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_FILE
)
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", 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
));
1091 /* defines handling */
1092 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1097 if (s
&& !macro_is_equal(s
->d
, str
))
1098 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1100 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1102 s
->next
= first_arg
;
1103 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1107 /* undefined a define symbol. Its name is just set to zero */
1108 ST_FUNC
void define_undef(Sym
*s
)
1112 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1114 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1118 ST_INLN Sym
*define_find(int v
)
1121 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1123 return table_ident
[v
]->sym_define
;
1126 /* free define stack until top reaches 'b' */
1127 ST_FUNC
void free_defines(Sym
*b
)
1135 /* do not free args or predefined defines */
1137 tok_str_free(top
->d
);
1139 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1140 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1147 ST_FUNC
void print_defines(void)
1155 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1156 s
= table_ident
[v
- TOK_IDENT
]->sym_define
;
1164 ST_FUNC Sym
*label_find(int v
)
1167 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1169 return table_ident
[v
]->sym_label
;
1172 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1175 s
= sym_push2(ptop
, v
, 0, 0);
1177 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1178 if (ptop
== &global_label_stack
) {
1179 /* modify the top most local identifier, so that
1180 sym_identifier will point to 's' when popped */
1182 ps
= &(*ps
)->prev_tok
;
1189 /* pop labels until element last is reached. Look if any labels are
1190 undefined. Define symbols if '&&label' was used. */
1191 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
)
1194 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1196 if (s
->r
== LABEL_DECLARED
) {
1197 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1198 } else if (s
->r
== LABEL_FORWARD
) {
1199 tcc_error("label '%s' used but not defined",
1200 get_tok_str(s
->v
, NULL
));
1203 /* define corresponding symbol. A size of
1205 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1209 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1215 /* eval an expression for #if/#elif */
1216 static int expr_preprocess(void)
1222 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1223 next(); /* do macro subst */
1224 if (tok
== TOK_DEFINED
) {
1229 c
= define_find(tok
) != 0;
1234 } else if (tok
>= TOK_IDENT
) {
1235 /* if undefined macro */
1239 tok_str_add_tok(&str
);
1241 tok_str_add(&str
, -1); /* simulate end of file */
1242 tok_str_add(&str
, 0);
1243 /* now evaluate C constant expression */
1244 macro_ptr
= str
.str
;
1248 tok_str_free(str
.str
);
1252 /* parse after #define */
1253 ST_FUNC
void parse_define(void)
1255 Sym
*s
, *first
, **ps
;
1256 int v
, t
, varg
, is_vaargs
, spc
, ptok
, macro_list_start
;
1261 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1262 /* XXX: should check if same macro (ANSI) */
1265 /* '(' must be just after macro definition for MACRO_FUNC */
1270 while (tok
!= ')') {
1274 if (varg
== TOK_DOTS
) {
1275 varg
= TOK___VA_ARGS__
;
1277 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1281 if (varg
< TOK_IDENT
)
1282 tcc_error( "\'%s\' may not appear in parameter list", get_tok_str(varg
, NULL
));
1283 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1295 /* EOF testing necessary for '-D' handling */
