG_SPACES
);
1598 /* '(' must be just after macro definition for MACRO_FUNC */
1600 parse_flags
&= ~PARSE_FLAG_SPACES
;
1602 int dotid
= set_idnum('.', 0);
1605 if (tok
!= ')') for (;;) {
1609 if (varg
== TOK_DOTS
) {
1610 varg
= TOK___VA_ARGS__
;
1612 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1616 if (varg
< TOK_IDENT
)
1618 tcc_error("bad macro parameter list");
1619 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1624 if (tok
!= ',' || is_vaargs
)
1628 parse_flags
|= PARSE_FLAG_SPACES
;
1631 set_idnum('.', dotid
);
1636 parse_flags
|= PARSE_FLAG_ACCEPT_STRAYS
| PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
;
1637 /* The body of a macro definition should be parsed such that identifiers
1638 are parsed like the file mode determines (i.e. with '.' being an
1639 ID character in asm mode). But '#' should be retained instead of
1640 regarded as line comment leader, so still don't set ASM_FILE
1642 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1643 /* remove spaces around ## and after '#' */
1644 if (TOK_TWOSHARPS
== tok
) {
1651 } else if ('#' == tok
) {
1653 } else if (check_space(tok
, &spc
)) {
1656 tok_str_add2(&tokstr_buf
, tok
, &tokc
);
1661 parse_flags
= saved_parse_flags
;
1663 --tokstr_buf
.len
; /* remove trailing space */
1664 tok_str_add(&tokstr_buf
, 0);
1667 tcc_error("'##' cannot appear at either end of macro");
1668 define_push(v
, t
, tok_str_dup(&tokstr_buf
), first
);
1671 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
, int add
)
1673 const unsigned char *s
;
1679 s
= (unsigned char *) filename
;
1682 h
= TOK_HASH_FUNC(h
, toup(*s
));
1684 h
= TOK_HASH_FUNC(h
, *s
);
1688 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1690 i
= s1
->cached_includes_hash
[h
];
1694 e
= s1
->cached_includes
[i
- 1];
1695 if (0 == PATHCMP(e
->filename
, filename
))
1702 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1703 strcpy(e
->filename
, filename
);
1704 e
->ifndef_macro
= e
->once
= 0;
1705 dynarray_add(&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1706 /* add in hash table */
1707 e
->hash_next
= s1
->cached_includes_hash
[h
];
1708 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1710 printf("adding cached '%s'\n", filename
);
1715 static void pragma_parse(TCCState
*s1
)
1718 if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1722 if (next(), tok
!= '(')
1724 if (next(), tok
!= TOK_STR
)
1726 v
= tok_alloc(tokc
.str
.data
, tokc
.str
.size
- 1)->tok
;
1727 if (next(), tok
!= ')')
1729 if (t
== TOK_push_macro
) {
1730 while (NULL
== (s
= define_find(v
)))
1731 define_push(v
, 0, NULL
, NULL
);
1732 s
->type
.ref
= s
; /* set push boundary */
1734 for (s
= define_stack
; s
; s
= s
->prev
)
1735 if (s
->v
== v
&& s
->type
.ref
== s
) {
1741 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1743 tcc_warning("unbalanced #pragma pop_macro");
1744 pp_debug_tok
= t
, pp_debug_symv
= v
;
1746 } else if (tok
== TOK_once
) {
1747 search_cached_include(s1
, file
->filename
, 1)->once
= pp_once
;
1749 } else if (s1
->output_type
== TCC_OUTPUT_PREPROCESS
) {
1750 /* tcc -E: keep pragmas below unchanged */
1752 unget_tok(TOK_PRAGMA
);
1754 unget_tok(TOK_LINEFEED
);
1756 } else if (tok
== TOK_pack
) {
1758 #pragma pack(1) // set
1759 #pragma pack() // reset to default
1760 #pragma pack(push) // push current
1761 #pragma pack(push,1) // push & set
1762 #pragma pack(pop) // restore previous */
1765 if (tok
== TOK_ASM_pop
) {
1767 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1769 tcc_error("out of pack stack");
1771 s1
->pack_stack_ptr
--;
1775 if (tok
== TOK_ASM_push
) {
1777 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1779 val
= *s1
->pack_stack_ptr
++;
1784 if (tok
!= TOK_CINT
)
1787 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1792 *s1
->pack_stack_ptr
= val
;
1797 } else if (tok
== TOK_comment
) {
1806 p
= tcc_strdup((char *)tokc
.str
.data
);
1811 dynarray_add(&s1
->pragma_libs
, &s1
->nb_pragma_libs
, p
);
1813 if (t
== TOK_option
)
1814 tcc_set_options(s1
, p
);
1819 tcc_warning_c(warn_unsupported
)("#pragma %s ignored", get_tok_str(tok
, &tokc
));
1823 tcc_error("malformed #pragma directive");
1827 /* is_bof is true if first non space token at beginning of file */
1828 ST_FUNC
void preprocess(int is_bof
)
1830 TCCState
*s1
= tcc_state
;
1831 int c
, n
, saved_parse_flags
;
1835 saved_parse_flags
= parse_flags
;
1836 parse_flags
= PARSE_FLAG_PREPROCESS
1837 | PARSE_FLAG_TOK_NUM
1838 | PARSE_FLAG_TOK_STR
1839 | PARSE_FLAG_LINEFEED
1840 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
1849 pp_debug_symv
= tok
;
1855 pp_debug_symv
= tok
;
1856 s
= define_find(tok
);
1857 /* undefine symbol by putting an invalid name */
1862 case TOK_INCLUDE_NEXT
:
1863 parse_include(s1
, tok
- TOK_INCLUDE
, 0);
1869 c
= expr_preprocess(s1
);
1875 if (tok
< TOK_IDENT
)
1876 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1880 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1882 file
->ifndef_macro
= tok
;
1885 if (define_find(tok
)
1886 || tok
== TOK___HAS_INCLUDE
1887 || tok
== TOK___HAS_INCLUDE_NEXT
)
1890 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1891 tcc_error("memory full (ifdef)");
1892 *s1
->ifdef_stack_ptr
++ = c
;
1895 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1896 tcc_error("#else without matching #if");
1897 if (s1
->ifdef_stack_ptr
[-1] & 2)
1898 tcc_error("#else after #else");
