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 int *macro_ptr_allocated
;
44 static const int *unget_saved_macro_ptr
;
45 static int unget_saved_buffer
[TOK_MAX_SIZE
+ 1];
46 static int unget_buffer_enabled
;
47 static TokenSym
*hash_ident
[TOK_HASH_SIZE
];
48 static char token_buf
[STRING_MAX_SIZE
+ 1];
49 /* true if isid(c) || isnum(c) */
50 static unsigned char isidnum_table
[256-CH_EOF
];
52 static const char tcc_keywords
[] =
53 #define DEF(id, str) str "\0"
58 /* WARNING: the content of this string encodes token numbers */
59 static const unsigned char tok_two_chars
[] =
61 "<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
62 "-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";
83 '.','.', 0xa8, // C++ token ?
84 '#','#', TOK_TWOSHARPS
,
89 struct macro_level
*prev
;
93 static void next_nomacro_spc(void);
94 static void macro_subst(
98 struct macro_level
**can_read_stream
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 /* CString handling */
115 static void cstr_realloc(CString
*cstr
, int new_size
)
120 size
= cstr
->size_allocated
;
122 size
= 8; /* no need to allocate a too small first string */
123 while (size
< new_size
)
125 data
= tcc_realloc(cstr
->data_allocated
, size
);
126 cstr
->data_allocated
= data
;
127 cstr
->size_allocated
= size
;
132 ST_FUNC
void cstr_ccat(CString
*cstr
, int ch
)
135 size
= cstr
->size
+ 1;
136 if (size
> cstr
->size_allocated
)
137 cstr_realloc(cstr
, size
);
138 ((unsigned char *)cstr
->data
)[size
- 1] = ch
;
142 ST_FUNC
void cstr_cat(CString
*cstr
, const char *str
)
154 /* add a wide char */
155 ST_FUNC
void cstr_wccat(CString
*cstr
, int ch
)
158 size
= cstr
->size
+ sizeof(nwchar_t
);
159 if (size
> cstr
->size_allocated
)
160 cstr_realloc(cstr
, size
);
161 *(nwchar_t
*)(((unsigned char *)cstr
->data
) + size
- sizeof(nwchar_t
)) = ch
;
165 ST_FUNC
void cstr_new(CString
*cstr
)
167 memset(cstr
, 0, sizeof(CString
));
170 /* free string and reset it to NULL */
171 ST_FUNC
void cstr_free(CString
*cstr
)
173 tcc_free(cstr
->data_allocated
);
177 /* reset string to empty */
178 ST_FUNC
void cstr_reset(CString
*cstr
)
184 static void add_char(CString
*cstr
, int c
)
186 if (c
== '\'' || c
== '\"' || c
== '\\') {
187 /* XXX: could be more precise if char or string */
188 cstr_ccat(cstr
, '\\');
190 if (c
>= 32 && c
<= 126) {
193 cstr_ccat(cstr
, '\\');
195 cstr_ccat(cstr
, 'n');
197 cstr_ccat(cstr
, '0' + ((c
>> 6) & 7));
198 cstr_ccat(cstr
, '0' + ((c
>> 3) & 7));
199 cstr_ccat(cstr
, '0' + (c
& 7));
204 /* ------------------------------------------------------------------------- */
205 /* allocate a new token */
206 static TokenSym
*tok_alloc_new(TokenSym
**pts
, const char *str
, int len
)
208 TokenSym
*ts
, **ptable
;
211 if (tok_ident
>= SYM_FIRST_ANOM
)
212 tcc_error("memory full (symbols)");
214 /* expand token table if needed */
215 i
= tok_ident
- TOK_IDENT
;
216 if ((i
% TOK_ALLOC_INCR
) == 0) {
217 ptable
= tcc_realloc(table_ident
, (i
+ TOK_ALLOC_INCR
) * sizeof(TokenSym
*));
218 table_ident
= ptable
;
221 ts
= tcc_malloc(sizeof(TokenSym
) + len
);
223 ts
->tok
= tok_ident
++;
224 ts
->sym_define
= NULL
;
225 ts
->sym_label
= NULL
;
226 ts
->sym_struct
= NULL
;
227 ts
->sym_identifier
= NULL
;
229 ts
->hash_next
= NULL
;
230 memcpy(ts
->str
, str
, len
);
236 #define TOK_HASH_INIT 1
237 #define TOK_HASH_FUNC(h, c) ((h) * 263 + (c))
239 /* find a token and add it if not found */
240 ST_FUNC TokenSym
*tok_alloc(const char *str
, int len
)
248 h
= TOK_HASH_FUNC(h
, ((unsigned char *)str
)[i
]);
249 h
&= (TOK_HASH_SIZE
- 1);
251 pts
= &hash_ident
[h
];
256 if (ts
->len
== len
&& !memcmp(ts
->str
, str
, len
))
258 pts
= &(ts
->hash_next
);
260 return tok_alloc_new(pts
, str
, len
);
263 /* XXX: buffer overflow */
264 /* XXX: float tokens */
265 ST_FUNC
char *get_tok_str(int v
, CValue
*cv
)
267 static char buf
[STRING_MAX_SIZE
+ 1];
268 static CString cstr_buf
;
273 /* NOTE: to go faster, we give a fixed buffer for small strings */
274 cstr_reset(&cstr_buf
);
276 cstr_buf
.size_allocated
= sizeof(buf
);
279 /* just an explanation, should never happen:
280 if (v <= TOK_LINENUM && v >= TOK_CINT && cv == NULL)
281 tcc_error("internal error: get_tok_str"); */
286 /* XXX: not quite exact, but only useful for testing */
287 sprintf(p
, "%u", cv
->ui
);
291 /* XXX: not quite exact, but only useful for testing */
293 sprintf(p
, "%u", (unsigned)cv
->ull
);
295 sprintf(p
, "%llu", cv
->ull
);
299 cstr_ccat(&cstr_buf
, 'L');
301 cstr_ccat(&cstr_buf
, '\'');
302 add_char(&cstr_buf
, cv
->i
);
303 cstr_ccat(&cstr_buf
, '\'');
304 cstr_ccat(&cstr_buf
, '\0');
308 len
= cstr
->size
- 1;
310 add_char(&cstr_buf
, ((unsigned char *)cstr
->data
)[i
]);
311 cstr_ccat(&cstr_buf
, '\0');
314 cstr_ccat(&cstr_buf
, 'L');
317 cstr_ccat(&cstr_buf
, '\"');
319 len
= cstr
->size
- 1;
321 add_char(&cstr_buf
, ((unsigned char *)cstr
->data
)[i
]);
323 len
= (cstr
->size
/ sizeof(nwchar_t
)) - 1;
325 add_char(&cstr_buf
, ((nwchar_t
*)cstr
->data
)[i
]);
327 cstr_ccat(&cstr_buf
, '\"');
328 cstr_ccat(&cstr_buf
, '\0');
335 return NULL
; /* should not happen */
337 /* above tokens have value, the ones below don't */
346 return strcpy(p
, "...");
348 return strcpy(p
, "<<=");
350 return strcpy(p
, ">>=");
353 /* search in two bytes table */
354 const unsigned char *q
= tok_two_chars
;
367 } else if (v
< tok_ident
) {
368 return table_ident
[v
- TOK_IDENT
]->str
;
369 } else if (v
>= SYM_FIRST_ANOM
) {
370 /* special name for anonymous symbol */
371 sprintf(p
, "L.%u", v
- SYM_FIRST_ANOM
);
373 /* should never happen */
378 return cstr_buf
.data
;
381 /* fill input buffer and peek next char */
382 static int tcc_peekc_slow(BufferedFile
*bf
)
385 /* only tries to read if really end of buffer */
386 if (bf
->buf_ptr
>= bf
->buf_end
) {
388 #if defined(PARSE_DEBUG)
393 len
= read(bf
->fd
, bf
->buffer
, len
);
400 bf
->buf_ptr
= bf
->buffer
;
401 bf
->buf_end
= bf
->buffer
+ len
;
402 *bf
->buf_end
= CH_EOB
;
404 if (bf
->buf_ptr
< bf
->buf_end
) {
405 return bf
->buf_ptr
[0];
407 bf
->buf_ptr
= bf
->buf_end
;
412 /* return the current character, handling end of block if necessary
414 ST_FUNC
int handle_eob(void)
416 return tcc_peekc_slow(file
);