1297 macro_list_start
= 1;
1298 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1299 if (macro_list_start
&& spc
== 2 && tok
== TOK_TWOSHARPS
)
1300 tcc_error("'##' invalid at start of macro");
1302 /* remove spaces around ## and after '#' */
1303 if (TOK_TWOSHARPS
== tok
) {
1307 } else if ('#' == tok
) {
1309 } else if (check_space(tok
, &spc
)) {
1312 tok_str_add2(&str
, tok
, &tokc
);
1313 macro_list_start
= 0;
1317 if (ptok
== TOK_TWOSHARPS
)
1318 tcc_error("'##' invalid at end of macro");
1320 --str
.len
; /* remove trailing space */
1321 tok_str_add(&str
, 0);
1322 define_push(v
, t
, str
.str
, first
);
1325 static inline int hash_cached_include(const char *filename
)
1327 const unsigned char *s
;
1331 s
= (unsigned char *) filename
;
1333 h
= TOK_HASH_FUNC(h
, *s
);
1336 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1340 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
)
1344 h
= hash_cached_include(filename
);
1345 i
= s1
->cached_includes_hash
[h
];
1349 e
= s1
->cached_includes
[i
- 1];
1350 if (0 == PATHCMP(e
->filename
, filename
))
1357 static inline void add_cached_include(TCCState
*s1
, const char *filename
, int ifndef_macro
)
1362 if (search_cached_include(s1
, filename
))
1365 printf("adding cached '%s' %s\n", filename
, get_tok_str(ifndef_macro
, NULL
));
1367 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1368 strcpy(e
->filename
, filename
);
1369 e
->ifndef_macro
= ifndef_macro
;
1370 dynarray_add((void ***)&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1371 /* add in hash table */
1372 h
= hash_cached_include(filename
);
1373 e
->hash_next
= s1
->cached_includes_hash
[h
];
1374 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1377 static void pragma_parse(TCCState
*s1
)
1380 if (tok
== TOK_pack
) {
1383 #pragma pack(1) // set
1384 #pragma pack() // reset to default
1385 #pragma pack(push,1) // push & set
1386 #pragma pack(pop) // restore previous
1390 if (tok
== TOK_ASM_pop
) {
1392 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1394 tcc_error("out of pack stack");
1396 s1
->pack_stack_ptr
--;
1400 if (tok
== TOK_ASM_push
) {
1402 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1404 s1
->pack_stack_ptr
++;
1407 if (tok
!= TOK_CINT
)
1410 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1414 *s1
->pack_stack_ptr
= val
;
1418 } else if (tok
== TOK_comment
) {
1419 if (s1
->ms_extensions
) {
1422 if (tok
== TOK_lib
) {
1425 if (tok
!= TOK_STR
) {
1426 tcc_error("invalid library specification");
1428 int len
= strlen((char *)tokc
.cstr
->data
);
1429 char *file
= tcc_malloc(len
+ 4); /* filetype, "-l", and \0 at the end */
1430 file
[0] = TCC_FILETYPE_BINARY
;
1433 strcpy(&file
[3],(char *)tokc
.cstr
->data
);
1434 dynarray_add((void ***)&s1
->files
, &s1
->nb_files
, file
);
1439 tcc_warning("unknown specifier '%s' in #pragma comment", get_tok_str(tok
, &tokc
));
1442 tcc_warning("#pragma comment(lib) is ignored");
1444 } else if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1448 if (next_nomacro(), tok
!= '(')
1450 if (next_nomacro(), tok
!= TOK_STR
)
1452 v
= tok_alloc(tokc
.cstr
->data
, tokc
.cstr
->size
- 1)->tok
;
1453 if (next_nomacro(), tok
!= ')')
1455 if (t
== TOK_push_macro
) {
1456 while (NULL
== (s
= define_find(v
)))
1457 define_push(v
, 0, NULL
, NULL
);
1458 s
->type
.ref
= s
; /* set push boundary */
1460 for (s
= define_stack
; s
; s
= s
->prev
)
1461 if (s
->v
== v
&& s
->type
.ref
== s
) {
1467 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1469 tcc_warning("unbalanced #pragma pop_macro");