1899 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1902 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1903 tcc_error("#elif without matching #if");
1904 c
= s1
->ifdef_stack_ptr
[-1];
1906 tcc_error("#elif after #else");
1907 /* last #if/#elif expression was true: we skip */
1911 c
= expr_preprocess(s1
);
1912 s1
->ifdef_stack_ptr
[-1] = c
;
1915 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1916 file
->ifndef_macro
= 0;
1925 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1926 tcc_error("#endif without matching #if");
1927 s1
->ifdef_stack_ptr
--;
1928 /* '#ifndef macro' was at the start of file. Now we check if
1929 an '#endif' is exactly at the end of file */
1930 if (file
->ifndef_macro
&&
1931 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1932 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1933 /* need to set to zero to avoid false matches if another
1934 #ifndef at middle of file */
1935 file
->ifndef_macro
= 0;
1936 while (tok
!= TOK_LINEFEED
)
1938 tok_flags
|= TOK_FLAG_ENDIF
;
1943 n
= strtoul((char*)tokc
.str
.data
, &q
, 10);
1947 if (tok
!= TOK_CINT
)
1949 tcc_error("wrong #line format");
1953 if (tok
!= TOK_LINEFEED
) {
1954 if (tok
== TOK_STR
) {
1955 if (file
->true_filename
== file
->filename
)
1956 file
->true_filename
= tcc_strdup(file
->filename
);
1957 q
= (char *)tokc
.str
.data
;
1959 if (!IS_ABSPATH(q
)) {
1960 /* prepend directory from real file */
1961 pstrcpy(buf
, sizeof buf
, file
->true_filename
);
1962 *tcc_basename(buf
) = 0;
1964 pstrcat(buf
, sizeof buf
, q
);
1965 tcc_debug_putfile(s1
, buf
);
1966 } else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
1973 total_lines
+= file
->line_num
- n
;
1980 while (c
!= '\n' && c
!= CH_EOF
) {
1981 if ((q
- buf
) < sizeof(buf
) - 1)
1986 if (tok
== TOK_ERROR
)
1987 tcc_error("#error %s", buf
);
1989 tcc_warning("#warning %s", buf
);
1997 /* ignore gas line comment in an 'S' file. */
1998 if (saved_parse_flags
& PARSE_FLAG_ASM_FILE
)
2000 if (tok
== '!' && is_bof
)
2001 /* '!' is ignored at beginning to allow C scripts. */
2003 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
2005 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
- 1);
2008 /* ignore other preprocess commands or #! for C scripts */
2009 while (tok
!= TOK_LINEFEED
)
2012 parse_flags
= saved_parse_flags
;
2015 /* evaluate escape codes in a string. */
2016 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
2031 case '0': case '1': case '2': case '3':
2032 case '4': case '5': case '6': case '7':
2033 /* at most three octal digits */
2038 n
= n
* 8 + c
- '0';
2042 n
= n
* 8 + c
- '0';
2047 goto add_char_nonext
;
2048 case 'x': i
= 0; goto parse_hex_or_ucn
;
2049 case 'u': i
= 4; goto parse_hex_or_ucn
;
2050 case 'U': i
= 8; goto parse_hex_or_ucn
;
2056 if (c
>= 'a' && c
<= 'f')
2058 else if (c
>= 'A' && c
<= 'F')
2063 expect("more hex digits in universal-character-name");
2065 goto add_hex_or_ucn
;
2072 goto add_char_nonext
;
2074 cstr_u8cat(outstr
, n
);
2099 goto invalid_escape
;
2109 if (c
>= '!' && c
<= '~')
2110 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
2112 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
2115 } else if (is_long
&& c
>= 0x80) {
2116 /* assume we are processing UTF-8 sequence */
2117 /* reference: The Unicode Standard, Version 10.0, ch3.9 */
2119 int cont
; /* count of continuation bytes */
2120 int skip
; /* how many bytes should skip when error occurred */
2123 /* decode leading byte */
2125 skip
= 1; goto invalid_utf8_sequence
;
2126 } else if (c
<= 0xDF) {
2127 cont
= 1; n
= c
& 0x1f;
2128 } else if (c
<= 0xEF) {
2129 cont
= 2; n
= c
& 0xf;
2130 } else if (c
<= 0xF4) {
2131 cont
= 3; n
= c
& 0x7;
2133 skip
= 1; goto invalid_utf8_sequence
;
2136 /* decode continuation bytes */
2137 for (i
= 1; i
<= cont
; i
++) {
2138 int l
= 0x80, h
= 0xBF;
2140 /* adjust limit for second byte */
2143 case 0xE0: l
= 0xA0; break;
2144 case 0xED: h
= 0x9F; break;
2145 case 0xF0: l
= 0x90; break;
2146 case 0xF4: h
= 0x8F; break;
2150 if (p
[i
] < l
|| p
[i
] > h
) {
2151 skip
= i
; goto invalid_utf8_sequence
;
2154 n
= (n
<< 6) | (p
[i
] & 0x3f);
2157 /* advance pointer */
2160 goto add_char_nonext
;
2162 /* error handling */
2163 invalid_utf8_sequence
:
2164 tcc_warning("ill-formed UTF-8 subsequence starting with: \'\\x%x\'", c
);
2167 goto add_char_nonext
;
2173 cstr_ccat(outstr
, c
);
2175 #ifdef TCC_TARGET_PE
2176 /* store as UTF-16 */
2178 cstr_wccat(outstr
, c
);
2181 cstr_wccat(outstr
, (c
>> 10) + 0xD800);
2182 cstr_wccat(outstr
, (c
& 0x3FF) + 0xDC00);
2185 cstr_wccat(outstr
, c
);
2189 /* add a trailing '\0' */
2191 cstr_ccat(outstr
, '\0');
2193 cstr_wccat(outstr
, '\0');
2196 static void parse_string(const char *s
, int len
)
2198 uint8_t buf
[1000], *p
= buf
;
2201 if ((is_long
= *s
== 'L'))
2205 if (len
>= sizeof buf
)
2206 p
= tcc_malloc(len
+ 1);
2210 cstr_reset(&tokcstr
);
2211 parse_escape_string(&tokcstr
, p
, is_long
);
2216 int char_size
, i
, n
, c
;
2217 /* XXX: make it portable */
2219 tok
= TOK_CCHAR
, char_size
= 1;
2221 tok
= TOK_LCHAR
, char_size
= sizeof(nwchar_t
);
2222 n
= tokcstr
.size
/ char_size
- 1;
2224 tcc_error("empty character constant");
2226 tcc_warning_c(warn_all
)("multi-character character constant");
2227 for (c
= i
= 0; i
< n
; ++i
) {
2229 c
= ((nwchar_t
*)tokcstr
.data
)[i
];
2231 c
= (c
<< 8) | ((char *)tokcstr
.data
)[i
];
2235 tokc
.str
.size
= tokcstr
.size
;
2236 tokc
.str
.data
