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 unsigned char isidnum_table
[256 - CH_EOF
];
46 /* isidnum_table flags: */
51 static TokenString
*macro_stack
;
53 static const char tcc_keywords
[] =
54 #define DEF(id, str) str "\0"
59 /* WARNING: the content of this string encodes token numbers */
60 static const unsigned char tok_two_chars
[] =
62 "<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
63 "-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";
84 '.','.', 0xa8, // C++ token ?
85 '#','#', TOK_TWOSHARPS
,
89 static void next_nomacro_spc(void);
91 ST_FUNC
void skip(int c
)
94 tcc_error("'%c' expected (got \"%s\")", c
, get_tok_str(tok
, &tokc
));
98 ST_FUNC
void expect(const char *msg
)
100 tcc_error("%s expected", msg
);
103 ST_FUNC
void begin_macro(TokenString
*str
, int alloc
)
106 str
->prev
= macro_stack
;
107 str
->prev_ptr
= macro_ptr
;
108 macro_ptr
= str
->str
;
112 ST_FUNC
void end_macro(void)
114 TokenString
*str
= macro_stack
;
115 macro_stack
= str
->prev
;
116 macro_ptr
= str
->prev_ptr
;
117 if (str
->alloc
== 2) {
118 str
->alloc
= 3; /* just mark as finished */
120 tok_str_free(str
->str
);
126 /* ------------------------------------------------------------------------- */
127 /* CString handling */
128 static void cstr_realloc(CString
*cstr
, int new_size
)
133 size
= cstr
->size_allocated
;
135 size
= 8; /* no need to allocate a too small first string */
136 while (size
< new_size
)
138 data
= tcc_realloc(cstr
->data_allocated
, size
);
139 cstr
->data_allocated
= data
;
140 cstr
->size_allocated
= size
;
145 ST_FUNC
void cstr_ccat(CString
*cstr
, int ch
)
148 size
= cstr
->size
+ 1;
149 if (size
> cstr
->size_allocated
)
150 cstr_realloc(cstr
, size
);
151 ((unsigned char *)cstr
->data
)[size
- 1] = ch
;
155 ST_FUNC
void cstr_cat(CString
*cstr
, const char *str
)
167 /* add a wide char */
168 ST_FUNC
void cstr_wccat(CString
*cstr
, int ch
)
171 size
= cstr
->size
+ sizeof(nwchar_t
);
172 if (size
> cstr
->size_allocated
)
173 cstr_realloc(cstr
, size
);
174 *(nwchar_t
*)(((unsigned char *)cstr
->data
) + size
- sizeof(nwchar_t
)) = ch
;
178 ST_FUNC
void cstr_new(CString
*cstr
)
180 memset(cstr
, 0, sizeof(CString
));
183 /* free string and reset it to NULL */
184 ST_FUNC
void cstr_free(CString
*cstr
)
186 tcc_free(cstr
->data_allocated
);
190 /* reset string to empty */
191 ST_FUNC
void cstr_reset(CString
*cstr
)
197 static void add_char(CString
*cstr
, int c
)
199 if (c
== '\'' || c
== '\"' || c
== '\\') {
200 /* XXX: could be more precise if char or string */
201 cstr_ccat(cstr
, '\\');
203 if (c
>= 32 && c
<= 126) {
206 cstr_ccat(cstr
, '\\');
208 cstr_ccat(cstr
, 'n');
210 cstr_ccat(cstr
, '0' + ((c
>> 6) & 7));
211 cstr_ccat(cstr
, '0' + ((c
>> 3) & 7));
212 cstr_ccat(cstr
, '0' + (c
& 7));
217 /* ------------------------------------------------------------------------- */
218 /* allocate a new token */
219 static TokenSym
*tok_alloc_new(TokenSym
**pts
, const char *str
, int len
)
221 TokenSym
*ts
, **ptable
;
224 if (tok_ident
>= SYM_FIRST_ANOM
)
225 tcc_error("memory full (symbols)");
227 /* expand token table if needed */
228 i
= tok_ident
- TOK_IDENT
;
229 if ((i
% TOK_ALLOC_INCR
) == 0) {
230 ptable
= tcc_realloc(table_ident
, (i
+ TOK_ALLOC_INCR
) * sizeof(TokenSym
*));
231 table_ident
= ptable
;
234 ts
= tcc_malloc(sizeof(TokenSym
) + len
);
236 ts
->tok
= tok_ident
++;
237 ts
->sym_define
= NULL
;
238 ts
->sym_label
= NULL
;
239 ts
->sym_struct
= NULL
;
240 ts
->sym_identifier
= NULL
;
242 ts
->hash_next
= NULL
;
243 memcpy(ts
->str
, str
, len
);
249 #define TOK_HASH_INIT 1
250 #define TOK_HASH_FUNC(h, c) ((h) * 263 + (c))
252 /* find a token and add it if not found */
253 ST_FUNC TokenSym
*tok_alloc(const char *str
, int len
)
261 h
= TOK_HASH_FUNC(h
, ((unsigned char *)str
)[i
]);
262 h
&= (TOK_HASH_SIZE
- 1);
264 pts
= &hash_ident
[h
];
269 if (ts
->len
== len
&& !memcmp(ts
->str
, str
, len
))
271 pts
= &(ts
->hash_next
);
273 return tok_alloc_new(pts
, str
, len
);
276 /* XXX: buffer overflow */
277 /* XXX: float tokens */
278 ST_FUNC
const char *get_tok_str(int v
, CValue
*cv
)
280 static char buf
[STRING_MAX_SIZE
+ 1];
281 static CString cstr_buf
;
286 /* NOTE: to go faster, we give a fixed buffer for small strings */
287 cstr_reset(&cstr_buf
);
289 cstr_buf
.size_allocated
= sizeof(buf
);
295 /* XXX: not quite exact, but only useful for testing */
296 sprintf(p
, "%u", cv
->ui
);
300 /* XXX: not quite exact, but only useful for testing */
302 sprintf(p
, "%u", (unsigned)cv
->ull
);
304 sprintf(p
, "%llu", cv
->ull
);
308 cstr_ccat(&cstr_buf
, 'L');
310 cstr_ccat(&cstr_buf
, '\'');
311 add_char(&cstr_buf
, cv
->i
);
312 cstr_ccat(&cstr_buf
, '\'');
313 cstr_ccat(&cstr_buf
, '\0');
317 return (char*)cv
->cstr
->data
;
319 cstr_ccat(&cstr_buf
, 'L');
322 cstr_ccat(&cstr_buf
, '\"');
324 len
= cstr
->size
- 1;
326 add_char(&cstr_buf
, ((unsigned char *)cstr
->data
)[i
]);
328 len
= (cstr
->size
/ sizeof(nwchar_t
)) - 1;
330 add_char(&cstr_buf
, ((nwchar_t
*)cstr
->data
)[i
]);
332 cstr_ccat(&cstr_buf
, '\"');
333 cstr_ccat(&cstr_buf
, '\0');
340 return NULL
; /* should not happen */
342 /* above tokens have value, the ones below don't */
351 return strcpy(p
, "...");
353 return strcpy(p
, "<<=");
355 return strcpy(p
, ">>=");
358 /* search in two bytes table */
359 const unsigned char *q
= tok_two_chars
;
370 sprintf(buf
, "<%02x>", v
);
376 } else if (v
< tok_ident
) {
377 return table_ident
[v
- TOK_IDENT
]->str
;
378 } else if (v
>= SYM_FIRST_ANOM
) {
379 /* special name for anonymous symbol */
380 sprintf(p
, "L.%u", v
- SYM_FIRST_ANOM
);
382 /* should never happen */
387 return cstr_buf
.data
;
390 /* return the current character, handling end of block if necessary
392 ST_FUNC
int handle_eob(void)
394 BufferedFile
*bf
= file
;
397 /* only tries to read if really end of buffer */
398 if (bf
->buf_ptr
>= bf
->buf_end
) {
400 #if defined(PARSE_DEBUG)
405 len
= read(bf
->fd
, bf
->buffer
, len
);
412 bf
->buf_ptr
= bf
->buffer
;
413 bf
->buf_end
= bf
->buffer
+ len
;
414 *bf
->buf_end
= CH_EOB
;
416 if (bf
->buf_ptr
< bf
->buf_end
) {
417 return bf
->buf_ptr
[0];
419 bf
->buf_ptr
= bf
->buf_end
;
424 /* read next char from current input file and handle end of input buffer */
425 ST_INLN
void inp(void)
427 ch
= *(++(file
->buf_ptr
));