419 /* read next char from current input file and handle end of input buffer */
420 ST_INLN
void inp(void)
422 ch
= *(++(file
->buf_ptr
));
423 /* end of buffer/file handling */
428 /* handle '\[\r]\n' */
429 static int handle_stray_noerror(void)
436 } else if (ch
== '\r') {
450 static void handle_stray(void)
452 if (handle_stray_noerror())
453 tcc_error("stray '\\' in program");
456 /* skip the stray and handle the \\n case. Output an error if
457 incorrect char after the stray */
458 static int handle_stray1(uint8_t *p
)
462 if (p
>= file
->buf_end
) {
479 /* handle just the EOB case, but not stray */
480 #define PEEKC_EOB(c, p)\
491 /* handle the complicated stray case */
497 c = handle_stray1(p);\
502 /* input with '\[\r]\n' handling. Note that this function cannot
503 handle other characters after '\', so you cannot call it inside
504 strings or comments */
505 ST_FUNC
void minp(void)
513 /* single line C++ comments */
514 static uint8_t *parse_line_comment(uint8_t *p
)
522 if (c
== '\n' || c
== CH_EOF
) {
524 } else if (c
== '\\') {
533 } else if (c
== '\r') {
551 ST_FUNC
uint8_t *parse_comment(uint8_t *p
)
560 if (c
== '\n' || c
== '*' || c
== '\\')
564 if (c
== '\n' || c
== '*' || c
== '\\')
568 /* now we can handle all the cases */
572 } else if (c
== '*') {
578 } else if (c
== '/') {
580 } else if (c
== '\\') {
585 /* skip '\[\r]\n', otherwise just skip the stray */
591 } else if (c
== '\r') {
608 /* stray, eob or eof */
613 tcc_error("unexpected end of file in comment");
614 } else if (c
== '\\') {
626 static inline void skip_spaces(void)
632 static inline int check_space(int t
, int *spc
)
643 /* parse a string without interpreting escapes */
644 static uint8_t *parse_pp_string(uint8_t *p
,
645 int sep
, CString
*str
)
653 } else if (c
== '\\') {
659 /* XXX: indicate line number of start of string */
660 tcc_error("missing terminating %c character", sep
);
661 } else if (c
== '\\') {
662 /* escape : just skip \[\r]\n */
667 } else if (c
== '\r') {
670 expect("'\n' after '\r'");
673 } else if (c
== CH_EOF
) {
674 goto unterminated_string
;
677 cstr_ccat(str
, '\\');
683 } else if (c
== '\n') {
686 } else if (c
== '\r') {
690 cstr_ccat(str
, '\r');
706 /* skip block of text until #else, #elif or #endif. skip also pairs of
708 static void preprocess_skip(void)
710 int a
, start_of_line
, c
, in_warn_or_error
;
717 in_warn_or_error
= 0;
738 } else if (c
== '\\') {
739 ch
= file
->buf_ptr
[0];
740 handle_stray_noerror();
747 if (in_warn_or_error
)
749 p
= parse_pp_string(p
, c
, NULL
);
753 if (in_warn_or_error
)
760 p
= parse_comment(p
);
761 } else if (ch
== '/') {
762 p
= parse_line_comment(p
);
772 (tok
== TOK_ELSE
|| tok
== TOK_ELIF
|| tok
== TOK_ENDIF
))
774 if (tok
== TOK_IF
|| tok
== TOK_IFDEF
|| tok
== TOK_IFNDEF
)
776 else if (tok
== TOK_ENDIF
)
778 else if( tok
== TOK_ERROR
|| tok
== TOK_WARNING
)
779 in_warn_or_error
= 1;
780 else if (tok
== TOK_LINEFEED
)
782 else if (parse_flags
& PARSE_FLAG_ASM_COMMENTS
)
783 p
= parse_line_comment(p
);
785 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
786 p
= parse_line_comment(p
);
799 /* ParseState handling */
801 /* XXX: currently, no include file info is stored. Thus, we cannot display
802 accurate messages if the function or data definition spans multiple
805 /* save current parse state in 's' */
806 ST_FUNC
void save_parse_state(ParseState
*s
)
808 s
->line_num
= file
->line_num
;
809 s
->macro_ptr
= macro_ptr
;
814 /* restore parse state from 's' */
815 ST_FUNC
void restore_parse_state(ParseState
*s
)
817 file
->line_num
= s
->line_num
;
818 macro_ptr
= s
->macro_ptr
;
823 /* return the number of additional 'ints' necessary to store the
825 static inline int tok_ext_size(int t
)
839 tcc_error("unsupported token");
846 return LDOUBLE_SIZE
/ 4;
852 /* token string handling */
854 ST_INLN
void tok_str_new(TokenString
*s
)
858 s
->allocated_len
= 0;
859 s
->last_line_num
= -1;
862 ST_FUNC
void tok_str_free(int *str
)
867 static int *tok_str_realloc(TokenString
*s
)
871 if (s
->allocated_len
== 0) {
874 len
= s
->allocated_len
* 2;
876 str
= tcc_realloc(s
->str
, len
* sizeof(int));
877 s
->allocated_len
= len
;
882 ST_FUNC
void tok_str_add(TokenString
*s
, int t
)
888 if (len
>= s
->allocated_len
)
889 str
= tok_str_realloc(s
);
894 static void tok_str_add2(TokenString
*s
, int t
, CValue
*cv
)
901 /* allocate space for worst case */
902 if (len
+ TOK_MAX_SIZE
> s
->allocated_len
)
903 str
= tok_str_realloc(s
);
912 str
[len
++] = cv
->tab
[0];
921 nb_words
= (sizeof(CString
) + cv
->cstr
->size
+ 3) >> 2;
922 while ((len
+ nb_words
) > s
->allocated_len
)
923 str
= tok_str_realloc(s
);
924 cstr
= (CString
*)(str
+ len
);
926 cstr
->size
= cv
->cstr
->size
;
927 cstr
->data_allocated
= NULL
;
928 cstr
->size_allocated
= cstr
->size
;
929 memcpy((char *)cstr
+ sizeof(CString
),
930 cv
->cstr
->data
, cstr
->size
);
937 #if LDOUBLE_SIZE == 8
940 str
[len
++] = cv
->tab
[0];
941 str
[len
++] = cv
->tab
[1];
943 #if LDOUBLE_SIZE == 12
945 str
[len
++] = cv
->tab
[0];
946 str
[len
++] = cv
->tab
[1];
947 str
[len
++] = cv
->tab
[2];
948 #elif LDOUBLE_SIZE == 16
950 str
[len
++] = cv
->tab
[0];
951 str
[len
++] = cv
->tab
[1];
952 str
[len
++] = cv
->tab
[2];
953 str
[len
++] = cv
->tab
[3];
954 #elif LDOUBLE_SIZE != 8
955 #error add long double size support
964 /* add the current parse token in token string 's' */
965 ST_FUNC
void tok_str_add_tok(TokenString
*s
)
969 /* save line number info */
970 if (file
->line_num
!= s
->last_line_num
) {
971 s
->last_line_num
= file
->line_num
;
972 cval
.i
= s
->last_line_num
;
973 tok_str_add2(s
, TOK_LINENUM
, &cval
);
975 tok_str_add2(s
, tok
, &tokc
);
978 /* get a token from an integer array and increment pointer
979 accordingly. we code it as a macro to avoid pointer aliasing. */
980 static inline void TOK_GET(int *t
, const int **pp
, CValue
*cv
)
998 cv
->cstr
= (CString
*)p
;
999 cv
->cstr
->data
= (char *)p
+ sizeof(CString
);
1000 p
+= (sizeof(CString
) + cv
->cstr
->size
+ 3) >> 2;
1008 #if LDOUBLE_SIZE == 16
1010 #elif LDOUBLE_SIZE == 12
1012 #elif LDOUBLE_SIZE == 8
1015 # error add long double size support
1028 static int macro_is_equal(const int *a
, const int *b
)
1030 char buf
[STRING_MAX_SIZE
+ 1];
1034 TOK_GET(&t
, &a