= tokcstr
.data
;
2244 /* we use 64 bit numbers */
2247 /* bn = (bn << shift) | or_val */
2248 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
2252 for(i
=0;i
<BN_SIZE
;i
++) {
2254 bn
[i
] = (v
<< shift
) | or_val
;
2255 or_val
= v
>> (32 - shift
);
2259 static void bn_zero(unsigned int *bn
)
2262 for(i
=0;i
<BN_SIZE
;i
++) {
2267 /* parse number in null terminated string 'p' and return it in the
2269 static void parse_number(const char *p
)
2271 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
2273 unsigned int bn
[BN_SIZE
];
2284 goto float_frac_parse
;
2285 } else if (t
== '0') {
2286 if (ch
== 'x' || ch
== 'X') {
2290 } else if (tcc_state
->tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
2296 /* parse all digits. cannot check octal numbers at this stage
2297 because of floating point constants */
2299 if (ch
>= 'a' && ch
<= 'f')
2301 else if (ch
>= 'A' && ch
<= 'F')
2309 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
2311 tcc_error("number too long");
2317 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
2318 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
2320 /* NOTE: strtox should support that for hexa numbers, but
2321 non ISOC99 libcs do not support it, so we prefer to do
2323 /* hexadecimal or binary floats */
2324 /* XXX: handle overflows */
2336 } else if (t
>= 'a') {
2338 } else if (t
>= 'A') {
2343 bn_lshift(bn
, shift
, t
);
2350 if (t
>= 'a' && t
<= 'f') {
2352 } else if (t
>= 'A' && t
<= 'F') {
2354 } else if (t
>= '0' && t
<= '9') {
2360 tcc_error("invalid digit");
2361 bn_lshift(bn
, shift
, t
);
2366 if (ch
!= 'p' && ch
!= 'P')
2373 } else if (ch
== '-') {
2377 if (ch
< '0' || ch
> '9')
2378 expect("exponent digits");
2379 while (ch
>= '0' && ch
<= '9') {
2380 exp_val
= exp_val
* 10 + ch
- '0';
2383 exp_val
= exp_val
* s
;
2385 /* now we can generate the number */
2386 /* XXX: should patch directly float number */
2387 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2388 d
= ldexp(d
, exp_val
- frac_bits
);
2393 /* float : should handle overflow */
2395 } else if (t
== 'L') {
2397 #ifdef TCC_USING_DOUBLE_FOR_LDOUBLE
2402 /* XXX: not large enough */
2403 tokc
.ld
= (long double)d
;
2410 /* decimal floats */
2412 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2417 while (ch
>= '0' && ch
<= '9') {
2418 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2424 if (ch
== 'e' || ch
== 'E') {
2425 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2429 if (ch
== '-' || ch
== '+') {
2430 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2435 if (ch
< '0' || ch
> '9')
2436 expect("exponent digits");
2437 while (ch
>= '0' && ch
<= '9') {
2438 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2450 tokc
.f
= strtof(token_buf
, NULL
);
2451 } else if (t
== 'L') {
2453 #ifdef TCC_USING_DOUBLE_FOR_LDOUBLE
2455 tokc
.d
= strtod(token_buf
, NULL
);
2458 tokc
.ld
= strtold(token_buf
, NULL
);
2462 tokc
.d
= strtod(token_buf
, NULL
);
2466 unsigned long long n
, n1
;
2467 int lcount
, ucount
, ov
= 0;
2470 /* integer number */
2473 if (b
== 10 && *q
== '0') {
2480 /* no need for checks except for base 10 / 8 errors */
2490 tcc_error("invalid digit");
2493 /* detect overflow */
2494 if (n1
>= 0x1000000000000000ULL
&& n
/ b
!= n1
)
2498 /* Determine the characteristics (unsigned and/or 64bit) the type of
2499 the constant must have according to the constant suffix(es) */
2500 lcount
= ucount
= 0;
2506 tcc_error("three 'l's in integer constant");
2507 if (lcount
&& *(p
- 1) != ch
)
2508 tcc_error("incorrect integer suffix: %s", p1
);
2511 } else if (t
== 'U') {
2513 tcc_error("two 'u's in integer constant");
2521 /* Determine if it needs 64 bits and/or unsigned in order to fit */
2522 if (ucount
== 0 && b
== 10) {
2523 if (lcount
<= (LONG_SIZE
== 4)) {
2524 if (n
>= 0x80000000U
)
2525 lcount
= (LONG_SIZE
== 4) + 1;
2527 if (n
>= 0x8000000000000000ULL
)
2530 if (lcount
<= (LONG_SIZE
== 4)) {
2531 if (n
>= 0x100000000ULL
)
2532 lcount
= (LONG_SIZE
== 4) + 1;
2533 else if (n
>= 0x80000000U
)
2536 if (n
>= 0x8000000000000000ULL
)
2541 tcc_warning("integer constant overflow");
2550 ++tok
; /* TOK_CU... */
2554 tcc_error("invalid number");
2558 #define PARSE2(c1, tok1, c2, tok2) \
2569 /* return next token without macro substitution */
2570 static inline void next_nomacro1(void)
2572 int t
, c
, is_long
, len
;
2586 if (parse_flags
& PARSE_FLAG_SPACES
)
2587 goto keep_tok_flags
;
2588 while (isidnum_table
[*p
- CH_EOF
] & IS_SPC
)
2597 /* first look if it is in fact an end of buffer */
2598 c
= handle_stray(&p
);
2602 TCCState
*s1
= tcc_state
;
2603 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2604 && !(tok_flags
& TOK_FLAG_EOF
)) {
2605 tok_flags
|= TOK_FLAG_EOF
;
2607 goto keep_tok_flags
;
2608 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2610 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2611 tcc_error("missing #endif");
2612 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2613 /* no include left : end of file. */
2616 tok_flags
&= ~TOK_FLAG_EOF
;
2617 /* pop include file */
2619 /* test if previous '#endif' was after a #ifdef at
2621 if (tok_flags
& TOK_FLAG_ENDIF
) {
2623 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2625 search_cached_include(s1
, file
->filename
, 1)
2626 ->ifndef_macro
= file
->ifndef_macro_saved
;
2627 tok_flags
&= ~TOK_FLAG_ENDIF
;
2630 /* add end of include file debug info */
2631 tcc_debug_eincl(tcc_state