428 /* end of buffer/file handling */
433 /* handle '\[\r]\n' */
434 static int handle_stray_noerror(void)
441 } else if (ch
== '\r') {
455 static void handle_stray(void)
457 if (handle_stray_noerror())
458 tcc_error("stray '\\' in program");
461 /* skip the stray and handle the \\n case. Output an error if
462 incorrect char after the stray */
463 static int handle_stray1(uint8_t *p
)
468 if (p
>= file
->buf_end
) {
475 if (handle_stray_noerror()) {
476 if (!(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
477 tcc_error("stray '\\' in program");
478 *--file
->buf_ptr
= '\\';
485 /* handle just the EOB case, but not stray */
486 #define PEEKC_EOB(c, p)\
497 /* handle the complicated stray case */
503 c = handle_stray1(p);\
508 /* input with '\[\r]\n' handling. Note that this function cannot
509 handle other characters after '\', so you cannot call it inside
510 strings or comments */
511 ST_FUNC
void minp(void)
519 /* single line C++ comments */
520 static uint8_t *parse_line_comment(uint8_t *p
)
528 if (c
== '\n' || c
== CH_EOF
) {
530 } else if (c
== '\\') {
539 } else if (c
== '\r') {
557 ST_FUNC
uint8_t *parse_comment(uint8_t *p
)
566 if (c
== '\n' || c
== '*' || c
== '\\')
570 if (c
== '\n' || c
== '*' || c
== '\\')
574 /* now we can handle all the cases */
578 } else if (c
== '*') {
584 } else if (c
== '/') {
586 } else if (c
== '\\') {
591 /* skip '\[\r]\n', otherwise just skip the stray */
597 } else if (c
== '\r') {
614 /* stray, eob or eof */
619 tcc_error("unexpected end of file in comment");
620 } else if (c
== '\\') {
632 static inline void skip_spaces(void)
634 while (isidnum_table
[ch
- CH_EOF
] & IS_SPC
)
638 static inline int check_space(int t
, int *spc
)
640 if (t
< 256 && (isidnum_table
[t
- CH_EOF
] & IS_SPC
)) {
649 /* parse a string without interpreting escapes */
650 static uint8_t *parse_pp_string(uint8_t *p
,
651 int sep
, CString
*str
)
659 } else if (c
== '\\') {
665 /* XXX: indicate line number of start of string */
666 tcc_error("missing terminating %c character", sep
);
667 } else if (c
== '\\') {
668 /* escape : just skip \[\r]\n */
673 } else if (c
== '\r') {
676 expect("'\n' after '\r'");
679 } else if (c
== CH_EOF
) {
680 goto unterminated_string
;
683 cstr_ccat(str
, '\\');
689 } else if (c
== '\n') {
692 } else if (c
== '\r') {
696 cstr_ccat(str
, '\r');
712 /* skip block of text until #else, #elif or #endif. skip also pairs of
714 static void preprocess_skip(void)
716 int a
, start_of_line
, c
, in_warn_or_error
;
723 in_warn_or_error
= 0;
744 } else if (c
== '\\') {
745 ch
= file
->buf_ptr
[0];
746 handle_stray_noerror();
753 if (in_warn_or_error
)
755 p
= parse_pp_string(p
, c
, NULL
);
759 if (in_warn_or_error
)
766 p
= parse_comment(p
);
767 } else if (ch
== '/') {
768 p
= parse_line_comment(p
);
778 (tok
== TOK_ELSE
|| tok
== TOK_ELIF
|| tok
== TOK_ENDIF
))
780 if (tok
== TOK_IF
|| tok
== TOK_IFDEF
|| tok
== TOK_IFNDEF
)
782 else if (tok
== TOK_ENDIF
)
784 else if( tok
== TOK_ERROR
|| tok
== TOK_WARNING
)
785 in_warn_or_error
= 1;
786 else if (tok
== TOK_LINEFEED
)
788 } else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
789 p
= parse_line_comment(p
);
802 /* ParseState handling */
804 /* XXX: currently, no include file info is stored. Thus, we cannot display
805 accurate messages if the function or data definition spans multiple
808 /* save current parse state in 's' */
809 ST_FUNC
void save_parse_state(ParseState
*s
)
811 s
->line_num
= file
->line_num
;
812 s
->macro_ptr
= macro_ptr
;
817 /* restore parse state from 's' */
818 ST_FUNC
void restore_parse_state(ParseState
*s
)
820 file
->line_num
= s
->line_num
;
821 macro_ptr
= s
->macro_ptr
;
826 /* return the number of additional 'ints' necessary to store the
828 static inline int tok_size(const int *p
)
843 return 1 + ((sizeof(CString
) + ((CString
*)(p
+1))->size
+ 3) >> 2);
849 return 1 + LDOUBLE_SIZE
/ 4;
855 /* token string handling */
857 ST_INLN
void tok_str_new(TokenString
*s
)
861 s
->allocated_len
= 0;
862 s
->last_line_num
= -1;
865 ST_FUNC
void tok_str_free(int *str
)
870 static int *tok_str_realloc(TokenString
*s
)
874 if (s
->allocated_len
== 0) {
877 len
= s
->allocated_len
* 2;
879 str
= tcc_realloc(s
->str
, len
* sizeof(int));
880 s
->allocated_len
= len
;
885 ST_FUNC
void tok_str_add(TokenString
*s
, int t
)
891 if (len
>= s
->allocated_len
)
892 str
= tok_str_realloc(s
);
897 static void tok_str_add2(TokenString
*s
, int t
, CValue
*cv
)
904 /* allocate space for worst case */
905 if (len
+ TOK_MAX_SIZE
> s
->allocated_len
)
906 str
= tok_str_realloc(s
);
915 str
[len
++] = cv
->tab
[0];
925 nb_words
= (sizeof(CString
) + cv
->cstr
->size
+ 3) >> 2;
926 while ((len
+ nb_words
) > s
->allocated_len
)
927 str
= tok_str_realloc(s
);
928 cstr
= (CString
*)(str
+ len
);
930 cstr
->size
= cv
->cstr
->size
;
931 cstr
->data_allocated
= NULL
;
932 cstr
->size_allocated
= cstr
->size
;
933 memcpy((char *)cstr
+ sizeof(CString
),
934 cv
->cstr
->data
, cstr
->size
);
941 #if LDOUBLE_SIZE == 8
944 str
[len
++] = cv
->tab
[0];
945 str
[len
++] = cv
->tab
[1];
947 #if LDOUBLE_SIZE == 12
949 str
[len
++] = cv
->tab
[0];
950 str
[len
++] = cv
->tab
[1];
951 str
[len
++] = cv
->tab
[2];
952 #elif LDOUBLE_SIZE == 16
954 str
[len
++] = cv
->tab
[0];
955 str
[len
++] = cv
->tab
[1];
956 str
[len
++] = cv
->tab
[2];
957 str
[len
++] = cv
->tab
[3];
958 #elif LDOUBLE_SIZE != 8
959 #error add long double size support
968 /* add the current parse token in token string 's' */
969 ST_FUNC
void tok_str_add_tok(TokenString
*s
)
973 /* save line number info */
974 if (file
->line_num
!= s
->last_line_num
) {
975 s
->last_line_num
= file
->line_num
;
976 cval
.i
= s
->last_line_num
;
977 tok_str_add2(s
, TOK_LINENUM
, &cval
);
979 tok_str_add2(s
, tok
, &tokc
);
982 /* get a token from an integer array and increment pointer
983 accordingly. we code it as a macro to avoid pointer aliasing. */
984 static inline void TOK_GET(int *t
, const int **pp
, CValue
*cv
)
1003 cv
->cstr
= (CString
*)p
;
1004 cv
->cstr
->data
= (char *)p
+ sizeof(CString
);
1005 p
+= (sizeof(CString
) + cv
->cstr
->size
+ 3) >> 2;
1013 #if LDOUBLE_SIZE == 16
1015 #elif LDOUBLE_SIZE == 12
1017 #elif LDOUBLE_SIZE == 8
1020 # error add long double size support
1033 static int macro_is_equal(const int *a
, const int *b
)