, &cv
);
1035 pstrcpy(buf
, sizeof buf
, get_tok_str(t
, &cv
));
1036 TOK_GET(&t
, &b
, &cv
);
1037 if (strcmp(buf
, get_tok_str(t
, &cv
)))
1043 static void define_print(Sym
*s
, int is_undef
)
1050 if (tcc_state
->dflag
== 0 || !s
|| !tcc_state
->ppfp
)
1054 c
= file
->line_num
- file
->line_ref
- 1;
1057 fputs("\n", tcc_state
->ppfp
);
1058 file
->line_ref
= file
->line_num
;
1063 fprintf(tcc_state
->ppfp
, "// #undef %s\n", get_tok_str(s
->v
, NULL
));
1067 fprintf(tcc_state
->ppfp
, "// #define %s", get_tok_str(s
->v
, NULL
));
1072 fprintf(tcc_state
->ppfp
, "%s%s", sep
, get_tok_str(arg
->v
& ~SYM_FIELD
, NULL
));
1076 fprintf(tcc_state
->ppfp
, ")");
1081 fprintf(tcc_state
->ppfp
, " ");
1084 TOK_GET(&t
, &str
, &cval
);
1087 fprintf(tcc_state
->ppfp
, "%s", get_tok_str(t
, &cval
));
1089 fprintf(tcc_state
->ppfp
, "\n");
1092 /* defines handling */
1093 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1098 if (s
&& !macro_is_equal(s
->d
, str
))
1099 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1101 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1103 s
->next
= first_arg
;
1104 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1108 /* undefined a define symbol. Its name is just set to zero */
1109 ST_FUNC
void define_undef(Sym
*s
)
1113 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1115 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1119 ST_INLN Sym
*define_find(int v
)
1122 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1124 return table_ident
[v
]->sym_define
;
1127 /* free define stack until top reaches 'b' */
1128 ST_FUNC
void free_defines(Sym
*b
)
1136 /* do not free args or predefined defines */
1138 tok_str_free(top
->d
);
1140 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1141 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1148 ST_FUNC
void print_defines(void)
1156 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1157 s
= table_ident
[v
- TOK_IDENT
]->sym_define
;
1165 ST_FUNC Sym
*label_find(int v
)
1168 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1170 return table_ident
[v
]->sym_label
;
1173 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1176 s
= sym_push2(ptop
, v
, 0, 0);
1178 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1179 if (ptop
== &global_label_stack
) {
1180 /* modify the top most local identifier, so that
1181 sym_identifier will point to 's' when popped */
1183 ps
= &(*ps
)->prev_tok
;
1190 /* pop labels until element last is reached. Look if any labels are
1191 undefined. Define symbols if '&&label' was used. */
1192 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
)
1195 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1197 if (s
->r
== LABEL_DECLARED
) {
1198 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1199 } else if (s
->r
== LABEL_FORWARD
) {
1200 tcc_error("label '%s' used but not defined",
1201 get_tok_str(s
->v
, NULL
));
1204 /* define corresponding symbol. A size of
1206 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1210 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1216 /* eval an expression for #if/#elif */
1217 static int expr_preprocess(void)
1223 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1224 next(); /* do macro subst */
1225 if (tok
== TOK_DEFINED
) {
1230 c
= define_find(tok
) != 0;
1235 } else if (tok
>= TOK_IDENT
) {
1236 /* if undefined macro */
1240 tok_str_add_tok(&str
);
1242 tok_str_add(&str
, -1); /* simulate end of file */
1243 tok_str_add(&str
, 0);
1244 /* now evaluate C constant expression */
1245 macro_ptr
= str
.str
;
1249 tok_str_free(str
.str
);
1253 /* parse after #define */
1254 ST_FUNC
void parse_define(void)
1256 Sym
*s
, *first
, **ps
;
1257 int v
, t
, varg
, is_vaargs
, spc
, ptok
, macro_list_start
;
1262 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1263 /* XXX: should check if same macro (ANSI) */
1266 /* '(' must be just after macro definition for MACRO_FUNC */
1271 while (tok
!= ')') {
1275 if (varg
== TOK_DOTS
) {
1276 varg
= TOK___VA_ARGS__
;
1278 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1282 if (varg
< TOK_IDENT
)
1283 tcc_error( "\'%s\' may not appear in parameter list", get_tok_str(varg
, NULL
));
1284 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1296 /* EOF testing necessary for '-D' handling */
1298 macro_list_start
= 1;
1299 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1300 if (!macro_list_start
&& spc
== 2 && tok
== TOK_TWOSHARPS
)
1301 tcc_error("'##' invalid at start of macro");
1303 /* remove spaces around ## and after '#' */
1304 if (TOK_TWOSHARPS
== tok
) {
1308 } else if ('#' == tok
) {
1310 } else if (check_space(tok
, &spc
)) {
1313 tok_str_add2(&str
, tok
, &tokc
);
1316 macro_list_start
= 0;
1318 if (ptok
== TOK_TWOSHARPS
)
1319 tcc_error("'##' invalid at end of macro");
1321 --str
.len
; /* remove trailing space */
1322 tok_str_add(&str
, 0);
1323 define_push(v
, t
, str
.str
, first
);
1326 static inline int hash_cached_include(const char *filename
)
1328 const unsigned char *s
;
1332 s
= (unsigned char *) filename
;
1334 h
= TOK_HASH_FUNC(h
, *s
);
1337 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1341 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
)
1345 h
= hash_cached_include(filename
);
1346 i
= s1
->cached_includes_hash
[h
];
1350 e
= s1
->cached_includes
[i
- 1];
1351 if (0 == PATHCMP(e
->filename
, filename
))
1358 static inline void add_cached_include(TCCState
*s1
, const char *filename
, int ifndef_macro
)
1363 if (search_cached_include(s1
, filename
))
1366 printf("adding cached '%s' %s\n", filename
, get_tok_str(ifndef_macro
, NULL
));
1368 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1369 strcpy(e
->filename
, filename
);
1370 e
->ifndef_macro
= ifndef_macro
;
1371 dynarray_add((void ***)&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1372 /* add in hash table */
1373 h
= hash_cached_include(filename
);
1374 e
->hash_next
= s1
->cached_includes_hash
[h
];
1375 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1378 static void pragma_parse(TCCState
*s1
)
1381 if (tok
== TOK_pack
) {
1384 #pragma pack(1) // set
1385 #pragma pack() // reset to default
1386 #pragma pack(push,1) // push & set
1387 #pragma pack(pop) // restore previous