);
2632 /* pop include stack */
2634 s1
->include_stack_ptr
--;
2636 if (p
== file
->buffer
)
2637 tok_flags
= TOK_FLAG_BOF
;
2638 tok_flags
|= TOK_FLAG_BOL
;
2648 tok_flags
|= TOK_FLAG_BOL
;
2651 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2654 goto keep_tok_flags
;
2659 if ((tok_flags
& TOK_FLAG_BOL
) &&
2660 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2662 preprocess(tok_flags
& TOK_FLAG_BOF
);
2668 tok
= TOK_TWOSHARPS
;
2670 #if !defined(TCC_TARGET_ARM)
2671 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2672 p
= parse_line_comment(p
- 1);
2683 /* dollar is allowed to start identifiers when not parsing asm */
2685 if (!(isidnum_table
[c
- CH_EOF
] & IS_ID
)
2686 || (parse_flags
& PARSE_FLAG_ASM_FILE
))
2689 case 'a': case 'b': case 'c': case 'd':
2690 case 'e': case 'f': case 'g': case 'h':
2691 case 'i': case 'j': case 'k': case 'l':
2692 case 'm': case 'n': case 'o': case 'p':
2693 case 'q': case 'r': case 's': case 't':
2694 case 'u': case 'v': case 'w': case 'x':
2696 case 'A': case 'B': case 'C': case 'D':
2697 case 'E': case 'F': case 'G': case 'H':
2698 case 'I': case 'J': case 'K':
2699 case 'M': case 'N': case 'O': case 'P':
2700 case 'Q': case 'R': case 'S': case 'T':
2701 case 'U': case 'V': case 'W': case 'X':
2707 h
= TOK_HASH_FUNC(h
, c
);
2708 while (c
= *++p
, isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2709 h
= TOK_HASH_FUNC(h
, c
);
2714 /* fast case : no stray found, so we have the full token
2715 and we have already hashed it */
2716 h
&= (TOK_HASH_SIZE
- 1);
2717 pts
= &hash_ident
[h
];
2722 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2724 pts
= &(ts
->hash_next
);
2726 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2730 cstr_reset(&tokcstr
);
2731 cstr_cat(&tokcstr
, (char *) p1
, len
);
2735 while (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2737 cstr_ccat(&tokcstr
, c
);
2740 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2746 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2748 goto parse_ident_fast
;
2751 if (c
== '\'' || c
== '\"') {
2755 cstr_reset(&tokcstr
);
2756 cstr_ccat(&tokcstr
, 'L');
2757 goto parse_ident_slow
;
2762 case '0': case '1': case '2': case '3':
2763 case '4': case '5': case '6': case '7':
2767 /* after the first digit, accept digits, alpha, '.' or sign if
2768 prefixed by 'eEpP' */
2770 cstr_reset(&tokcstr
);
2772 cstr_ccat(&tokcstr
, t
);
2773 if (!((isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2775 || ((c
== '+' || c
== '-')
2776 && (((t
== 'e' || t
== 'E')
2777 && !(parse_flags
& PARSE_FLAG_ASM_FILE
2778 /* 0xe+1 is 3 tokens in asm */
2779 && ((char*)tokcstr
.data
)[0] == '0'
2780 && toup(((char*)tokcstr
.data
)[1]) == 'X'))
2781 || t
== 'p' || t
== 'P'))))
2786 /* We add a trailing '\0' to ease parsing */
2787 cstr_ccat(&tokcstr
, '\0');
2788 tokc
.str
.size
= tokcstr
.size
;
2789 tokc
.str
.data
= tokcstr
.data
;
2794 /* special dot handling because it can also start a number */
2799 } else if ((isidnum_table
['.' - CH_EOF
] & IS_ID
)
2800 && (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))) {
2802 goto parse_ident_fast
;
2803 } else if (c
== '.') {
2809 *--p
= '.'; /* may underflow into file->unget[] */
2820 cstr_reset(&tokcstr
);
2822 cstr_ccat(&tokcstr
, 'L');
2823 cstr_ccat(&tokcstr
, c
);
2824 p
= parse_pp_string(p
, c
, &tokcstr
);
2825 cstr_ccat(&tokcstr
, c
);
2826 cstr_ccat(&tokcstr
, '\0');
2827 tokc
.str
.size
= tokcstr
.size
;
2828 tokc
.str
.data
= tokcstr
.data
;
2837 } else if (c
== '<') {
2854 } else if (c
== '>') {
2872 } else if (c
== '=') {
2885 } else if (c
== '=') {
2898 } else if (c
== '=') {
2911 } else if (c
== '=') {
2914 } else if (c
== '>') {
2922 PARSE2('!', '!', '=', TOK_NE
)
2923 PARSE2('=', '=', '=', TOK_EQ
)
2924 PARSE2('*', '*', '=', TOK_A_MUL
)
2925 PARSE2('%', '%', '=', TOK_A_MOD
)
2926 PARSE2('^', '^', '=', TOK_A_XOR
)
2928 /* comments or operator */
2932 p
= parse_comment(p
);
2933 /* comments replaced by a blank */
2936 } else if (c
== '/') {
2937 p
= parse_line_comment(p
);
2940 } else if (c
== '=') {
2960 case '@': /* only used in assembler */
2966 if (c
>= 0x80 && c
<= 0xFF) /* utf8 identifiers */
2967 goto parse_ident_fast
;
2968 if (parse_flags
& PARSE_FLAG_ASM_FILE
)
2970 tcc_error("unrecognized character \\x%02x", c
);
2976 #if defined(PARSE_DEBUG)
2977 printf("token = %d %s\n", tok
, get_tok_str(tok
, &tokc
));
2981 static void macro_subst(
2982 TokenString
*tok_str
,
2984 const int *macro_str
2987 /* substitute arguments in replacement lists in macro_str by the values in
2988 args (field d) and return allocated string */
2989 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
3001 TOK_GET(&t
, ¯o_str
, &cval
);
3006 TOK_GET(&t
, ¯o_str
, &cval
);
3009 s
= sym_find2(args
, t
);
3012 cstr_ccat(&cstr
, '\"');
3016 TOK_GET(&t
, &st
, &cval
);
3017 if (t
!= TOK_PLCHLDR
3019 && 0 == check_space(t
, &spc
)) {
3020 const char *s
= get_tok_str(t
, &cval
);
3022 if (t
== TOK_PPSTR
&& *s
!= '\'')
3023 add_char(&cstr
, *s
);
3025 cstr_ccat(&cstr
, *s
);
3031 cstr_ccat(&cstr
, '\"');
3032 cstr_ccat(&cstr
, '\0');
3034 printf("\nstringize: <%s>\n", (char *)cstr
.data
);
3037 cval
.str
.size
= cstr
.size
;
3038 cval
.str
.data
= cstr
.data
;
3039 tok_str_add2(&str
, TOK_PPSTR
, &cval
);
3043 expect("macro parameter after '#'");
3045 } else if (t
>= TOK_IDENT
) {
3046 s
= sym_find2(args
, t
);
3050 /* if '##' is present before or after, no arg substitution */