1035 char buf
[STRING_MAX_SIZE
+ 1];
1039 TOK_GET(&t
, &a
, &cv
);
1040 pstrcpy(buf
, sizeof buf
, get_tok_str(t
, &cv
));
1041 TOK_GET(&t
, &b
, &cv
);
1042 if (strcmp(buf
, get_tok_str(t
, &cv
)))
1048 static void pp_line(TCCState
*s1
, BufferedFile
*f
, int level
)
1050 int d
= f
->line_num
- f
->line_ref
;
1051 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_NONE
1052 || (level
== 0 && f
->line_ref
&& d
< 8))
1055 fputs("\n", s1
->ppfp
), --d
;
1058 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_STD
) {
1059 fprintf(s1
->ppfp
, "#line %d \"%s\"\n", f
->line_num
, f
->filename
);
1062 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
,
1063 level
> 0 ? " 1" : level
< 0 ? " 2" : "");
1065 f
->line_ref
= f
->line_num
;
1068 static void tok_print(const char *msg
, const int *str
)
1070 FILE *pr
= tcc_state
->ppfp
;
1074 fprintf(pr
, "%s ", msg
);
1076 TOK_GET(&t
, &str
, &cval
);
1079 fprintf(pr
,"%s", get_tok_str(t
, &cval
));
1084 static int define_print_prepared(Sym
*s
)
1086 if (!s
|| !tcc_state
->ppfp
|| tcc_state
->dflag
== 0)
1089 if (s
->v
< TOK_IDENT
|| s
->v
>= tok_ident
)
1094 pp_line(tcc_state
, file
, 0);
1095 file
->line_ref
= ++file
->line_num
;
1100 static void define_print(int v
)
1102 FILE *pr
= tcc_state
->ppfp
;
1106 if (define_print_prepared(s
) == 0)
1109 fprintf(pr
, "// #define %s", get_tok_str(v
, NULL
));
1110 if (s
->type
.t
== MACRO_FUNC
) {
1115 fprintf(pr
,"%s", get_tok_str(a
->v
& ~SYM_FIELD
, NULL
));
1122 tok_print("", s
->d
);
1125 static void undef_print(int v
)
1127 FILE *pr
= tcc_state
->ppfp
;
1131 if (define_print_prepared(s
) == 0)
1134 fprintf(pr
, "// #undef %s\n", get_tok_str(s
->v
, NULL
));
1137 ST_FUNC
void print_defines(void)
1139 Sym
*top
= define_stack
;
1141 define_print(top
->v
);
1146 /* defines handling */
1147 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1152 if (s
&& !macro_is_equal(s
->d
, str
))
1153 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1155 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1157 s
->next
= first_arg
;
1158 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1161 /* undefined a define symbol. Its name is just set to zero */
1162 ST_FUNC
void define_undef(Sym
*s
)
1166 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1167 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1170 ST_INLN Sym
*define_find(int v
)
1173 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1175 return table_ident
[v
]->sym_define
;
1178 /* free define stack until top reaches 'b' */
1179 ST_FUNC
void free_defines(Sym
*b
)
1187 /* do not free args or predefined defines */
1189 tok_str_free(top
->d
);
1191 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1192 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1200 ST_FUNC Sym
*label_find(int v
)
1203 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1205 return table_ident
[v
]->sym_label
;
1208 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1211 s
= sym_push2(ptop
, v
, 0, 0);
1213 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1214 if (ptop
== &global_label_stack
) {
1215 /* modify the top most local identifier, so that
1216 sym_identifier will point to 's' when popped */
1218 ps
= &(*ps
)->prev_tok
;
1225 /* pop labels until element last is reached. Look if any labels are
1226 undefined. Define symbols if '&&label' was used. */
1227 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
)
1230 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1232 if (s
->r
== LABEL_DECLARED
) {
1233 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1234 } else if (s
->r
== LABEL_FORWARD
) {
1235 tcc_error("label '%s' used but not defined",
1236 get_tok_str(s
->v
, NULL
));
1239 /* define corresponding symbol. A size of
1241 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1245 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1251 /* eval an expression for #if/#elif */
1252 static int expr_preprocess(void)
1258 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1259 next(); /* do macro subst */
1260 if (tok
== TOK_DEFINED
) {
1265 c
= define_find(tok
) != 0;
1270 } else if (tok
>= TOK_IDENT
) {
1271 /* if undefined macro */
1275 tok_str_add_tok(&str
);
1277 tok_str_add(&str
, -1); /* simulate end of file */
1278 tok_str_add(&str
, 0);
1279 /* now evaluate C constant expression */
1280 begin_macro(&str
, 0);
1288 /* parse after #define */
1289 ST_FUNC
void parse_define(void)
1291 Sym
*s
, *first
, **ps
;
1292 int v
, t
, varg
, is_vaargs
, spc
;
1293 int saved_parse_flags
= parse_flags
;
1298 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1299 /* XXX: should check if same macro (ANSI) */
1302 /* '(' must be just after macro definition for MACRO_FUNC */
1303 parse_flags
|= PARSE_FLAG_SPACES
;
1308 if (tok
!= ')') for (;;) {
1312 if (varg
== TOK_DOTS
) {
1313 varg
= TOK___VA_ARGS__
;
1315 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1319 if (varg
< TOK_IDENT
)
1321 tcc_error("bad macro parameter list");
1322 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1327 if (tok
!= ',' || is_vaargs
)
1336 parse_flags
|= PARSE_FLAG_ACCEPT_STRAYS
| PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
;
1337 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1338 /* remove spaces around ## and after '#' */
1339 if (TOK_TWOSHARPS
== tok
) {
1345 } else if ('#' == tok
) {
1347 } else if (check_space(tok
, &spc
)) {
1350 tok_str_add2(&str
, tok
, &tokc
);
1355 parse_flags
= saved_parse_flags
;
1357 --str
.len
; /* remove trailing space */
1358 tok_str_add(&str
, 0);
1361 tcc_error("'##' cannot appear at either end of macro");
1362 define_push(v
, t
, str
.str
, first
);
1366 static inline int hash_cached_include(const char *filename
)
1368 const unsigned char *s
;
1372 s
= (unsigned char *) filename
;
1374 h
= TOK_HASH_FUNC(h
, *s
);
1377 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1381 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
)
1385 h
= hash_cached_include(filename
);
1386 i
= s1
->cached_includes_hash
[h
];
1390 e
= s1
->cached_includes
[i
- 1];
1391 if (0 == PATHCMP(e
->filename
, filename
))
1398 static inline void add_cached_include(TCCState
*s1
, const char *filename
, int ifndef_macro
)
1403 if (search_cached_include(s1
, filename
))
1406 printf("adding cached '%s' %s\n", filename
, get_tok_str(ifndef_macro