1471 /* print info when tcc is called with "-E -dD" switches */
1472 if (s1
->dflag
&& s1
->ppfp
) {
1474 int c
= file
->line_num
- file
->line_ref
- 1;
1477 fputs("\n", tcc_state
->ppfp
);
1478 file
->line_ref
= file
->line_num
;
1481 fprintf(s1
->ppfp
, "// #pragma %s_macro(\"%s\")",
1482 (t
== TOK_push_macro
) ? "push" : "pop",
1483 get_tok_str(v
, NULL
));
1485 } else if (tok
== TOK_once
) {
1486 add_cached_include(s1
, file
->filename
, TOK_once
);
1488 tcc_warning("unknown #pragma %s", get_tok_str(tok
, &tokc
));
1492 tcc_error("malformed #pragma directive");
1495 /* is_bof is true if first non space token at beginning of file */
1496 ST_FUNC
void preprocess(int is_bof
)
1498 TCCState
*s1
= tcc_state
;
1499 int i
, c
, n
, saved_parse_flags
;
1503 saved_parse_flags
= parse_flags
;
1504 parse_flags
= PARSE_FLAG_PREPROCESS
| PARSE_FLAG_TOK_NUM
|
1505 PARSE_FLAG_LINEFEED
;
1506 parse_flags
|= (saved_parse_flags
& PARSE_FLAG_ASM_FILE
);
1516 s
= define_find(tok
);
1517 /* undefine symbol by putting an invalid name */
1522 case TOK_INCLUDE_NEXT
:
1523 ch
= file
->buf_ptr
[0];
1524 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1529 } else if (ch
== '\"') {
1534 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1535 if ((q
- buf
) < sizeof(buf
) - 1)
1538 if (handle_stray_noerror() == 0)
1546 /* eat all spaces and comments after include */
1547 /* XXX: slightly incorrect */
1548 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1552 /* computed #include : either we have only strings or
1553 we have anything enclosed in '<>' */
1556 if (tok
== TOK_STR
) {
1557 while (tok
!= TOK_LINEFEED
) {
1558 if (tok
!= TOK_STR
) {
1560 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1562 pstrcat(buf
, sizeof(buf
), (char *)tokc
.cstr
->data
);
1568 while (tok
!= TOK_LINEFEED
) {
1569 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1573 /* check syntax and remove '<>' */
1574 if (len
< 2 || buf
[0] != '<' || buf
[len
- 1] != '>')
1575 goto include_syntax
;
1576 memmove(buf
, buf
+ 1, len
- 2);
1577 buf
[len
- 2] = '\0';
1582 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1583 tcc_error("#include recursion too deep");
1584 /* store current file in stack, but increment stack later below */
1585 *s1
->include_stack_ptr
= file
;
1587 n
= s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1588 for (i
= -2; i
< n
; ++i
) {
1589 char buf1
[sizeof file
->filename
];
1595 /* check absolute include path */
1596 if (!IS_ABSPATH(buf
))
1599 i
= n
; /* force end loop */
1601 } else if (i
== -1) {
1602 /* search in current dir if "header.h" */
1605 path
= file
->filename
;
1606 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1609 /* search in all the include paths */
1610 if (i
< s1
->nb_include_paths
)
1611 path
= s1
->include_paths
[i
];
1613 path
= s1
->sysinclude_paths
[i
- s1
->nb_include_paths
];
1614 pstrcpy(buf1
, sizeof(buf1
), path
);
1615 pstrcat(buf1
, sizeof(buf1
), "/");
1618 pstrcat(buf1
, sizeof(buf1
), buf
);
1620 if (tok
== TOK_INCLUDE_NEXT
)
1621 for (f
= s1
->include_stack_ptr
; f
>= s1
->include_stack
; --f
)
1622 if (0 == PATHCMP((*f
)->filename
, buf1
)) {
1624 printf("%s: #include_next skipping %s\n", file
->filename
, buf1
);
1626 goto include_trynext
;
1629 e
= search_cached_include(s1
, buf1
);
1630 if (e
&& (define_find(e
->ifndef_macro
) || e
->ifndef_macro
== TOK_once
)) {
1631 /* no need to parse the include because the 'ifndef macro'
1634 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1639 if (tcc_open(s1
, buf1
) < 0)