3051 if (*macro_str
== TOK_PPJOIN
|| t1
== TOK_PPJOIN
) {
3052 /* special case for var arg macros : ## eats the ','
3053 if empty VA_ARGS variable. */
3054 if (t1
== TOK_PPJOIN
&& t0
== ',' && gnu_ext
&& s
->type
.t
) {
3056 /* suppress ',' '##' */
3059 /* suppress '##' and add variable */
3067 /* Expand arguments tokens and store them. In most
3068 cases we could also re-expand each argument if
3069 used multiple times, but not if the argument
3070 contains the __COUNTER__ macro. */
3072 sym_push2(&s
->next
, s
->v
, s
->type
.t
, 0);
3074 macro_subst(&str2
, nested_list
, st
);
3075 tok_str_add(&str2
, 0);
3076 s
->next
->d
= str2
.str
;
3082 TOK_GET(&t2
, &st
, &cval
);
3085 tok_str_add2(&str
, t2
, &cval
);
3087 if (str
.len
== l0
) /* expanded to empty string */
3088 tok_str_add(&str
, TOK_PLCHLDR
);
3090 tok_str_add(&str
, t
);
3093 tok_str_add2(&str
, t
, &cval
);
3097 tok_str_add(&str
, 0);
3101 static char const ab_month_name
[12][4] =
3103 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
3104 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
3107 static int paste_tokens(int t1
, CValue
*v1
, int t2
, CValue
*v2
)
3113 if (t1
!= TOK_PLCHLDR
)
3114 cstr_cat(&cstr
, get_tok_str(t1
, v1
), -1);
3116 if (t2
!= TOK_PLCHLDR
)
3117 cstr_cat(&cstr
, get_tok_str(t2
, v2
), -1);
3118 cstr_ccat(&cstr
, '\0');
3120 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3121 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3125 if (0 == *file
->buf_ptr
)
3129 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid"
3130 " preprocessing token", n
, (char *)cstr
.data
, (char*)cstr
.data
+ n
);
3135 //printf("paste <%s>\n", (char*)cstr.data);
3140 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
3141 return the resulting string (which must be freed). */
3142 static inline int *macro_twosharps(const int *ptr0
)
3146 TokenString macro_str1
;
3147 int start_of_nosubsts
= -1;
3150 /* we search the first '##' */
3151 for (ptr
= ptr0
;;) {
3152 TOK_GET(&t
, &ptr
, &cval
);
3153 if (t
== TOK_PPJOIN
)
3159 tok_str_new(¯o_str1
);
3161 //tok_print(" $$$", ptr0);
3162 for (ptr
= ptr0
;;) {
3163 TOK_GET(&t
, &ptr
, &cval
);
3166 if (t
== TOK_PPJOIN
)
3168 while (*ptr
== TOK_PPJOIN
) {
3170 /* given 'a##b', remove nosubsts preceding 'a' */
3171 if (start_of_nosubsts
>= 0)
3172 macro_str1
.len
= start_of_nosubsts
;
3173 /* given 'a##b', remove nosubsts preceding 'b' */
3174 while ((t1
= *++ptr
) == TOK_NOSUBST
)
3176 if (t1
&& t1
!= TOK_PPJOIN
) {
3177 TOK_GET(&t1
, &ptr
, &cv1
);
3178 if (t
!= TOK_PLCHLDR
|| t1
!= TOK_PLCHLDR
) {
3179 if (paste_tokens(t
, &cval
, t1
, &cv1
)) {
3180 t
= tok
, cval
= tokc
;
3182 tok_str_add2(¯o_str1
, t
, &cval
);
3188 if (t
== TOK_NOSUBST
) {
3189 if (start_of_nosubsts
< 0)
3190 start_of_nosubsts
= macro_str1
.len
;
3192 start_of_nosubsts
= -1;
3194 tok_str_add2(¯o_str1
, t
, &cval
);
3196 tok_str_add(¯o_str1
, 0);
3197 //tok_print(" ###", macro_str1.str);
3198 return macro_str1
.str
;
3201 /* peek or read [ws_str == NULL] next token from function macro call,
3202 walking up macro levels up to the file if necessary */
3203 static int next_argstream(Sym
**nested_list
, TokenString
*ws_str
)
3211 p
= macro_ptr
, t
= *p
;
3213 while (is_space(t
) || TOK_LINEFEED
== t
|| TOK_PLCHLDR
== t
)
3214 tok_str_add(ws_str
, t
), t
= *++p
;
3218 /* also, end of scope for nested defined symbol */
3220 while (sa
&& sa
->v
== 0)
3227 uint8_t *p
= file
->buf_ptr
;
3228 int ch
= handle_bs(&p
);
3230 while (is_space(ch
) || ch
== '\n' || ch
== '/') {
3235 p
= parse_comment(p
) - 1;
3236 } else if (c
== '/') {
3237 p
= parse_line_comment(p
) - 1;
3246 if (!(ch
== '\f' || ch
== '\v' || ch
== '\r'))
3247 tok_str_add(ws_str
, ch
);
3262 /* do macro substitution of current token with macro 's' and add
3263 result to (tok_str,tok_len). 'nested_list' is the list of all
3264 macros we got inside to avoid recursing. Return non zero if no
3265 substitution needs to be done */
3266 static int macro_subst_tok(
3267 TokenString
*tok_str
,
3271 Sym
*args
, *sa
, *sa1
;
3272 int parlevel
, t
, t1
, spc
;
3279 /* if symbol is a macro, prepare substitution */
3280 /* special macros */
3281 if (tok
== TOK___LINE__
|| tok
== TOK___COUNTER__
) {
3282 t
= tok
== TOK___LINE__
? file
->line_num
: pp_counter
++;
3283 snprintf(buf
, sizeof(buf
), "%d", t
);
3287 } else if (tok
== TOK___FILE__
) {
3288 cstrval
= file
->filename
;
3290 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
3295 tm
= localtime(&ti
);
3296 if (tok
== TOK___DATE__
) {
3297 snprintf(buf
, sizeof(buf
), "%s %2d %d",
3298 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
3300 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
3301 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
3308 cstr_cat(&cstr
, cstrval
, 0);
3309 cval
.str
.size
= cstr
.size
;
3310 cval
.str
.data
= cstr
.data
;
3311 tok_str_add2(tok_str
, t1
, &cval
);
3314 int saved_parse_flags
= parse_flags
;
3315 int *joined_str
= NULL
;
3318 if (s
->type
.t
== MACRO_FUNC
) {
3319 /* whitespace between macro name and argument list */
3321 tok_str_new(&ws_str
);
3324 parse_flags
|= PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
3325 | PARSE_FLAG_ACCEPT_STRAYS
;
3327 /* get next token from argument stream */
3328 t
= next_argstream(nested_list
, &ws_str
);
3330 /* not a macro substitution after all, restore the
3331 * macro token plus all whitespace we've read.