, NULL
));
1408 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1409 strcpy(e
->filename
, filename
);
1410 e
->ifndef_macro
= ifndef_macro
;
1411 dynarray_add((void ***)&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1412 /* add in hash table */
1413 h
= hash_cached_include(filename
);
1414 e
->hash_next
= s1
->cached_includes_hash
[h
];
1415 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1418 static void pragma_parse(TCCState
*s1
)
1421 if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1425 if (next(), tok
!= '(')
1427 if (next(), tok
!= TOK_STR
)
1429 v
= tok_alloc(tokc
.cstr
->data
, tokc
.cstr
->size
- 1)->tok
;
1430 if (next(), tok
!= ')')
1432 if (t
== TOK_push_macro
) {
1433 while (NULL
== (s
= define_find(v
)))
1434 define_push(v
, 0, NULL
, NULL
);
1435 s
->type
.ref
= s
; /* set push boundary */
1437 for (s
= define_stack
; s
; s
= s
->prev
)
1438 if (s
->v
== v
&& s
->type
.ref
== s
) {
1444 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1446 tcc_warning("unbalanced #pragma pop_macro");
1448 } else if (tok
== TOK_once
) {
1449 add_cached_include(s1
, file
->filename
, TOK_once
);
1451 } else if (s1
->ppfp
) {
1452 /* tcc -E: keep pragmas below unchanged */
1454 unget_tok(TOK_PRAGMA
);
1456 unget_tok(TOK_LINEFEED
);
1458 } else if (tok
== TOK_pack
) {
1460 #pragma pack(1) // set
1461 #pragma pack() // reset to default
1462 #pragma pack(push,1) // push & set
1463 #pragma pack(pop) // restore previous */
1466 if (tok
== TOK_ASM_pop
) {
1468 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1470 tcc_error("out of pack stack");
1472 s1
->pack_stack_ptr
--;
1476 if (tok
== TOK_ASM_push
) {
1478 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1480 s1
->pack_stack_ptr
++;
1483 if (tok
!= TOK_CINT
)
1486 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1490 *s1
->pack_stack_ptr
= val
;
1495 } else if (tok
== TOK_comment
) {
1505 file
= tcc_strdup((char *)tokc
.cstr
->data
);
1506 dynarray_add((void ***)&s1
->pragma_libs
, &s1
->nb_pragma_libs
, file
);
1512 if (s1
->warn_unsupported
)
1513 tcc_warning("#pragma %s is ignored", get_tok_str(tok
, &tokc
));
1518 tcc_error("malformed #pragma directive");
1522 /* is_bof is true if first non space token at beginning of file */
1523 ST_FUNC
void preprocess(int is_bof
)
1525 TCCState
*s1
= tcc_state
;
1526 int i
, c
, n
, saved_parse_flags
;
1530 saved_parse_flags
= parse_flags
;
1531 parse_flags
= PARSE_FLAG_PREPROCESS
1532 | PARSE_FLAG_TOK_NUM
1533 | PARSE_FLAG_TOK_STR
1534 | PARSE_FLAG_LINEFEED
1535 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
1547 s
= define_find(tok
);
1548 /* undefine symbol by putting an invalid name */
1553 case TOK_INCLUDE_NEXT
:
1554 ch
= file
->buf_ptr
[0];
1555 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1560 } else if (ch
== '\"') {
1565 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1566 if ((q
- buf
) < sizeof(buf
) - 1)
1569 if (handle_stray_noerror() == 0)
1577 /* eat all spaces and comments after include */
1578 /* XXX: slightly incorrect */
1579 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1583 /* computed #include : either we have only strings or
1584 we have anything enclosed in '<>' */
1587 if (tok
== TOK_STR
) {
1588 while (tok
!= TOK_LINEFEED
) {
1589 if (tok
!= TOK_STR
) {
1591 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1593 pstrcat(buf
, sizeof(buf
), (char *)tokc
.cstr
->data
);
1599 while (tok
!= TOK_LINEFEED
) {
1600 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1604 /* check syntax and remove '<>' */
1605 if (len
< 2 || buf
[0] != '<' || buf
[len
- 1] != '>')
1606 goto include_syntax
;
1607 memmove(buf
, buf
+ 1, len
- 2);
1608 buf
[len
- 2] = '\0';
1613 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1614 tcc_error("#include recursion too deep");
1615 /* store current file in stack, but increment stack later below */
1616 *s1
->include_stack_ptr
= file
;
1618 n
= s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1619 for (i
= -2; i
< n
; ++i
) {
1620 char buf1
[sizeof file
->filename
];
1626 /* check absolute include path */
1627 if (!IS_ABSPATH(buf
))
1630 i
= n
; /* force end loop */
1632 } else if (i
== -1) {
1633 /* search in current dir if "header.h" */
1636 path
= file
->filename
;
1637 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1640 /* search in all the include paths */
1641 if (i
< s1
->nb_include_paths
)
1642 path
= s1
->include_paths
[i
];
1644 path
= s1
->sysinclude_paths
[i
- s1
->nb_include_paths
];
1645 pstrcpy(buf1
, sizeof(buf1
), path
);
1646 pstrcat(buf1
, sizeof(buf1
), "/");
1649 pstrcat(buf1
, sizeof(buf1
), buf
);
1651 if (tok
== TOK_INCLUDE_NEXT
)
1652 for (f
= s1
->include_stack_ptr
; f
>= s1
->include_stack
; --f
)
1653 if (0 == PATHCMP((*f
)->filename
, buf1
)) {
1655 printf("%s: #include_next skipping %s\n", file
->filename
, buf1
);
1657 goto include_trynext
;
1660 e
= search_cached_include(s1
, buf1
);
1661 if (e
&& (define_find(e
->ifndef_macro
) || e
->ifndef_macro
== TOK_once
)) {
1662 /* no need to parse the include because the 'ifndef macro'
1665 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1670 if (tcc_open(s1
, buf1
) < 0)
1675 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1677 /* update target deps */
1678 dynarray_add((void ***)&s1
->target_deps
, &s1
->nb_target_deps
,
1680 /* push current file in stack */
1681 ++s1
->include_stack_ptr
;
1682 /* add include file debug info */
1684 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1685 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1686 ch
= file
->buf_ptr
[0];
1689 tcc_error("include file '%s' not found", buf
);
1696 c
= expr_preprocess();
1702 if (tok
< TOK_IDENT
)
1703 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1707 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1709 file
->ifndef_macro
= tok
;
1712 c
= (define_find(tok
) != 0) ^ c
;
1714 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1715 tcc_error("memory full (ifdef)");
1716 *s1
->ifdef_stack_ptr
++ = c
;
1719 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1720 tcc_error("#else without matching #if");
1721 if (s1
->ifdef_stack_ptr
[-1] & 2)
1722 tcc_error("#else after #else");
1723 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1726 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1727 tcc_error("#elif without matching #if");