1644 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1646 /* update target deps */
1647 dynarray_add((void ***)&s1
->target_deps
, &s1
->nb_target_deps
,
1649 /* push current file in stack */
1650 ++s1
->include_stack_ptr
;
1651 /* add include file debug info */
1653 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1654 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1655 ch
= file
->buf_ptr
[0];
1658 tcc_error("include file '%s' not found", buf
);
1665 c
= expr_preprocess();
1671 if (tok
< TOK_IDENT
)
1672 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1676 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1678 file
->ifndef_macro
= tok
;
1681 c
= (define_find(tok
) != 0) ^ c
;
1683 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1684 tcc_error("memory full (ifdef)");
1685 *s1
->ifdef_stack_ptr
++ = c
;
1688 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1689 tcc_error("#else without matching #if");
1690 if (s1
->ifdef_stack_ptr
[-1] & 2)
1691 tcc_error("#else after #else");
1692 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1695 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1696 tcc_error("#elif without matching #if");
1697 c
= s1
->ifdef_stack_ptr
[-1];
1699 tcc_error("#elif after #else");
1700 /* last #if/#elif expression was true: we skip */
1703 c
= expr_preprocess();
1704 s1
->ifdef_stack_ptr
[-1] = c
;
1706 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1707 file
->ifndef_macro
= 0;
1717 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1718 tcc_error("#endif without matching #if");
1719 s1
->ifdef_stack_ptr
--;
1720 /* '#ifndef macro' was at the start of file. Now we check if
1721 an '#endif' is exactly at the end of file */
1722 if (file
->ifndef_macro
&&
1723 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1724 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1725 /* need to set to zero to avoid false matches if another
1726 #ifndef at middle of file */
1727 file
->ifndef_macro
= 0;
1728 while (tok
!= TOK_LINEFEED
)
1730 tok_flags
|= TOK_FLAG_ENDIF
;
1736 if (tok
!= TOK_CINT
)
1737 tcc_error("A #line format is wrong");
1739 if (tok
!= TOK_CINT
) {
1740 char *p
= tokc
.cstr
->data
;
1741 tokc
.i
= strtoul(p
, (char **)&p
, 10);
1744 file
->line_num
= tokc
.i
- 1;
1746 if (tok
!= TOK_LINEFEED
) {
1747 if (tok
!= TOK_STR
) {
1748 if ((parse_flags
& PARSE_FLAG_ASM_FILE
) == 0) {
1750 tcc_error("#line format is wrong");
1754 pstrcpy(file
->filename
, sizeof(file
->filename
),
1755 (char *)tokc
.cstr
->data
);
1758 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1763 ch
= file
->buf_ptr
[0];
1766 while (ch
!= '\n' && ch
!= CH_EOF
) {
1767 if ((q
- buf
) < sizeof(buf
) - 1)
1770 if (handle_stray_noerror() == 0)
1777 tcc_error("#error %s", buf
);
1779 tcc_warning("#warning %s", buf
);
1785 if (tok
== TOK_LINEFEED
|| tok
== '!' || tok
== TOK_PPNUM
) {
1786 /* '!' is ignored to allow C scripts. numbers are ignored
1787 to emulate cpp behaviour */
1789 if (!(parse_flags
& PARSE_FLAG_ASM_FILE
))
1790 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
1792 /* this is a gas line comment in an 'S' file. */
1793 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
);
1799 /* ignore other preprocess commands or #! for C scripts */
1800 while (tok
!= TOK_LINEFEED
)
1803 parse_flags
= saved_parse_flags
;
1806 /* evaluate escape codes in a string. */
1807 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
1822 case '0': case '1': case '2': case '3':
1823 case '4': case '5': case '6': case '7':
1824 /* at most three octal digits */
1829 n
= n
* 8 + c
- '0';
1833 n
= n
* 8 + c
- '0';
1838 goto add_char_nonext