3332 * whitespace is intentionally not merged to preserve
3334 parse_flags
= saved_parse_flags
;
3335 tok_str_add(tok_str
, tok
);
3336 if (parse_flags
& PARSE_FLAG_SPACES
) {
3338 for (i
= 0; i
< ws_str
.len
; i
++)
3339 tok_str_add(tok_str
, ws_str
.str
[i
]);
3341 if (ws_str
.len
&& ws_str
.str
[ws_str
.len
- 1] == '\n')
3342 tok_flags
|= TOK_FLAG_BOL
;
3343 tok_str_free_str(ws_str
.str
);
3346 tok_str_free_str(ws_str
.str
);
3349 next_nomacro(); /* eat '(' */
3350 } while (tok
== TOK_PLCHLDR
|| is_space(tok
));
3352 /* argument macro */
3355 /* NOTE: empty args are allowed, except if no args */
3358 next_argstream(nested_list
, NULL
);
3359 } while (tok
== TOK_PLCHLDR
|| is_space(tok
) ||
3360 TOK_LINEFEED
== tok
);
3362 /* handle '()' case */
3363 if (!args
&& !sa
&& tok
== ')')
3366 tcc_error("macro '%s' used with too many args",
3367 get_tok_str(s
->v
, 0));
3370 /* NOTE: non zero sa->t indicates VA_ARGS */
3371 while ((parlevel
> 0 ||
3373 (tok
!= ',' || sa
->type
.t
)))) {
3374 if (tok
== TOK_EOF
|| tok
== 0)
3378 else if (tok
== ')')
3380 if (tok
== TOK_LINEFEED
)
3382 if (!check_space(tok
, &spc
))
3383 tok_str_add2(&str
, tok
, &tokc
);
3384 next_argstream(nested_list
, NULL
);
3389 tok_str_add(&str
, -1);
3390 tok_str_add(&str
, 0);
3391 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
3395 /* special case for gcc var args: add an empty
3396 var arg argument if it is omitted */
3397 if (sa
&& sa
->type
.t
&& gnu_ext
)
3405 tcc_error("macro '%s' used with too few args",
3406 get_tok_str(s
->v
, 0));
3409 /* now subst each arg */
3410 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
3415 tok_str_free_str(sa
->d
);
3417 tok_str_free_str(sa
->next
->d
);
3423 parse_flags
= saved_parse_flags
;
3426 sym_push2(nested_list
, s
->v
, 0, 0);
3427 parse_flags
= saved_parse_flags
;
3428 joined_str
= macro_twosharps(mstr
);
3429 macro_subst(tok_str
, nested_list
, joined_str
? joined_str
: mstr
);
3431 /* pop nested defined symbol */
3433 *nested_list
= sa1
->prev
;
3436 tok_str_free_str(joined_str
);
3438 tok_str_free_str(mstr
);
3443 /* do macro substitution of macro_str and add result to
3444 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3445 inside to avoid recursing. */
3446 static void macro_subst(
3447 TokenString
*tok_str
,
3449 const int *macro_str
3453 int t
, spc
, nosubst
;
3459 TOK_GET(&t
, ¯o_str
, &cval
);
3463 if (t
>= TOK_IDENT
&& 0 == nosubst
) {
3468 /* if nested substitution, do nothing */
3469 if (sym_find2(*nested_list
, t
)) {
3470 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3471 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3476 TokenString
*str
= tok_str_alloc();
3477 str
->str
= (int*)macro_str
;
3478 begin_macro(str
, 2);
3481 macro_subst_tok(tok_str
, nested_list
, s
);
3483 if (macro_stack
!= str
) {
3484 /* already finished by reading function macro arguments */
3488 macro_str
= macro_ptr
;
3492 spc
= is_space(t
= tok_str
->str
[tok_str
->lastlen
]);
3495 if (!check_space(t
, &spc
))
3496 tok_str_add2(tok_str
, t
, &cval
);
3499 if (nosubst
> 1 && (spc
|| (++nosubst
== 3 && t
== '(')))
3503 if (t
== TOK_NOSUBST
)
3506 /* GCC supports 'defined' as result of a macro substitution */
3507 if (t
== TOK_DEFINED
&& pp_expr
)
3512 /* return next token without macro substitution. Can read input from
3514 static void next_nomacro(void)
3520 if (TOK_HAS_VALUE(t
)) {
3521 tok_get(&tok
, ¯o_ptr
, &tokc
);
3522 if (t
== TOK_LINENUM
) {
3523 file
->line_num
= tokc
.i
;
3528 if (t
< TOK_IDENT
) {
3529 if (!(parse_flags
& PARSE_FLAG_SPACES
)
3530 && (isidnum_table
[t
- CH_EOF
] & IS_SPC
))
3540 /* return next token with macro substitution */
3541 ST_FUNC
void next(void)
3548 if (!TOK_HAS_VALUE(t
)) {
3549 if (t
== TOK_NOSUBST
|| t
== TOK_PLCHLDR
) {
3550 /* discard preprocessor markers */
3552 } else if (t
== 0) {
3553 /* end of macro or unget token string */
3556 } else if (t
== '\\') {
3557 if (!(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
3558 tcc_error("stray '\\' in program");
3562 } else if (t
>= TOK_IDENT
&& (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3563 /* if reading from file, try to substitute macros */
3564 Sym
*s
= define_find(t
);
3566 Sym
*nested_list
= NULL
;
3568 macro_subst_tok(&tokstr_buf
, &nested_list
, s
);
3569 tok_str_add(&tokstr_buf
, 0);
3570 begin_macro(&tokstr_buf
, 0);
3575 /* convert preprocessor tokens into C tokens */
3576 if (t
== TOK_PPNUM
) {
3577 if (parse_flags
& PARSE_FLAG_TOK_NUM
)
3578 parse_number((char *)tokc
.str
.data
);
3579 } else if (t
== TOK_PPSTR
) {
3580 if (parse_flags
& PARSE_FLAG_TOK_STR
)
3581 parse_string((char *)tokc
.str
.data
, tokc
.str
.size
- 1);
3585 /* push back current token and set current token to 'last_tok'. Only