1391 if (tok
== TOK_ASM_pop
) {
1393 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1395 tcc_error("out of pack stack");
1397 s1
->pack_stack_ptr
--;
1401 if (tok
== TOK_ASM_push
) {
1403 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1405 s1
->pack_stack_ptr
++;
1408 if (tok
!= TOK_CINT
)
1411 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1415 *s1
->pack_stack_ptr
= val
;
1419 } else if (tok
== TOK_comment
) {
1420 if (s1
->ms_extensions
) {
1423 if (tok
== TOK_lib
) {
1426 if (tok
!= TOK_STR
) {
1427 tcc_error("invalid library specification");
1429 int len
= strlen((char *)tokc
.cstr
->data
);
1430 char *file
= tcc_malloc(len
+ 4); /* filetype, "-l", and \0 at the end */
1431 file
[0] = TCC_FILETYPE_BINARY
;
1434 strcpy(&file
[3],(char *)tokc
.cstr
->data
);
1435 dynarray_add((void ***)&s1
->files
, &s1
->nb_files
, file
);
1440 tcc_warning("unknown specifier '%s' in #pragma comment", get_tok_str(tok
, &tokc
));
1443 tcc_warning("#pragma comment(lib) is ignored");
1445 } else if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1449 if (next_nomacro(), tok
!= '(')
1451 if (next_nomacro(), tok
!= TOK_STR
)
1453 v
= tok_alloc(tokc
.cstr
->data
, tokc
.cstr
->size
- 1)->tok
;
1454 if (next_nomacro(), tok
!= ')')
1456 if (t
== TOK_push_macro
) {
1457 while (NULL
== (s
= define_find(v
)))
1458 define_push(v
, 0, NULL
, NULL
);
1459 s
->type
.ref
= s
; /* set push boundary */
1461 for (s
= define_stack
; s
; s
= s
->prev
)
1462 if (s
->v
== v
&& s
->type
.ref
== s
) {
1468 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1470 tcc_warning("unbalanced #pragma pop_macro");
1471 } else if (tok
== TOK_once
) {
1472 add_cached_include(s1
, file
->filename
, TOK_once
);
1474 tcc_warning("unknown #pragma %s", get_tok_str(tok
, &tokc
));
1478 tcc_error("malformed #pragma directive");
1481 /* is_bof is true if first non space token at beginning of file */
1482 ST_FUNC
void preprocess(int is_bof
)
1484 TCCState
*s1
= tcc_state
;
1485 int i
, c
, n
, saved_parse_flags
;
1489 saved_parse_flags
= parse_flags
;
1490 parse_flags
= PARSE_FLAG_PREPROCESS
| PARSE_FLAG_TOK_NUM
|
1491 PARSE_FLAG_LINEFEED
;
1492 parse_flags
|= (saved_parse_flags
& (PARSE_FLAG_ASM_FILE
| PARSE_FLAG_ASM_COMMENTS
));
1502 s
= define_find(tok
);
1503 /* undefine symbol by putting an invalid name */
1508 case TOK_INCLUDE_NEXT
:
1509 ch
= file
->buf_ptr
[0];
1510 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1515 } else if (ch
== '\"') {
1520 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1521 if ((q
- buf
) < sizeof(buf
) - 1)
1524 if (handle_stray_noerror() == 0)
1532 /* eat all spaces and comments after include */
1533 /* XXX: slightly incorrect */
1534 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1538 /* computed #include : either we have only strings or
1539 we have anything enclosed in '<>' */
1542 if (tok
== TOK_STR
) {
1543 while (tok
!= TOK_LINEFEED
) {
1544 if (tok
!= TOK_STR
) {
1546 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1548 pstrcat(buf
, sizeof(buf
), (char *)tokc
.cstr
->data
);
1554 while (tok
!= TOK_LINEFEED
) {
1555 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1559 /* check syntax and remove '<>' */
1560 if (len
< 2 || buf
[0] != '<' || buf
[len
- 1] != '>')
1561 goto include_syntax
;
1562 memmove(buf
, buf
+ 1, len
- 2);
1563 buf
[len
- 2] = '\0';
1568 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1569 tcc_error("#include recursion too deep");
1570 /* store current file in stack, but increment stack later below */
1571 *s1
->include_stack_ptr
= file
;
1573 n
= s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1574 for (i
= -2; i
< n
; ++i
) {
1575 char buf1
[sizeof file
->filename
];
1581 /* check absolute include path */
1582 if (!IS_ABSPATH(buf
))
1585 i
= n
; /* force end loop */
1587 } else if (i
== -1) {
1588 /* search in current dir if "header.h" */
1591 path
= file
->filename
;
1592 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1595 /* search in all the include paths */
1596 if (i
< s1
->nb_include_paths
)
1597 path
= s1
->include_paths
[i
];
1599 path
= s1
->sysinclude_paths
[i
- s1
->nb_include_paths
];
1600 pstrcpy(buf1
, sizeof(buf1
), path
);
1601 pstrcat(buf1
, sizeof(buf1
), "/");
1604 pstrcat(buf1
, sizeof(buf1
), buf
);
1606 if (tok
== TOK_INCLUDE_NEXT
)
1607 for (f
= s1
->include_stack_ptr
; f
>= s1
->include_stack
; --f
)
1608 if (0 == PATHCMP((*f
)->filename
, buf1
)) {
1610 printf("%s: #include_next skipping %s\n", file
->filename
, buf1
);
1612 goto include_trynext
;
1615 e
= search_cached_include(s1
, buf1
);
1616 if (e
&& (define_find(e
->ifndef_macro
) || e
->ifndef_macro
== TOK_once
)) {
1617 /* no need to parse the include because the 'ifndef macro'
1620 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1625 if (tcc_open(s1
, buf1
) < 0)
1630 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1632 /* update target deps */
1633 dynarray_add((void ***)&s1
->target_deps
, &s1
->nb_target_deps
,
1635 /* push current file in stack */
1636 ++s1
->include_stack_ptr
;
1637 /* add include file debug info */
1639 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1640 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1641 ch
= file
->buf_ptr
[0];
1644 tcc_error("include file '%s' not found", buf
);
1651 c
= expr_preprocess();
1657 if (tok
< TOK_IDENT
)
1658 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1662 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1664 file
->ifndef_macro
= tok
;
1667 c
= (define_find(tok
) != 0) ^ c
;
1669 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1670 tcc_error("memory full (ifdef)");
1671 *s1
->ifdef_stack_ptr
++ = c
;
1674 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1675 tcc_error("#else without matching #if");
1676 if (s1
->ifdef_stack_ptr
[-1] & 2)
1677 tcc_error("#else after #else");
1678 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1681 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1682 tcc_error("#elif without matching #if");
1683 c
= s1
->ifdef_stack_ptr
[-1];
1685 tcc_error("#elif after #else");
1686 /* last #if/#elif expression was true: we skip */
1689 c
= expr_preprocess();
1690 s1
->ifdef_stack_ptr
[-1] = c
;