;
1846 if (c
>= 'a' && c
<= 'f')
1848 else if (c
>= 'A' && c
<= 'F')
1858 goto add_char_nonext
;
1882 goto invalid_escape
;
1892 if (c
>= '!' && c
<= '~')
1893 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
1895 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
1902 cstr_ccat(outstr
, c
);
1904 cstr_wccat(outstr
, c
);
1906 /* add a trailing '\0' */
1908 cstr_ccat(outstr
, '\0');
1910 cstr_wccat(outstr
, '\0');
1913 /* we use 64 bit numbers */
1916 /* bn = (bn << shift) | or_val */
1917 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
1921 for(i
=0;i
<BN_SIZE
;i
++) {
1923 bn
[i
] = (v
<< shift
) | or_val
;
1924 or_val
= v
>> (32 - shift
);
1928 static void bn_zero(unsigned int *bn
)
1931 for(i
=0;i
<BN_SIZE
;i
++) {
1936 /* parse number in null terminated string 'p' and return it in the
1938 static void parse_number(const char *p
)
1940 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
1942 unsigned int bn
[BN_SIZE
];
1953 goto float_frac_parse
;
1954 } else if (t
== '0') {
1955 if (ch
== 'x' || ch
== 'X') {
1959 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
1965 /* parse all digits. cannot check octal numbers at this stage
1966 because of floating point constants */
1968 if (ch
>= 'a' && ch
<= 'f')
1970 else if (ch
>= 'A' && ch
<= 'F')
1978 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
1980 tcc_error("number too long");
1986 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
1987 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
1989 /* NOTE: strtox should support that for hexa numbers, but
1990 non ISOC99 libcs do not support it, so we prefer to do
1992 /* hexadecimal or binary floats */
1993 /* XXX: handle overflows */
2005 } else if (t
>= 'a') {
2007 } else if (t
>= 'A') {
2012 bn_lshift(bn
, shift
, t
);
2019 if (t
>= 'a' && t
<= 'f') {
2021 } else if (t
>= 'A' && t
<= 'F') {
2023 } else if (t
>= '0' && t
<= '9') {
2029 tcc_error("invalid digit");
2030 bn_lshift(bn
, shift
, t
);
2035 if (ch
!= 'p' && ch
!= 'P')
2042 } else if (ch
== '-') {
2046 if (ch
< '0' || ch
> '9')
2047 expect("exponent digits");
2048 while (ch
>= '0' && ch
<= '9') {
2049 exp_val
= exp_val
* 10 + ch
- '0';
2052 exp_val
= exp_val
* s
;
2054 /* now we can generate the number */
2055 /* XXX: should patch directly float number */
2056 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2057 d
= ldexp(d
, exp_val
- frac_bits
);
2062 /* float : should handle overflow */
2064 } else if (t
== 'L') {
2066 #ifdef TCC_TARGET_PE
2071 /* XXX: not large enough */
2072 tokc
.ld
= (long double)d
;
2079 /* decimal floats */
2081 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2086 while (ch
>= '0' && ch
<= '9') {
2087 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2093 if (ch
== 'e' || ch
== 'E') {
2094 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2098 if (ch
== '-' || ch
== '+') {
2099 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2104 if (ch
< '0' || ch
> '9')
2105 expect("exponent digits");
2106 while (ch
>= '0' && ch
<= '9') {
2107 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2119 tokc
.f
= strtof(token_buf
, NULL
);
2120 } else if (t
== 'L') {
2122 #ifdef TCC_TARGET_PE
2124 tokc
.d
= strtod(token_buf
, NULL
);
2127 tokc
.ld
= strtold(token_buf
, NULL
);
2131 tokc
.d
= strtod(token_buf
, NULL
);
2135 unsigned long long n
, n1
;
2136 int lcount
, ucount
, must_64bit
;
2139 /* integer number */
2142 if (b
== 10 && *q
== '0') {
2149 /* no need for checks except for base 10 / 8 errors */
2159 tcc_error("invalid digit");
2162 /* detect overflow */
2163 /* XXX: this test is not reliable */