1728 c
= s1
->ifdef_stack_ptr
[-1];
1730 tcc_error("#elif after #else");
1731 /* last #if/#elif expression was true: we skip */
1734 c
= expr_preprocess();
1735 s1
->ifdef_stack_ptr
[-1] = c
;
1737 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1738 file
->ifndef_macro
= 0;
1748 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1749 tcc_error("#endif without matching #if");
1750 s1
->ifdef_stack_ptr
--;
1751 /* '#ifndef macro' was at the start of file. Now we check if
1752 an '#endif' is exactly at the end of file */
1753 if (file
->ifndef_macro
&&
1754 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1755 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1756 /* need to set to zero to avoid false matches if another
1757 #ifndef at middle of file */
1758 file
->ifndef_macro
= 0;
1759 while (tok
!= TOK_LINEFEED
)
1761 tok_flags
|= TOK_FLAG_ENDIF
;
1766 n
= strtoul((char*)tokc
.cstr
->data
, &q
, 10);
1770 if (tok
!= TOK_CINT
)
1772 tcc_error("wrong #line format");
1776 if (tok
!= TOK_LINEFEED
) {
1778 pstrcpy(file
->filename
, sizeof(file
->filename
), (char *)tokc
.cstr
->data
);
1779 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
1786 total_lines
+= file
->line_num
- n
;
1789 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1794 ch
= file
->buf_ptr
[0];
1797 while (ch
!= '\n' && ch
!= CH_EOF
) {
1798 if ((q
- buf
) < sizeof(buf
) - 1)
1801 if (handle_stray_noerror() == 0)
1808 tcc_error("#error %s", buf
);
1810 tcc_warning("#warning %s", buf
);
1818 /* ignore gas line comment in an 'S' file. */
1819 if (saved_parse_flags
& PARSE_FLAG_ASM_FILE
)
1821 if (tok
== '!' && is_bof
)
1822 /* '!' is ignored at beginning to allow C scripts. */
1824 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
1826 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
);
1829 /* ignore other preprocess commands or #! for C scripts */
1830 while (tok
!= TOK_LINEFEED
)
1833 parse_flags
= saved_parse_flags
;
1836 /* evaluate escape codes in a string. */
1837 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
1852 case '0': case '1': case '2': case '3':
1853 case '4': case '5': case '6': case '7':
1854 /* at most three octal digits */
1859 n
= n
* 8 + c
- '0';
1863 n
= n
* 8 + c
- '0';
1868 goto add_char_nonext
;
1876 if (c
>= 'a' && c
<= 'f')
1878 else if (c
>= 'A' && c
<= 'F')
1888 goto add_char_nonext
;
1912 goto invalid_escape
;
1922 if (c
>= '!' && c
<= '~')
1923 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
1925 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
1932 cstr_ccat(outstr
, c
);
1934 cstr_wccat(outstr
, c
);
1936 /* add a trailing '\0' */
1938 cstr_ccat(outstr
, '\0');
1940 cstr_wccat(outstr
, '\0');
1943 void parse_string(const char *s
, int len
)
1945 uint8_t buf
[1000], *p
= buf
;
1948 if ((is_long
= *s
== 'L'))
1952 if (len
>= sizeof buf
)
1953 p
= tcc_malloc(len
+ 1);
1957 cstr_reset(&tokcstr
);
1958 parse_escape_string(&tokcstr
, p
, is_long
);
1964 /* XXX: make it portable */
1968 char_size
= sizeof(nwchar_t
);
1969 if (tokcstr
.size
<= char_size
)
1970 tcc_error("empty character constant");
1971 if (tokcstr
.size
> 2 * char_size
)
1972 tcc_warning("multi-character character constant");
1974 tokc
.i
= *(int8_t *)tokcstr
.data
;
1977 tokc
.i
= *(nwchar_t
*)tokcstr
.data
;
1981 tokc
.cstr
= &tokcstr
;
1989 /* we use 64 bit numbers */
1992 /* bn = (bn << shift) | or_val */
1993 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
1997 for(i
=0;i
<BN_SIZE
;i
++) {
1999 bn
[i
] = (v
<< shift
) | or_val
;
2000 or_val
= v
>> (32 - shift
);
2004 static void bn_zero(unsigned int *bn
)
2007 for(i
=0;i
<BN_SIZE
;i
++) {
2012 /* parse number in null terminated string 'p' and return it in the
2014 static void parse_number(const char *p
)
2016 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
2018 unsigned int bn
[BN_SIZE
];
2029 goto float_frac_parse
;
2030 } else if (t
== '0') {
2031 if (ch
== 'x' || ch
== 'X') {
2035 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
2041 /* parse all digits. cannot check octal numbers at this stage
2042 because of floating point constants */
2044 if (ch
>= 'a' && ch
<= 'f')
2046 else if (ch
>= 'A' && ch
<= 'F')
2054 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
2056 tcc_error("number too long");
2062 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
2063 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
2065 /* NOTE: strtox should support that for hexa numbers, but
2066 non ISOC99 libcs do not support it, so we prefer to do
2068 /* hexadecimal or binary floats */
2069 /* XXX: handle overflows */
2081 } else if (t
>= 'a') {
2083 } else if (t
>= 'A') {
2088 bn_lshift(bn
, shift
, t
);
2095 if (t
>= 'a' && t
<= 'f') {
2097 } else if (t
>= 'A' && t
<= 'F') {
2099 } else if (t
>= '0' && t
<= '9') {
2105 tcc_error("invalid digit");
2106 bn_lshift(bn
, shift
, t
);
2111 if (ch
!= 'p' && ch
!= 'P')
2118 } else if (ch
== '-') {
2122 if (ch
< '0' || ch
> '9')
2123 expect("exponent digits");
2124 while (ch
>= '0' && ch
<= '9') {
2125 exp_val
= exp_val
* 10 + ch
- '0';
2128 exp_val
= exp_val
* s
;
2130 /* now we can generate the number */
2131 /* XXX: should patch directly float number */
2132 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2133 d
= ldexp(d
, exp_val
- frac_bits
);
2138 /* float : should handle overflow */
2140 } else if (t
== 'L') {
2142 #ifdef TCC_TARGET_PE
2147 /* XXX: not large enough */
2148 tokc
.ld
= (long double)d
;
2155 /* decimal floats */
2157 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2162 while (ch
>= '0' && ch
<= '9') {
2163 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2169 if (ch
== 'e' || ch
== 'E') {
2170 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2174 if (ch
== '-' || ch
== '+') {
2175 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2180 if (ch
< '0' || ch
> '9')
2181 expect("exponent digits");
2182 while (ch
>= '0' && ch
<= '9') {
2183 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2195 tokc
.f
= strtof(token_buf
, NULL
);
2196 } else if (t
== 'L') {
2198 #ifdef TCC_TARGET_PE
2200 tokc
.d
= strtod(token_buf
, NULL
);
2203 tokc
.ld
= strtold(token_buf
, NULL
);
2207 tokc
.d
= strtod(token_buf
, NULL
);
2211 unsigned long long n
, n1
;
2212 int lcount
, ucount
, must_64bit
;
2215 /* integer number */