3586 identifier case handled for labels. */
3587 ST_INLN
void unget_tok(int last_tok
)
3590 TokenString
*str
= tok_str_alloc();
3591 tok_str_add2(str
, tok
, &tokc
);
3592 tok_str_add(str
, 0);
3593 begin_macro(str
, 1);
3597 /* ------------------------------------------------------------------------- */
3598 /* init preprocessor */
3600 static const char * const target_os_defs
=
3601 #ifdef TCC_TARGET_PE
3607 # if defined TCC_TARGET_MACHO
3609 # elif TARGETOS_FreeBSD
3611 # elif TARGETOS_FreeBSD_kernel
3612 "__FreeBSD_kernel__\0"
3613 # elif TARGETOS_NetBSD
3615 # elif TARGETOS_OpenBSD
3620 # if TARGETOS_ANDROID
3629 static void putdef(CString
*cs
, const char *p
)
3631 cstr_printf(cs
, "#define %s%s\n", p
, &" 1"[!!strchr(p
, ' ')*2]);
3634 static void putdefs(CString
*cs
, const char *p
)
3637 putdef(cs
, p
), p
= strchr(p
, 0) + 1;
3640 static void tcc_predefs(TCCState
*s1
, CString
*cs
, int is_asm
)
3644 sscanf(TCC_VERSION
, "%d.%d.%d", &a
, &b
, &c
);
3645 cstr_printf(cs
, "#define __TINYC__ %d\n", a
*10000 + b
*100 + c
);
3647 putdefs(cs
, target_machine_defs
);
3648 putdefs(cs
, target_os_defs
);
3650 #ifdef TCC_TARGET_ARM
3651 if (s1
->float_abi
== ARM_HARD_FLOAT
)
3652 putdef(cs
, "__ARM_PCS_VFP");
3655 putdef(cs
, "__ASSEMBLER__");
3656 if (s1
->output_type
== TCC_OUTPUT_PREPROCESS
)
3657 putdef(cs
, "__TCC_PP__");
3658 if (s1
->output_type
== TCC_OUTPUT_MEMORY
)
3659 putdef(cs
, "__TCC_RUN__");
3660 if (s1
->char_is_unsigned
)
3661 putdef(cs
, "__CHAR_UNSIGNED__");
3662 if (s1
->optimize
> 0)
3663 putdef(cs
, "__OPTIMIZE__");
3664 if (s1
->option_pthread
)
3665 putdef(cs
, "_REENTRANT");
3666 if (s1
->leading_underscore
)
3667 putdef(cs
, "__leading_underscore");
3668 #ifdef CONFIG_TCC_BCHECK
3669 if (s1
->do_bounds_check
)
3670 putdef(cs
, "__BOUNDS_CHECKING_ON");
3672 #ifdef CONFIG_TCC_BACKTRACE
3673 if (s1
->do_backtrace
)
3674 putdef(cs
, "__TCC_BACKTRACE_ENABLED__");
3676 cstr_printf(cs
, "#define __SIZEOF_POINTER__ %d\n", PTR_SIZE
);
3677 cstr_printf(cs
, "#define __SIZEOF_LONG__ %d\n", LONG_SIZE
);
3679 putdef(cs
, "__STDC__");
3680 cstr_printf(cs
, "#define __STDC_VERSION__ %dL\n", s1
->cversion
);
3682 /* load more predefs and __builtins */
3683 #if CONFIG_TCC_PREDEFS
3684 #include "tccdefs_.h" /* include as strings */
3686 "#include <tccdefs.h>\n" /* load at runtime */
3690 cstr_printf(cs
, "#define __BASE_FILE__ \"%s\"\n", file
->filename
);
3693 ST_FUNC
void preprocess_start(TCCState
*s1
, int filetype
)
3695 int is_asm
= !!(filetype
& (AFF_TYPE_ASM
|AFF_TYPE_ASMPP
));
3700 s1
->include_stack_ptr
= s1
->include_stack
;
3701 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3702 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3705 pp_debug_tok
= pp_debug_symv
= 0;
3707 s1
->pack_stack
[0] = 0;
3708 s1
->pack_stack_ptr
= s1
->pack_stack
;
3710 set_idnum('$', !is_asm
&& s1
->dollars_in_identifiers
? IS_ID
: 0);
3711 set_idnum('.', is_asm
? IS_ID
: 0);
3713 if (!(filetype
& AFF_TYPE_ASM
)) {
3715 tcc_predefs(s1
, &cstr
, is_asm
);
3716 if (s1
->cmdline_defs
.size
)
3717 cstr_cat(&cstr
, s1
->cmdline_defs
.data
, s1
->cmdline_defs
.size
);
3718 if (s1
->cmdline_incl
.size
)
3719 cstr_cat(&cstr
, s1
->cmdline_incl
.data
, s1
->cmdline_incl
.size
);
3720 //printf("%s\n", (char*)cstr.data);
3721 *s1
->include_stack_ptr
++ = file
;
3722 tcc_open_bf(s1
, "<command line>", cstr
.size
);
3723 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3727 parse_flags
= is_asm
? PARSE_FLAG_ASM_FILE
: 0;
3728 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3731 /* cleanup from error/setjmp */
3732 ST_FUNC
void preprocess_end(TCCState
*s1
)
3742 ST_FUNC
int set_idnum(int c
, int val
)
3744 int prev
= isidnum_table
[c
- CH_EOF
];
3745 isidnum_table
[c
- CH_EOF
] = val
;
3749 ST_FUNC
void tccpp_new(TCCState
*s
)
3754 /* init isid table */
3755 for(i
= CH_EOF
; i
<128; i
++)
3757 is_space(i
) ? IS_SPC
3762 for(i
= 128; i
<256; i
++)
3763 set_idnum(i
, IS_ID
);
3765 /* init allocators */
3766 tal_new(&toksym_alloc
, TOKSYM_TAL_LIMIT
, TOKSYM_TAL_SIZE
);
3767 tal_new(&tokstr_alloc
, TOKSTR_TAL_LIMIT
, TOKSTR_TAL_SIZE
);
3769 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3770 memset(s
->cached_includes_hash
, 0, sizeof s
->cached_includes_hash
);
3772 cstr_new(&cstr_buf
);
3773 cstr_realloc(&cstr_buf
, STRING_MAX_SIZE
);
3774 tok_str_new(&tokstr_buf
);
3775 tok_str_realloc(&tokstr_buf
, TOKSTR_MAX_SIZE
);
3777 tok_ident
= TOK_IDENT
;
3786 tok_alloc(p
, r
- p
- 1);
3790 /* we add dummy defines for some special macros to speed up tests
3791 and to have working defined() */
3792 define_push(TOK___LINE__
, MACRO_OBJ
, NULL
, NULL
);
3793 define_push(TOK___FILE__
, MACRO_OBJ
, NULL
, NULL