1692 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1693 file
->ifndef_macro
= 0;
1703 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1704 tcc_error("#endif without matching #if");
1705 s1
->ifdef_stack_ptr
--;
1706 /* '#ifndef macro' was at the start of file. Now we check if
1707 an '#endif' is exactly at the end of file */
1708 if (file
->ifndef_macro
&&
1709 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1710 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1711 /* need to set to zero to avoid false matches if another
1712 #ifndef at middle of file */
1713 file
->ifndef_macro
= 0;
1714 while (tok
!= TOK_LINEFEED
)
1716 tok_flags
|= TOK_FLAG_ENDIF
;
1722 if (tok
!= TOK_CINT
)
1723 tcc_error("A #line format is wrong");
1725 if (tok
!= TOK_CINT
) {
1726 char *p
= tokc
.cstr
->data
;
1727 tokc
.i
= strtoul(p
, (char **)&p
, 10);
1730 file
->line_num
= tokc
.i
- 1;
1732 if (tok
!= TOK_LINEFEED
) {
1733 if (tok
!= TOK_STR
) {
1734 if ((parse_flags
& PARSE_FLAG_ASM_COMMENTS
) == 0) {
1736 tcc_error("#line format is wrong");
1740 pstrcpy(file
->filename
, sizeof(file
->filename
),
1741 (char *)tokc
.cstr
->data
);
1744 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1749 ch
= file
->buf_ptr
[0];
1752 while (ch
!= '\n' && ch
!= CH_EOF
) {
1753 if ((q
- buf
) < sizeof(buf
) - 1)
1756 if (handle_stray_noerror() == 0)
1763 tcc_error("#error %s", buf
);
1765 tcc_warning("#warning %s", buf
);
1771 if (tok
== TOK_LINEFEED
|| tok
== '!' || tok
== TOK_PPNUM
) {
1772 /* '!' is ignored to allow C scripts. numbers are ignored
1773 to emulate cpp behaviour */
1775 if (!(parse_flags
& PARSE_FLAG_ASM_COMMENTS
))
1776 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
1778 /* this is a gas line comment in an 'S' file. */
1779 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
);
1785 /* ignore other preprocess commands or #! for C scripts */
1786 while (tok
!= TOK_LINEFEED
)
1789 parse_flags
= saved_parse_flags
;
1792 /* evaluate escape codes in a string. */
1793 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
1808 case '0': case '1': case '2': case '3':
1809 case '4': case '5': case '6': case '7':
1810 /* at most three octal digits */
1815 n
= n
* 8 + c
- '0';
1819 n
= n
* 8 + c
- '0';
1824 goto add_char_nonext
;
1832 if (c
>= 'a' && c
<= 'f')
1834 else if (c
>= 'A' && c
<= 'F')
1844 goto add_char_nonext
;
1868 goto invalid_escape
;
1878 if (c
>= '!' && c
<= '~')
1879 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
1881 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
1888 cstr_ccat(outstr
, c
);
1890 cstr_wccat(outstr
, c
);
1892 /* add a trailing '\0' */
1894 cstr_ccat(outstr
, '\0');
1896 cstr_wccat(outstr
, '\0');
1899 /* we use 64 bit numbers */
1902 /* bn = (bn << shift) | or_val */
1903 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
1907 for(i
=0;i
<BN_SIZE
;i
++) {
1909 bn
[i
] = (v
<< shift
) | or_val
;
1910 or_val
= v
>> (32 - shift
);
1914 static void bn_zero(unsigned int *bn
)
1917 for(i
=0;i
<BN_SIZE
;i
++) {
1922 /* parse number in null terminated string 'p' and return it in the
1924 static void parse_number(const char *p
)
1926 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
1928 unsigned int bn
[BN_SIZE
];
1939 goto float_frac_parse
;
1940 } else if (t
== '0') {
1941 if (ch
== 'x' || ch
== 'X') {
1945 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
1951 /* parse all digits. cannot check octal numbers at this stage
1952 because of floating point constants */
1954 if (ch
>= 'a' && ch
<= 'f')
1956 else if (ch
>= 'A' && ch
<= 'F')
1964 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
1966 tcc_error("number too long");
1972 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
1973 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
1975 /* NOTE: strtox should support that for hexa numbers, but
1976 non ISOC99 libcs do not support it, so we prefer to do
1978 /* hexadecimal or binary floats */
1979 /* XXX: handle overflows */
1991 } else if (t
>= 'a') {
1993 } else if (t
>= 'A') {
1998 bn_lshift(bn
, shift
, t
);
2005 if (t
>= 'a' && t
<= 'f') {
2007 } else if (t
>= 'A' && t
<= 'F') {
2009 } else if (t
>= '0' && t
<= '9') {
2015 tcc_error("invalid digit");
2016 bn_lshift(bn
, shift
, t
);
2021 if (ch
!= 'p' && ch
!= 'P')
2028 } else if (ch
== '-') {
2032 if (ch
< '0' || ch
> '9')
2033 expect("exponent digits");
2034 while (ch
>= '0' && ch
<= '9') {
2035 exp_val
= exp_val
* 10 + ch
- '0';
2038 exp_val
= exp_val
* s
;
2040 /* now we can generate the number */
2041 /* XXX: should patch directly float number */
2042 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2043 d
= ldexp(d
, exp_val
- frac_bits
);
2048 /* float : should handle overflow */
2050 } else if (t
== 'L') {
2052 #ifdef TCC_TARGET_PE
2057 /* XXX: not large enough */
2058 tokc
.ld
= (long double)d
;
2065 /* decimal floats */
2067 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2072 while (ch
>= '0' && ch
<= '9') {
2073 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2079 if (ch
== 'e' || ch
== 'E') {
2080 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2084 if (ch
== '-' || ch
== '+') {
2085 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2090 if (ch
< '0' || ch
> '9')
2091 expect("exponent digits");
2092 while (ch
>= '0' && ch
<= '9') {
2093 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2105 tokc
.f
= strtof(token_buf
, NULL
);
2106 } else if (t
== 'L') {
2108 #ifdef TCC_TARGET_PE
2110 tokc
.d
= strtod(token_buf
, NULL
);
2113 tokc
.ld
= strtold(token_buf
, NULL
);
2117 tokc
.d
= strtod(token_buf
, NULL
);
2121 unsigned long long n
, n1
;
2122 int lcount
, ucount
, must_64bit
;
2125 /* integer number */
2128 if (b
== 10 && *q
== '0') {
2135 /* no need for checks except for base 10 / 8 errors */
2145 tcc_error("invalid digit");
2148 /* detect overflow */
2149 /* XXX: this test is not reliable */
2151 tcc_error("integer constant overflow");
2154 /* Determine the characteristics (unsigned and/or 64bit) the type of
2155 the constant must have according to the constant suffix(es) */
2156 lcount
= ucount
= must_64bit
= 0;
2162 tcc_error("three 'l's in integer constant");
2163 if (lcount
&& *(p
- 1) != ch
)
2164 tcc_error("incorrect integer suffix: %s", p1
);