2218 if (b
== 10 && *q
== '0') {
2225 /* no need for checks except for base 10 / 8 errors */
2235 tcc_error("invalid digit");
2238 /* detect overflow */
2239 /* XXX: this test is not reliable */
2241 tcc_error("integer constant overflow");
2244 /* Determine the characteristics (unsigned and/or 64bit) the type of
2245 the constant must have according to the constant suffix(es) */
2246 lcount
= ucount
= must_64bit
= 0;
2252 tcc_error("three 'l's in integer constant");
2253 if (lcount
&& *(p
- 1) != ch
)
2254 tcc_error("incorrect integer suffix: %s", p1
);
2256 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
2261 } else if (t
== 'U') {
2263 tcc_error("two 'u's in integer constant");
2271 /* Whether 64 bits are needed to hold the constant's value */
2272 if (n
& 0xffffffff00000000LL
|| must_64bit
) {
2280 /* Whether type must be unsigned to hold the constant's value */
2281 if (ucount
|| ((n1
>> 31) && (b
!= 10))) {
2282 if (tok
== TOK_CLLONG
)
2286 /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
2287 } else if (n1
>> 31) {
2288 if (tok
== TOK_CINT
)
2291 tcc_error("integer constant overflow");
2294 if (tok
== TOK_CINT
|| tok
== TOK_CUINT
)
2300 tcc_error("invalid number\n");
2304 #define PARSE2(c1, tok1, c2, tok2) \
2315 /* return next token without macro substitution */
2316 static inline void next_nomacro1(void)
2331 if (parse_flags
& PARSE_FLAG_SPACES
)
2332 goto keep_tok_flags
;
2333 while (isidnum_table
[*p
- CH_EOF
] & IS_SPC
)
2342 /* first look if it is in fact an end of buffer */
2343 c
= handle_stray1(p
);
2350 TCCState
*s1
= tcc_state
;
2351 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2352 && !(tok_flags
& TOK_FLAG_EOF
)) {
2353 tok_flags
|= TOK_FLAG_EOF
;
2355 goto keep_tok_flags
;
2356 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2358 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2359 tcc_error("missing #endif");
2360 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2361 /* no include left : end of file. */
2364 tok_flags
&= ~TOK_FLAG_EOF
;
2365 /* pop include file */
2367 /* test if previous '#endif' was after a #ifdef at
2369 if (tok_flags
& TOK_FLAG_ENDIF
) {
2371 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2373 add_cached_include(s1
, file
->filename
, file
->ifndef_macro_saved
);
2374 tok_flags
&= ~TOK_FLAG_ENDIF
;
2377 /* add end of include file debug info */
2378 if (tcc_state
->do_debug
) {
2379 put_stabd(N_EINCL
, 0, 0);
2381 /* pop include stack */
2383 s1
->include_stack_ptr
--;
2392 tok_flags
|= TOK_FLAG_BOL
;
2395 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2398 goto keep_tok_flags
;
2403 if ((tok_flags
& TOK_FLAG_BOL
) &&
2404 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2406 preprocess(tok_flags
& TOK_FLAG_BOF
);
2412 tok
= TOK_TWOSHARPS
;
2414 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2415 p
= parse_line_comment(p
- 1);
2424 /* dollar is allowed to start identifiers when not parsing asm */
2426 if (!(isidnum_table
[c
- CH_EOF
] & IS_ID
)
2427 || (parse_flags
& PARSE_FLAG_ASM_FILE
))
2430 case 'a': case 'b': case 'c': case 'd':
2431 case 'e': case 'f': case 'g': case 'h':
2432 case 'i': case 'j': case 'k': case 'l':
2433 case 'm': case 'n': case 'o': case 'p':
2434 case 'q': case 'r': case 's': case 't':
2435 case 'u': case 'v': case 'w': case 'x':
2437 case 'A': case 'B': case 'C': case 'D':
2438 case 'E': case 'F': case 'G': case 'H':
2439 case 'I': case 'J': case 'K':
2440 case 'M': case 'N': case 'O': case 'P':
2441 case 'Q': case 'R': case 'S': case 'T':
2442 case 'U': case 'V': case 'W': case 'X':
2448 h
= TOK_HASH_FUNC(h
, c
);
2449 while (c
= *++p
, isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2450 h
= TOK_HASH_FUNC(h
, c
);
2455 /* fast case : no stray found, so we have the full token
2456 and we have already hashed it */
2458 h
&= (TOK_HASH_SIZE
- 1);
2459 pts
= &hash_ident
[h
];
2464 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2466 pts
= &(ts
->hash_next
);
2468 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2472 cstr_reset(&tokcstr
);
2475 cstr_ccat(&tokcstr
, *p1
);
2481 while (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
)) {
2482 cstr_ccat(&tokcstr
, c
);
2485 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2491 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2493 goto parse_ident_fast
;
2496 if (c
== '\'' || c
== '\"') {
2500 cstr_reset(&tokcstr
);
2501 cstr_ccat(&tokcstr
, 'L');
2502 goto parse_ident_slow
;
2507 case '0': case '1': case '2': case '3':
2508 case '4': case '5': case '6': case '7':
2510 cstr_reset(&tokcstr
);
2511 /* after the first digit, accept digits, alpha, '.' or sign if
2512 prefixed by 'eEpP' */
2516 cstr_ccat(&tokcstr
, c
);
2518 if (!((isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2520 || ((c
== '+' || c
== '-')
2521 && (t
== 'e' || t
== 'E' || t
== 'p' || t
== 'P')
2525 /* We add a trailing '\0' to ease parsing */
2526 cstr_ccat(&tokcstr
, '\0');
2527 tokc
.cstr
= &tokcstr
;
2532 /* special dot handling because it can also start a number */
2535 cstr_reset(&tokcstr
);
2536 cstr_ccat(&tokcstr
, '.');
2538 } else if (c
== '.') {
2552 cstr_reset(&tokcstr
);
2554 cstr_ccat(&tokcstr
, 'L');
2555 cstr_ccat(&tokcstr
, c
);
2556 p
= parse_pp_string(p
, c
, &tokcstr
);
2557 cstr_ccat(&tokcstr
, c
);
2558 cstr_ccat(&tokcstr
, '\0');
2559 tokc
.cstr
= &tokcstr
;
2568 } else if (c
== '<') {
2585 } else if (c
== '>') {
2603 } else if (c
== '=') {
2616 } else if (c
== '=') {
2629 } else if (c
== '=') {
2642 } else if (c
== '=') {
2645 } else if (c
== '>') {
2653 PARSE2('!', '!', '=', TOK_NE
)
2654 PARSE2('=', '=', '=', TOK_EQ
)
2655 PARSE2('*', '*', '=', TOK_A_MUL
)
2656 PARSE2('%', '%', '=', TOK_A_MOD
)
2657 PARSE2('^', '^', '=', TOK_A_XOR
)
2659 /* comments or operator */
2663 p
= parse_comment(p
);
2664 /* comments replaced by a blank */
2666 goto keep_tok_flags
;
2667 } else if (c
== '/') {
2668 p
= parse_line_comment(p
);
2670 goto keep_tok_flags
;
2671 } else if (c
== '=') {
2691 case '@': /* only used in assembler */
2697 tcc_error("unrecognized character \\x%02x", c
);
2703 #if defined(PARSE_DEBUG)
2704 printf("token = %s\n", get_tok_str(tok
, &tokc
));
2708 /* return next token without macro substitution. Can read input from
2710 static void next_nomacro_spc(void)
2716 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2717 if (tok