);
3794 define_push(TOK___DATE__
, MACRO_OBJ
, NULL
, NULL
);
3795 define_push(TOK___TIME__
, MACRO_OBJ
, NULL
, NULL
);
3796 define_push(TOK___COUNTER__
, MACRO_OBJ
, NULL
, NULL
);
3799 ST_FUNC
void tccpp_delete(TCCState
*s
)
3803 dynarray_reset(&s
->cached_includes
, &s
->nb_cached_includes
);
3806 n
= tok_ident
- TOK_IDENT
;
3807 if (n
> total_idents
)
3809 for(i
= 0; i
< n
; i
++)
3810 tal_free(toksym_alloc
, table_ident
[i
]);
3811 tcc_free(table_ident
);
3814 /* free static buffers */
3815 cstr_free(&tokcstr
);
3816 cstr_free(&cstr_buf
);
3817 cstr_free(¯o_equal_buf
);
3818 tok_str_free_str(tokstr_buf
.str
);
3820 /* free allocators */
3821 tal_delete(toksym_alloc
);
3822 toksym_alloc
= NULL
;
3823 tal_delete(tokstr_alloc
);
3824 tokstr_alloc
= NULL
;
3827 /* ------------------------------------------------------------------------- */
3828 /* tcc -E [-P[1]] [-dD} support */
3830 static void tok_print(const char *msg
, const int *str
)
3836 fp
= tcc_state
->ppfp
;
3837 fprintf(fp
, "%s", msg
);
3839 TOK_GET(&t
, &str
, &cval
);
3842 fprintf(fp
, &" %s"[s
], get_tok_str(t
, &cval
)), s
= 1;
3847 static void pp_line(TCCState
*s1
, BufferedFile
*f
, int level
)
3849 int d
= f
->line_num
- f
->line_ref
;
3854 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_NONE
) {
3856 } else if (level
== 0 && f
->line_ref
&& d
< 8) {
3858 fputs("\n", s1
->ppfp
), --d
;
3859 } else if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_STD
) {
3860 fprintf(s1
->ppfp
, "#line %d \"%s\"\n", f
->line_num
, f
->filename
);
3862 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
,
3863 level
> 0 ? " 1" : level
< 0 ? " 2" : "");
3865 f
->line_ref
= f
->line_num
;
3868 static void define_print(TCCState
*s1
, int v
)
3874 if (NULL
== s
|| NULL
== s
->d
)
3878 fprintf(fp
, "#define %s", get_tok_str(v
, NULL
));
3879 if (s
->type
.t
== MACRO_FUNC
) {
3884 fprintf(fp
,"%s", get_tok_str(a
->v
& ~SYM_FIELD
, NULL
));
3891 tok_print("", s
->d
);
3894 static void pp_debug_defines(TCCState
*s1
)
3905 pp_line(s1
, file
, 0);
3906 file
->line_ref
= ++file
->line_num
;
3910 vs
= get_tok_str(v
, NULL
);
3911 if (t
== TOK_DEFINE
) {
3912 define_print(s1
, v
);
3913 } else if (t
== TOK_UNDEF
) {
3914 fprintf(fp
, "#undef %s\n", vs
);
3915 } else if (t
== TOK_push_macro
) {
3916 fprintf(fp
, "#pragma push_macro(\"%s\")\n", vs
);
3917 } else if (t
== TOK_pop_macro
) {
3918 fprintf(fp
, "#pragma pop_macro(\"%s\")\n", vs
);
3923 static void pp_debug_builtins(TCCState
*s1
)
3926 for (v
= TOK_IDENT
; v
< tok_ident
; ++v
)
3927 define_print(s1
, v
);
3930 /* Add a space between tokens a and b to avoid unwanted textual pasting */
3931 static int pp_need_space(int a
, int b
)
3933 return 'E' == a
? '+' == b
|| '-' == b
3934 : '+' == a
? TOK_INC
== b
|| '+' == b
3935 : '-' == a
? TOK_DEC
== b
|| '-' == b
3936 : a
>= TOK_IDENT
? b
>= TOK_IDENT
3937 : a
== TOK_PPNUM
? b
>= TOK_IDENT
3941 /* maybe hex like 0x1e */
3942 static int pp_check_he0xE(int t
, const char *p
)
3944 if (t
== TOK_PPNUM
&& toup(strchr(p
, 0)[-1]) == 'E')
3949 /* Preprocess the current file */
3950 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3952 BufferedFile
**iptr
;
3953 int token_seen
, spcs
, level
;
3957 parse_flags
= PARSE_FLAG_PREPROCESS
3958 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
3959 | PARSE_FLAG_LINEFEED
3961 | PARSE_FLAG_ACCEPT_STRAYS
3963 /* Credits to Fabrice Bellard's initial revision to demonstrate its
3964 capability to compile and run itself, provided all numbers are
3965 given as decimals. tcc -E -P10 will do. */
3966 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_P10
)
3967 parse_flags
|= PARSE_FLAG_TOK_NUM
, s1
->Pflag
= 1;
3970 /* for PP benchmarks */
3971 do next(); while (tok
!= TOK_EOF
);
3975 if (s1
->dflag
& 1) {
3976 pp_debug_builtins(s1
);
3980 token_seen
= TOK_LINEFEED
, spcs
= 0, level
= 0;
3982 pp_line(s1
, file
->prev
, level
++);
3983 pp_line(s1
, file
, level
);
3985 iptr
= s1
->include_stack_ptr
;
3990 level
= s1
->include_stack_ptr
- iptr
;
3993 pp_line(s1
, *iptr
, 0);
3994 pp_line(s1
, file
, level
);
3996 if (s1
->dflag
& 7) {
3997 pp_debug_defines(s1
);
4002 if (is_space(tok
)) {
4003 if (spcs
< sizeof white
- 1)
4004 white
[spcs
++] = tok
;
4006 } else if (tok
== TOK_LINEFEED
) {
4008 if (token_seen
== TOK_LINEFEED
)
4011 } else if (token_seen
== TOK_LINEFEED
) {
4012 pp_line(s1
, file
, 0);
4013 } else if (spcs
== 0 && pp_need_space(token_seen
, tok
)) {
4014 white
[spcs
++] = ' ';
4017 white
[spcs
] = 0, fputs(white
, s1
->ppfp
), spcs
= 0;
4018 fputs(p
= get_tok_str(tok
, &tokc
), s1
->ppfp
);
4019 token_seen
= pp_check_he0xE(tok
, p
);
4024 /* ------------------------------------------------------------------------- */