2166 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
2171 } else if (t
== 'U') {
2173 tcc_error("two 'u's in integer constant");
2181 /* Whether 64 bits are needed to hold the constant's value */
2182 if (n
& 0xffffffff00000000LL
|| must_64bit
) {
2190 /* Whether type must be unsigned to hold the constant's value */
2191 if (ucount
|| ((n1
>> 31) && (b
!= 10))) {
2192 if (tok
== TOK_CLLONG
)
2196 /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
2197 } else if (n1
>> 31) {
2198 if (tok
== TOK_CINT
)
2201 tcc_error("integer constant overflow");
2204 if (tok
== TOK_CINT
|| tok
== TOK_CUINT
)
2210 tcc_error("invalid number\n");
2214 #define PARSE2(c1, tok1, c2, tok2) \
2225 /* return next token without macro substitution */
2226 static inline void next_nomacro1(void)
2241 goto keep_tok_flags
;
2248 /* first look if it is in fact an end of buffer */
2249 if (p
>= file
->buf_end
) {
2253 if (p
>= file
->buf_end
)
2266 TCCState
*s1
= tcc_state
;
2267 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2268 && !(tok_flags
& TOK_FLAG_EOF
)) {
2269 tok_flags
|= TOK_FLAG_EOF
;
2271 goto keep_tok_flags
;
2272 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2274 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2275 tcc_error("missing #endif");
2276 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2277 /* no include left : end of file. */
2280 tok_flags
&= ~TOK_FLAG_EOF
;
2281 /* pop include file */
2283 /* test if previous '#endif' was after a #ifdef at
2285 if (tok_flags
& TOK_FLAG_ENDIF
) {
2287 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2289 add_cached_include(s1
, file
->filename
, file
->ifndef_macro_saved
);
2290 tok_flags
&= ~TOK_FLAG_ENDIF
;
2293 /* add end of include file debug info */
2294 if (tcc_state
->do_debug
) {
2295 put_stabd(N_EINCL
, 0, 0);
2297 /* pop include stack */
2299 s1
->include_stack_ptr
--;
2308 tok_flags
|= TOK_FLAG_BOL
;
2311 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2314 goto keep_tok_flags
;
2319 if (is_space(c
) && (parse_flags
& PARSE_FLAG_ASM_FILE
)) {
2320 p
= parse_line_comment(p
);
2324 if ((tok_flags
& TOK_FLAG_BOL
) &&
2325 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2327 preprocess(tok_flags
& TOK_FLAG_BOF
);
2333 tok
= TOK_TWOSHARPS
;
2335 if (parse_flags
& PARSE_FLAG_ASM_COMMENTS
) {
2336 p
= parse_line_comment(p
- 1);
2345 /* dollar is allowed to start identifiers when not parsing asm */
2347 if (!tcc_state
->dollars_in_identifiers
2348 || (parse_flags
& PARSE_FLAG_ASM_FILE
)) goto parse_simple
;
2350 case 'a': case 'b': case 'c': case 'd':
2351 case 'e': case 'f': case 'g': case 'h':
2352 case 'i': case 'j': case 'k': case 'l':
2353 case 'm': case 'n': case 'o': case 'p':
2354 case 'q': case 'r': case 's': case 't':
2355 case 'u': case 'v': case 'w': case 'x':
2357 case 'A': case 'B': case 'C': case 'D':
2358 case 'E': case 'F': case 'G': case 'H':
2359 case 'I': case 'J': case 'K':
2360 case 'M': case 'N': case 'O': case 'P':
2361 case 'Q': case 'R': case 'S': case 'T':
2362 case 'U': case 'V': case 'W': case 'X':
2368 h
= TOK_HASH_FUNC(h
, c
);
2372 if (!isidnum_table
[c
-CH_EOF
]
2373 && (tcc_state
->dollars_in_identifiers
? (c
!= '$') : 1))
2375 h
= TOK_HASH_FUNC(h
, c
);
2382 /* fast case : no stray found, so we have the full token
2383 and we have already hashed it */
2385 h
&= (TOK_HASH_SIZE
- 1);
2386 pts
= &hash_ident
[h
];
2391 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2393 pts
= &(ts
->hash_next
);
2395 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2399 cstr_reset(&tokcstr
);
2402 cstr_ccat(&tokcstr
, *p1
);
2408 while (isidnum_table
[c
-CH_EOF
]
2409 || (tcc_state
->dollars_in_identifiers
? (c
== '$') : 0)) {
2410 cstr_ccat(&tokcstr
, c
);
2413 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2419 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2421 goto parse_ident_fast
;
2424 if (c
== '\'' || c
== '\"') {
2428 cstr_reset(&tokcstr
);
2429 cstr_ccat(&tokcstr
, 'L');
2430 goto parse_ident_slow
;
2434 case '0': case '1': case '2': case '3':
2435 case '4': case '5': case '6': case '7':
2438 cstr_reset(&tokcstr
);
2439 /* after the first digit, accept digits, alpha, '.' or sign if
2440 prefixed by 'eEpP' */
2444 cstr_ccat(&tokcstr
, c
);
2446 if (!(isnum(c
) || isid(c
) || c
== '.' ||
2447 ((c
== '+' || c
== '-') &&
2448 (t
== 'e' || t
== 'E' || t
== 'p' || t
== 'P'))))
2451 /* We add a trailing '\0' to ease parsing */
2452 cstr_ccat(&tokcstr
, '\0');
2453 tokc
.cstr
= &tokcstr
;
2457 /* special dot handling because it can also start a number */
2460 cstr_reset(&tokcstr
);
2461 cstr_ccat(&tokcstr
, '.');
2463 } else if (c
== '.') {
2466 if ((parse_flags
& PARSE_FLAG_ASM_COMMENTS
) == 0)
2487 /* parse the string */
2489 p
= parse_pp_string(p
, sep
, &str
);
2490 cstr_ccat(&str
, '\0');
2492 /* eval the escape (should be done as TOK_PPNUM) */
2493 cstr_reset(&tokcstr
);
2494 parse_escape_string(&tokcstr
, str
.data
, is_long
);
2499 /* XXX: make it portable */
2503 char_size
= sizeof(nwchar_t
);
2504 if (tokcstr
.size
<= char_size
)
2505 tcc_error("empty character constant");
2506 if (tokcstr
.size
> 2 * char_size
)
2507 tcc_warning("multi-character character constant");
2509 tokc
.i
= *(int8_t *)tokcstr
.data
;
2512 tokc
.i
= *(nwchar_t
*)tokcstr
.data
;
2516 tokc
.cstr
= &tokcstr
;
2530 } else if (c
== '<') {
2548 } else if (c
== '>') {
2566 } else if (c
== '=') {
2579 } else if (c
== '=') {
2592 } else if (c
== '=') {
2605 } else if (c
== '=') {
2608 } else if (c
== '>') {
2616 PARSE2('!', '!', '=', TOK_NE
)
2617 PARSE2('=', '=', '=', TOK_EQ
)
2618 PARSE2('*', '*', '=', TOK_A_MUL
)
2619 PARSE2('%', '%', '=', TOK_A_MOD
)
2620 PARSE2('^', '^', '=', TOK_A_XOR
)
2622 /* comments or operator */
2626 p
= parse_comment(p
);
2627 /* comments replaced by a blank */
2629 goto keep_tok_flags
;
2630 } else if (c
== '/') {
2631 p
= parse_line_comment(p
);
2633 goto keep_tok_flags
;
2634 } else if (c
== '=') {
2654 case '@': /* only used in assembler */
2660 if ((parse_flags
& PARSE_FLAG_ASM_FILE
) == 0)
2661 tcc_error("unrecognized character \\x%02x", c
);
2671 #if defined(PARSE_DEBUG)
2672 printf("token = %s\n", get_tok_str(tok
, &tokc
));
2676 /* return next token without macro substitution. Can read input from