== TOK_LINENUM
) {
2718 file
->line_num
= tokc
.i
;
2727 ST_FUNC
void next_nomacro(void)
2731 } while (tok
< 256 && (isidnum_table
[tok
- CH_EOF
] & IS_SPC
));
2735 static void macro_subst(
2736 TokenString
*tok_str
,
2738 const int *macro_str
,
2742 /* substitute arguments in replacement lists in macro_str by the values in
2743 args (field d) and return allocated string */
2744 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
2756 TOK_GET(&t
, ¯o_str
, &cval
);
2761 TOK_GET(&t
, ¯o_str
, &cval
);
2764 s
= sym_find2(args
, t
);
2767 cstr_ccat(&cstr
, '\"');
2771 TOK_GET(&t
, &st
, &cval
);
2772 if (t
!= TOK_PLCHLDR
2774 && 0 == check_space(t
, &spc
)) {
2775 const char *s
= get_tok_str(t
, &cval
);
2777 if (t
== TOK_PPSTR
&& *s
!= '\'')
2778 add_char(&cstr
, *s
);
2780 cstr_ccat(&cstr
, *s
);
2786 cstr_ccat(&cstr
, '\"');
2787 cstr_ccat(&cstr
, '\0');
2789 printf("\nstringize: <%s>\n", (char *)cstr
.data
);
2793 tok_str_add2(&str
, TOK_PPSTR
, &cval
);
2794 cstr_free(cval
.cstr
);
2797 expect("macro parameter after '#'");
2799 } else if (t
>= TOK_IDENT
) {
2800 s
= sym_find2(args
, t
);
2804 /* if '##' is present before or after, no arg substitution */
2805 if (*macro_str
== TOK_TWOSHARPS
|| t1
== TOK_TWOSHARPS
) {
2806 /* special case for var arg macros : ## eats the ','
2807 if empty VA_ARGS variable. */
2808 if (t1
== TOK_TWOSHARPS
&& t0
== ',' && gnu_ext
&& s
->type
.t
) {
2810 /* suppress ',' '##' */
2813 /* suppress '##' and add variable */
2820 TOK_GET(&t1
, &st
, &cval
);
2823 tok_str_add2(&str
, t1
, &cval
);
2829 /* NOTE: the stream cannot be read when macro
2830 substituing an argument */
2831 macro_subst(&str
, nested_list
, st
, 0);
2833 if (str
.len
== l0
) /* exanded to empty string */
2834 tok_str_add(&str
, TOK_PLCHLDR
);
2836 tok_str_add(&str
, t
);
2839 tok_str_add2(&str
, t
, &cval
);
2843 tok_str_add(&str
, 0);
2847 static char const ab_month_name
[12][4] =
2849 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
2850 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
2853 /* peek or read [ws_str == NULL] next token from function macro call,
2854 walking up macro levels up to the file if necessary */
2855 static int next_argstream(Sym
**nested_list
, int can_read_stream
, TokenString
*ws_str
)
2863 p
= macro_ptr
, t
= *p
;
2865 while (is_space(t
) || TOK_LINEFEED
== t
)
2866 tok_str_add(ws_str
, t
), t
= *++p
;
2868 if (t
== 0 && can_read_stream
) {
2870 /* also, end of scope for nested defined symbol */
2872 while (sa
&& sa
->v
== -1)
2881 while (is_space(ch
) || ch
== '\n' || ch
== '/') {
2884 uint8_t *p
= file
->buf_ptr
;
2887 p
= parse_comment(p
);
2888 file
->buf_ptr
= p
- 1;
2889 } else if (c
== '/') {
2890 p
= parse_line_comment(p
);
2891 file
->buf_ptr
= p
- 1;
2896 tok_str_add(ws_str
, ch
);
2910 /* do macro substitution of current token with macro 's' and add
2911 result to (tok_str,tok_len). 'nested_list' is the list of all
2912 macros we got inside to avoid recursing. Return non zero if no
2913 substitution needs to be done */
2914 static int macro_subst_tok(
2915 TokenString
*tok_str
,
2918 int can_read_stream
)
2920 Sym
*args
, *sa
, *sa1
;
2921 int parlevel
, *mstr
, t
, t1
, spc
;
2928 /* if symbol is a macro, prepare substitution */
2929 /* special macros */
2930 if (tok
== TOK___LINE__
) {
2931 snprintf(buf
, sizeof(buf
), "%d", file
->line_num
);
2935 } else if (tok
== TOK___FILE__
) {
2936 cstrval
= file
->filename
;
2938 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
2943 tm
= localtime(&ti
);
2944 if (tok
== TOK___DATE__
) {
2945 snprintf(buf
, sizeof(buf
), "%s %2d %d",
2946 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
2948 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
2949 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
2956 cstr_cat(&cstr
, cstrval
);
2957 cstr_ccat(&cstr
, '\0');
2959 tok_str_add2(tok_str
, t1
, &cval
);
2962 int saved_parse_flags
= parse_flags
;
2965 if (s
->type
.t
== MACRO_FUNC
) {
2966 /* whitespace between macro name and argument list */
2968 tok_str_new(&ws_str
);
2971 parse_flags
|= PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
2972 | PARSE_FLAG_ACCEPT_STRAYS
;
2974 /* get next token from argument stream */
2975 t
= next_argstream(nested_list
, can_read_stream
, &ws_str
);
2977 /* not a macro substitution after all, restore the
2978 * macro token plus all whitespace we've read.
2979 * whitespace is intentionally not merged to preserve
2981 parse_flags
= saved_parse_flags
;
2982 tok_str_add(tok_str
, tok
);
2983 if (parse_flags
& PARSE_FLAG_SPACES
) {
2985 for (i
= 0; i
< ws_str
.len
; i
++)
2986 tok_str_add(tok_str
, ws_str
.str
[i
]);
2988 tok_str_free(ws_str
.str
);
2991 tok_str_free(ws_str
.str
);
2993 next_nomacro(); /* eat '(' */
2995 /* argument macro */
2998 /* NOTE: empty args are allowed, except if no args */
3001 next_argstream(nested_list
, can_read_stream
, NULL
);
3002 } while (is_space(tok
) || TOK_LINEFEED
== tok
);
3004 /* handle '()' case */
3005 if (!args
&& !sa
&& tok
== ')')
3008 tcc_error("macro '%s' used with too many args",
3009 get_tok_str(s
->v
, 0));
3012 /* NOTE: non zero sa->t indicates VA_ARGS */
3013 while ((parlevel
> 0 ||
3015 (tok
!= ',' || sa
->type
.t
)))) {
3016 if (tok
== TOK_EOF
|| tok
== 0)
3020 else if (tok
== ')')
3022 if (tok
== TOK_LINEFEED
)
3024 if (!check_space(tok
, &spc
))
3025 tok_str_add2(&str
, tok
, &tokc
);
3026 next_argstream(nested_list
, can_read_stream
, NULL
);
3031 tok_str_add(&str
, 0);
3032 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
3036 /* special case for gcc var args: add an empty
3037 var arg argument if it is omitted */
3038 if (sa
&& sa
->type
.t
&& gnu_ext
)
3046 tcc_error("macro '%s' used with too few args",
3047 get_tok_str(s
->v
, 0));
3050 parse_flags
= saved_parse_flags
;
3052 /* now subst each arg */
3053 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
3058 tok_str_free(sa
->d
);
3064 sym_push2(nested_list
, s
->v
, 0, 0);
3065 parse_flags
= saved_parse_flags
;
3066 macro_subst(tok_str
, nested_list
, mstr
, can_read_stream
);
3068 /* pop nested defined symbol */
3070 *nested_list
= sa1
->prev
;
3078 int paste_tokens(int t1
, CValue
*v1
, int t2
, CValue
*v2
)
3084 if (t1
!= TOK_PLCHLDR
)
3085 cstr_cat(&cstr
, get_tok_str(t1
, v1
));
3087 if (t2
!= TOK_PLCHLDR
)