2678 static void next_nomacro_spc(void)
2684 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2685 if (tok
== TOK_LINENUM
) {
2686 file
->line_num
= tokc
.i
;
2695 ST_FUNC
void next_nomacro(void)
2699 } while (is_space(tok
));
2702 /* substitute arguments in replacement lists in macro_str by the values in
2703 args (field d) and return allocated string */
2704 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
2706 int last_tok
, t
, spc
;
2716 TOK_GET(&t
, ¯o_str
, &cval
);
2721 TOK_GET(&t
, ¯o_str
, &cval
);
2724 s
= sym_find2(args
, t
);
2730 TOK_GET(&t
, &st
, &cval
);
2731 if (!check_space(t
, &spc
))
2732 cstr_cat(&cstr
, get_tok_str(t
, &cval
));
2735 cstr_ccat(&cstr
, '\0');
2737 printf("stringize: %s\n", (char *)cstr
.data
);
2741 tok_str_add2(&str
, TOK_STR
, &cval
);
2744 tok_str_add2(&str
, t
, &cval
);
2746 } else if (t
>= TOK_IDENT
) {
2747 s
= sym_find2(args
, t
);
2750 /* if '##' is present before or after, no arg substitution */
2751 if (*macro_str
== TOK_TWOSHARPS
|| last_tok
== TOK_TWOSHARPS
) {
2752 /* special case for var arg macros : ## eats the
2753 ',' if empty VA_ARGS variable. */
2754 /* XXX: test of the ',' is not 100%
2755 reliable. should fix it to avoid security
2757 if (gnu_ext
&& s
->type
.t
&&
2758 last_tok
== TOK_TWOSHARPS
&&
2759 str
.len
>= 2 && str
.str
[str
.len
- 2] == ',') {
2760 if (*st
== TOK_PLCHLDR
) {
2761 /* suppress ',' '##' */
2764 /* suppress '##' and add variable */
2772 TOK_GET(&t1
, &st
, &cval
);
2775 tok_str_add2(&str
, t1
, &cval
);
2778 } else if (*st
!= TOK_PLCHLDR
) {
2779 /* NOTE: the stream cannot be read when macro
2780 substituing an argument */
2781 macro_subst(&str
, nested_list
, st
, NULL
);
2784 tok_str_add(&str
, t
);
2787 tok_str_add2(&str
, t
, &cval
);
2791 tok_str_add(&str
, 0);
2795 static char const ab_month_name
[12][4] =
2797 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
2798 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
2801 /* do macro substitution of current token with macro 's' and add
2802 result to (tok_str,tok_len). 'nested_list' is the list of all
2803 macros we got inside to avoid recursing. Return non zero if no
2804 substitution needs to be done */
2805 static int macro_subst_tok(TokenString
*tok_str
,
2806 Sym
**nested_list
, Sym
*s
, struct macro_level
**can_read_stream
)
2808 Sym
*args
, *sa
, *sa1
;
2809 int mstr_allocated
, parlevel
, *mstr
, t
, t1
, spc
;
2817 /* if symbol is a macro, prepare substitution */
2818 /* special macros */
2819 if (tok
== TOK___LINE__
) {
2820 snprintf(buf
, sizeof(buf
), "%d", file
->line_num
);
2824 } else if (tok
== TOK___FILE__
) {
2825 cstrval
= file
->filename
;
2827 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
2832 tm
= localtime(&ti
);
2833 if (tok
== TOK___DATE__
) {
2834 snprintf(buf
, sizeof(buf
), "%s %2d %d",
2835 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
2837 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
2838 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
2845 cstr_cat(&cstr
, cstrval
);
2846 cstr_ccat(&cstr
, '\0');
2848 tok_str_add2(tok_str
, t1
, &cval
);
2853 if (s
->type
.t
== MACRO_FUNC
) {
2854 /* NOTE: we do not use next_nomacro to avoid eating the
2855 next token. XXX: find better solution */
2859 while (is_space(t
= *p
) || TOK_LINEFEED
== t
)
2861 if (t
== 0 && can_read_stream
) {
2862 /* end of macro stream: we must look at the token
2863 after in the file */
2864 struct macro_level
*ml
= *can_read_stream
;
2870 *can_read_stream
= ml
-> prev
;
2872 /* also, end of scope for nested defined symbol */
2873 (*nested_list
)->v
= -1;
2877 ch
= file
->buf_ptr
[0];
2878 while (is_space(ch
) || ch
== '\n' || ch
== '/')
2883 uint8_t *p
= file
->buf_ptr
;
2886 p
= parse_comment(p
);
2887 file
->buf_ptr
= p
- 1;
2888 } else if (c
== '/') {
2889 p
= parse_line_comment(p
);
2890 file
->buf_ptr
= p
- 1;
2898 if (t
!= '(') /* no macro subst */
2901 /* argument macro */
2906 /* NOTE: empty args are allowed, except if no args */
2908 /* handle '()' case */
2909 if (!args
&& !sa
&& tok
== ')')
2912 tcc_error("macro '%s' used with too many args",
2913 get_tok_str(s
->v
, 0));
2916 /* NOTE: non zero sa->t indicates VA_ARGS */
2917 while ((parlevel
> 0 ||
2919 (tok
!= ',' || sa
->type
.t
))) &&
2923 else if (tok
== ')')
2925 if (tok
== TOK_LINEFEED
)
2927 if (!check_space(tok
, &spc
))
2928 tok_str_add2(&str
, tok
, &tokc
);
2932 tok_str_add(&str
, TOK_PLCHLDR
);
2934 tok_str_add(&str
, 0);
2935 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
2939 /* special case for gcc var args: add an empty
2940 var arg argument if it is omitted */
2941 if (sa
&& sa
->type
.t
&& gnu_ext
)
2951 tcc_error("macro '%s' used with too few args",
2952 get_tok_str(s
->v
, 0));
2955 /* now subst each arg */
2956 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
2961 tok_str_free(sa
->d
);
2967 sym_push2(nested_list
, s
->v
, 0, 0);
2968 macro_subst(tok_str
, nested_list
, mstr
, can_read_stream
);
2969 /* pop nested defined symbol */
2971 *nested_list
= sa1
->prev
;
2979 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
2980 return the resulting string (which must be freed). */
2981 static inline int *macro_twosharps(const int *macro_str
)
2985 TokenString macro_str1
;
2987 int n
, start_of_nosubsts
;
2989 /* we search the first '##' */
2990 for(ptr
= macro_str
;;) {
2992 TOK_GET(&t
, &ptr
, &cval
);
2993 if (t
== TOK_TWOSHARPS
)
2995 /* nothing more to do if end of string */
3000 /* we saw '##', so we need more processing to handle it */
3001 start_of_nosubsts
= -1;
3002 tok_str_new(¯o_str1
);
3003 for(ptr
= macro_str
;;) {
3004 TOK_GET(&tok
, &ptr
, &tokc
);
3007 if (tok
== TOK_TWOSHARPS
)
3009 if (tok
== TOK_NOSUBST
&& start_of_nosubsts
< 0)
3010 start_of_nosubsts
= macro_str1
.len
;
3011 while (*ptr
== TOK_TWOSHARPS
) {
3012 /* given 'a##b', remove nosubsts preceding 'a' */
3013 if (start_of_nosubsts
>= 0)
3014 macro_str1
.len
= start_of_nosubsts
;
3015 /* given 'a##b', skip '##' */
3017 /* given 'a##b', remove nosubsts preceding 'b' */
3018 while (t
== TOK_NOSUBST
)
3020 if (t
&& t
!= TOK_TWOSHARPS
) {
3022 TOK_GET(&t
, &ptr
, &cval
);
3023 /* We concatenate the two tokens */
3025 if (tok
!= TOK_PLCHLDR
)
3026 cstr_cat(&cstr
, get_tok_str(tok
, &tokc