3088 cstr_cat(&cstr
, get_tok_str(t2
, v2
));
3089 cstr_ccat(&cstr
, '\0');
3091 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3092 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3095 if (0 == *file
->buf_ptr
)
3099 tcc_warning("pasting <%.*s> and <%s> does not give a valid preprocessing token",
3100 n
, cstr
.data
, (char*)cstr
.data
+ n
);
3105 //printf("paste <%s>\n", (char*)cstr.data);
3110 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
3111 return the resulting string (which must be freed). */
3112 static inline int *macro_twosharps(const int *ptr0
)
3116 TokenString macro_str1
;
3117 int start_of_nosubsts
= -1;
3120 /* we search the first '##' */
3121 for (ptr
= ptr0
;;) {
3122 TOK_GET(&t
, &ptr
, &cval
);
3123 if (t
== TOK_TWOSHARPS
)
3129 tok_str_new(¯o_str1
);
3131 //tok_print(" $$$", ptr0);
3132 for (ptr
= ptr0
;;) {
3133 TOK_GET(&t
, &ptr
, &cval
);
3136 if (t
== TOK_TWOSHARPS
)
3138 while (*ptr
== TOK_TWOSHARPS
) {
3140 /* given 'a##b', remove nosubsts preceding 'a' */
3141 if (start_of_nosubsts
>= 0)
3142 macro_str1
.len
= start_of_nosubsts
;
3143 /* given 'a##b', remove nosubsts preceding 'b' */
3144 while ((t1
= *++ptr
) == TOK_NOSUBST
)
3146 if (t1
&& t1
!= TOK_TWOSHARPS
3147 && t1
!= ':') /* 'a##:' don't build a new token */
3149 TOK_GET(&t1
, &ptr
, &cv1
);
3150 if (t
!= TOK_PLCHLDR
|| t1
!= TOK_PLCHLDR
) {
3151 paste_tokens(t
, &cval
, t1
, &cv1
);
3152 t
= tok
, cval
= tokc
;
3156 if (t
== TOK_NOSUBST
) {
3157 if (start_of_nosubsts
< 0)
3158 start_of_nosubsts
= macro_str1
.len
;
3160 start_of_nosubsts
= -1;
3162 tok_str_add2(¯o_str1
, t
, &cval
);
3164 tok_str_add(¯o_str1
, 0);
3165 //tok_print(" ###", macro_str1.str);
3166 return macro_str1
.str
;
3169 /* do macro substitution of macro_str and add result to
3170 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3171 inside to avoid recursing. */
3172 static void macro_subst(
3173 TokenString
*tok_str
,
3175 const int *macro_str
,
3181 int t
, spc
, nosubst
;
3183 int *macro_str1
= NULL
;
3185 /* first scan for '##' operator handling */
3189 /* first scan for '##' operator handling */
3190 if (can_read_stream
) {
3191 macro_str1
= macro_twosharps(ptr
);
3197 TOK_GET(&t
, &ptr
, &cval
);
3201 if (t
>= TOK_IDENT
&& 0 == nosubst
) {
3206 /* if nested substitution, do nothing */
3207 if (sym_find2(*nested_list
, t
)) {
3208 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3209 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3215 str
.str
= (int*)ptr
;
3216 begin_macro(&str
, 2);
3219 macro_subst_tok(tok_str
, nested_list
, s
, can_read_stream
);
3221 if (str
.alloc
== 3) {
3222 /* already finished by reading function macro arguments */
3230 spc
= tok_str
->len
&& is_space(tok_str
->str
[tok_str
->len
-1]);
3234 if (t
== '\\' && !(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
3235 tcc_error("stray '\\' in program");
3238 if (!check_space(t
, &spc
))
3239 tok_str_add2(tok_str
, t
, &cval
);
3241 if (t
== TOK_NOSUBST
)
3246 tok_str_free(macro_str1
);
3250 /* return next token with macro substitution */
3251 ST_FUNC
void next(void)
3254 if (parse_flags
& PARSE_FLAG_SPACES
)
3260 if (tok
== TOK_NOSUBST
|| tok
== TOK_PLCHLDR
) {
3261 /* discard preprocessor markers */
3263 } else if (tok
== 0) {
3264 /* end of macro or unget token string */
3268 } else if (tok
>= TOK_IDENT
&& (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3270 /* if reading from file, try to substitute macros */
3271 s
= define_find(tok
);
3273 static TokenString str
; /* using static string for speed */
3274 Sym
*nested_list
= NULL
;
3277 macro_subst_tok(&str
, &nested_list
, s
, 1);
3278 tok_str_add(&str
, 0);
3279 begin_macro(&str
, 0);
3283 /* convert preprocessor tokens into C tokens */
3284 if (tok
== TOK_PPNUM
) {
3285 if (parse_flags
& PARSE_FLAG_TOK_NUM
)
3286 parse_number((char *)tokc
.cstr
->data
);
3287 } else if (tok
== TOK_PPSTR
) {
3288 if (parse_flags
& PARSE_FLAG_TOK_STR
)
3289 parse_string((char *)tokc
.cstr
->data
, tokc
.cstr
->size
- 1);
3293 /* push back current token and set current token to 'last_tok'. Only
3294 identifier case handled for labels. */
3295 ST_INLN
void unget_tok(int last_tok
)
3297 TokenString
*str
= tcc_malloc(sizeof *str
);
3299 tok_str_add2(str
, tok
, &tokc
);
3300 tok_str_add(str
, 0);
3301 begin_macro(str
, 1);
3305 /* better than nothing, but needs extension to handle '-E' option
3307 ST_FUNC
void preprocess_init(TCCState
*s1
)
3309 s1
->include_stack_ptr
= s1
->include_stack
;
3310 /* XXX: move that before to avoid having to initialize
3311 file->ifdef_stack_ptr ? */
3312 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3313 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3315 pvtop
= vtop
= vstack
- 1;
3316 s1
->pack_stack
[0] = 0;
3317 s1
->pack_stack_ptr
= s1
->pack_stack
;
3319 isidnum_table
['$' - CH_EOF
] =
3320 tcc_state
->dollars_in_identifiers
? IS_ID
: 0;
3323 ST_FUNC
void preprocess_new(void)
3328 /* init isid table */
3329 for(i
= CH_EOF
; i
<256; i
++)
3330 isidnum_table
[i
- CH_EOF
]
3331 = is_space(i
) ? IS_SPC
3336 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3338 tok_ident
= TOK_IDENT
;
3347 tok_alloc(p
, r
- p
- 1);
3352 ST_FUNC
void preprocess_delete(void)
3356 /* free -D and compiler defines */
3359 /* cleanup from error/setjmp */
3365 n
= tok_ident
- TOK_IDENT
;
3366 for(i
= 0; i
< n
; i
++)
3367 tcc_free(table_ident
[i
]);
3368 tcc_free(table_ident
);
3372 /* Preprocess the current file */
3373 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3375 BufferedFile
**iptr
;
3376 int token_seen
, spcs
, level
;
3378 preprocess_init(s1
);
3379 ch
= file
->buf_ptr
[0];
3380 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3381 parse_flags
= PARSE_FLAG_PREPROCESS
3382 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
3383 | PARSE_FLAG_LINEFEED
3385 | PARSE_FLAG_ACCEPT_STRAYS
3389 do next(); while (tok
!= TOK_EOF
); return 0;
3392 token_seen
= spcs
= 0;
3393 pp_line(s1
, file
, 0);
3396 iptr
= s1
->include_stack_ptr
;
3400 level
= s1
->include_stack_ptr
- iptr
;
3403 pp_line(s1
, *iptr
, 0);
3404 pp_line(s1
, file
, level
);
3407 if (0 == token_seen
) {
3412 if (tok
== TOK_LINEFEED
) {
3416 pp_line(s1
, file
, 0);
3418 fputs(" ", s1
->ppfp
), --spcs
;
3421 } else if (tok
== TOK_LINEFEED
) {
3426 fputs(get_tok_str(tok
, &tokc
), s1
->ppfp
);