));
3028 if (t
!= TOK_PLCHLDR
|| tok
== TOK_PLCHLDR
)
3029 cstr_cat(&cstr
, get_tok_str(t
, &cval
));
3030 cstr_ccat(&cstr
, '\0');
3032 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3033 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3036 if (0 == *file
->buf_ptr
)
3038 tok_str_add2(¯o_str1
, tok
, &tokc
);
3039 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid preprocessing token",
3040 n
, cstr
.data
, (char*)cstr
.data
+ n
);
3046 if (tok
!= TOK_NOSUBST
) {
3047 tok_str_add2(¯o_str1
, tok
, &tokc
);
3049 start_of_nosubsts
= -1;
3051 tok_str_add2(¯o_str1
, tok
, &tokc
);
3053 tok_str_add(¯o_str1
, 0);
3054 return macro_str1
.str
;
3058 /* do macro substitution of macro_str and add result to
3059 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3060 inside to avoid recursing. */
3061 static void macro_subst(TokenString
*tok_str
, Sym
**nested_list
,
3062 const int *macro_str
, struct macro_level
** can_read_stream
)
3069 struct macro_level ml
;
3072 /* first scan for '##' operator handling */
3074 macro_str1
= macro_twosharps(ptr
);
3082 /* NOTE: ptr == NULL can only happen if tokens are read from
3083 file stream due to a macro function call */
3086 TOK_GET(&t
, &ptr
, &cval
);
3089 if (t
== TOK_NOSUBST
) {
3090 /* following token has already been subst'd. just copy it on */
3091 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3092 TOK_GET(&t
, &ptr
, &cval
);
3097 /* if nested substitution, do nothing */
3098 if (sym_find2(*nested_list
, t
)) {
3099 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3100 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3104 if (can_read_stream
)
3105 ml
.prev
= *can_read_stream
, *can_read_stream
= &ml
;
3106 macro_ptr
= (int *)ptr
;
3108 ret
= macro_subst_tok(tok_str
, nested_list
, s
, can_read_stream
);
3109 ptr
= (int *)macro_ptr
;
3111 if (can_read_stream
&& *can_read_stream
== &ml
)
3112 *can_read_stream
= ml
.prev
;
3115 if (parse_flags
& PARSE_FLAG_SPACES
)
3120 tok_str_add(tok_str
, ' ');
3124 if (!check_space(t
, &spc
))
3125 tok_str_add2(tok_str
, t
, &cval
);
3129 tok_str_free(macro_str1
);
3132 /* return next token with macro substitution */
3133 ST_FUNC
void next(void)
3135 Sym
*nested_list
, *s
;
3137 struct macro_level
*ml
;
3140 if (parse_flags
& PARSE_FLAG_SPACES
)
3145 /* if not reading from macro substituted string, then try
3146 to substitute macros */
3147 if (tok
>= TOK_IDENT
&&
3148 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3149 s
= define_find(tok
);
3151 /* we have a macro: we try to substitute */
3155 if (macro_subst_tok(&str
, &nested_list
, s
, &ml
) == 0) {
3156 /* substitution done, NOTE: maybe empty */
3157 tok_str_add(&str
, 0);
3158 macro_ptr
= str
.str
;
3159 macro_ptr_allocated
= str
.str
;
3166 /* end of macro or end of unget buffer */
3167 if (unget_buffer_enabled
) {
3168 macro_ptr
= unget_saved_macro_ptr
;
3169 unget_buffer_enabled
= 0;
3171 /* end of macro string: free it */
3172 tok_str_free(macro_ptr_allocated
);
3173 macro_ptr_allocated
= NULL
;
3177 } else if (tok
== TOK_NOSUBST
) {
3178 /* discard preprocessor's nosubst markers */
3183 /* convert preprocessor tokens into C tokens */
3184 if (tok
== TOK_PPNUM
&&
3185 (parse_flags
& PARSE_FLAG_TOK_NUM
)) {
3186 parse_number((char *)tokc
.cstr
->data
);
3190 /* push back current token and set current token to 'last_tok'. Only
3191 identifier case handled for labels. */
3192 ST_INLN
void unget_tok(int last_tok
)
3196 if (unget_buffer_enabled
)
3198 /* assert(macro_ptr == unget_saved_buffer + 1);
3199 assert(*macro_ptr == 0); */
3203 unget_saved_macro_ptr
= macro_ptr
;
3204 unget_buffer_enabled
= 1;
3206 q
= unget_saved_buffer
;
3209 n
= tok_ext_size(tok
) - 1;
3212 *q
= 0; /* end of token string */
3217 /* better than nothing, but needs extension to handle '-E' option
3219 ST_FUNC
void preprocess_init(TCCState
*s1
)
3221 s1
->include_stack_ptr
= s1
->include_stack
;
3222 /* XXX: move that before to avoid having to initialize
3223 file->ifdef_stack_ptr ? */
3224 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3225 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3228 s1
->pack_stack
[0] = 0;
3229 s1
->pack_stack_ptr
= s1
->pack_stack
;
3232 ST_FUNC
void preprocess_new(void)
3237 /* init isid table */
3239 for(i
=CH_EOF
;i
<256;i
++)
3240 isidnum_table
[i
-CH_EOF
] = isid(i
) || isnum(i
);
3242 /* add all tokens */
3244 tcc_free (table_ident
);
3247 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3249 tok_ident
= TOK_IDENT
;
3258 tok_alloc(p
, r
- p
- 1);
3263 static void line_macro_output(BufferedFile
*f
, const char *s
, TCCState
*s1
)
3265 switch (s1
->Pflag
) {
3266 case LINE_MACRO_OUTPUT_FORMAT_STD
:
3267 /* "tcc -E -P1" case */
3268 fprintf(s1
->ppfp
, "# line %d \"%s\"\n", f
->line_num
, f
->filename
);
3271 case LINE_MACRO_OUTPUT_FORMAT_NONE
:
3272 /* "tcc -E -P" case: don't output a line directive */
3275 case LINE_MACRO_OUTPUT_FORMAT_GCC
:
3277 /* "tcc -E" case: a gcc standard by default */
3278 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
, s
);
3283 /* Preprocess the current file */
3284 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3286 BufferedFile
*file_ref
, **iptr
, **iptr_new
;
3290 preprocess_init(s1
);
3291 ch
= file
->buf_ptr
[0];
3292 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3293 parse_flags
= (parse_flags
& PARSE_FLAG_ASM_FILE
);
3294 parse_flags
|= PARSE_FLAG_ASM_COMMENTS
| PARSE_FLAG_PREPROCESS
|
3295 PARSE_FLAG_LINEFEED
| PARSE_FLAG_SPACES
;
3299 iptr
= s1
->include_stack_ptr
;
3303 if (tok
== TOK_EOF
) {
3305 } else if (file
!= file_ref
) {
3307 line_macro_output(file_ref
, "", s1
);
3309 } else if (tok
== TOK_LINEFEED
) {
3314 } else if (!token_seen
) {
3315 d
= file
->line_num
- file
->line_ref
;
3316 if (file
!= file_ref
|| d
>= 8) {
3319 if (tcc_state
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_GCC
) {
3320 iptr_new
= s1
->include_stack_ptr
;
3321 s
= iptr_new
> iptr
? " 1"
3322 : iptr_new
< iptr
? " 2"
3323 : iptr_new
> s1
->include_stack
? " 3"
3327 line_macro_output(file
, s
, s1
);
3330 fputs("\n", s1
->ppfp
), --d
;
3332 file
->line_ref
= (file_ref
= file
)->line_num
;
3333 token_seen
= tok
!= TOK_LINEFEED
;
3337 fputs(get_tok_str(tok
, &tokc
), s1
->ppfp
);