2165 tcc_error("integer constant overflow");
2168 /* Determine the characteristics (unsigned and/or 64bit) the type of
2169 the constant must have according to the constant suffix(es) */
2170 lcount
= ucount
= must_64bit
= 0;
2176 tcc_error("three 'l's in integer constant");
2177 if (lcount
&& *(p
- 1) != ch
)
2178 tcc_error("incorrect integer suffix: %s", p1
);
2180 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
2185 } else if (t
== 'U') {
2187 tcc_error("two 'u's in integer constant");
2195 /* Whether 64 bits are needed to hold the constant's value */
2196 if (n
& 0xffffffff00000000LL
|| must_64bit
) {
2204 /* Whether type must be unsigned to hold the constant's value */
2205 if (ucount
|| ((n1
>> 31) && (b
!= 10))) {
2206 if (tok
== TOK_CLLONG
)
2210 /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
2211 } else if (n1
>> 31) {
2212 if (tok
== TOK_CINT
)
2215 tcc_error("integer constant overflow");
2218 if (tok
== TOK_CINT
|| tok
== TOK_CUINT
)
2224 tcc_error("invalid number\n");
2228 #define PARSE2(c1, tok1, c2, tok2) \
2239 /* return next token without macro substitution */
2240 static inline void next_nomacro1(void)
2255 goto keep_tok_flags
;
2262 /* first look if it is in fact an end of buffer */
2263 if (p
>= file
->buf_end
) {
2267 if (p
>= file
->buf_end
)
2280 TCCState
*s1
= tcc_state
;
2281 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2282 && !(tok_flags
& TOK_FLAG_EOF
)) {
2283 tok_flags
|= TOK_FLAG_EOF
;
2285 goto keep_tok_flags
;
2286 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2288 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2289 tcc_error("missing #endif");
2290 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2291 /* no include left : end of file. */
2294 tok_flags
&= ~TOK_FLAG_EOF
;
2295 /* pop include file */
2297 /* test if previous '#endif' was after a #ifdef at
2299 if (tok_flags
& TOK_FLAG_ENDIF
) {
2301 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2303 add_cached_include(s1
, file
->filename
, file
->ifndef_macro_saved
);
2304 tok_flags
&= ~TOK_FLAG_ENDIF
;
2307 /* add end of include file debug info */
2308 if (tcc_state
->do_debug
) {
2309 put_stabd(N_EINCL
, 0, 0);
2311 /* pop include stack */
2313 s1
->include_stack_ptr
--;
2322 tok_flags
|= TOK_FLAG_BOL
;
2325 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2328 goto keep_tok_flags
;
2333 if (is_space(c
) && (parse_flags
& PARSE_FLAG_ASM_FILE
)) {
2334 p
= parse_line_comment(p
);
2338 if ((tok_flags
& TOK_FLAG_BOL
) &&
2339 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2341 preprocess(tok_flags
& TOK_FLAG_BOF
);
2347 tok
= TOK_TWOSHARPS
;
2349 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2350 p
= parse_line_comment(p
- 1);
2359 /* dollar is allowed to start identifiers when not parsing asm */
2361 if (!tcc_state
->dollars_in_identifiers
2362 || (parse_flags
& PARSE_FLAG_ASM_FILE
)) goto parse_simple
;
2364 case 'a': case 'b': case 'c': case 'd':
2365 case 'e': case 'f': case 'g': case 'h':
2366 case 'i': case 'j': case 'k': case 'l':
2367 case 'm': case 'n': case 'o': case 'p':
2368 case 'q': case 'r': case 's': case 't':
2369 case 'u': case 'v': case 'w': case 'x':
2371 case 'A': case 'B': case 'C': case 'D':
2372 case 'E': case 'F': case 'G': case 'H':
2373 case 'I': case 'J': case 'K':
2374 case 'M': case 'N': case 'O': case 'P':
2375 case 'Q': case 'R': case 'S': case 'T':
2376 case 'U': case 'V': case 'W': case 'X':
2382 h
= TOK_HASH_FUNC(h
, c
);
2386 if (!isidnum_table
[c
-CH_EOF
]
2387 && (tcc_state
->dollars_in_identifiers
? (c
!= '$') : 1))
2389 h
= TOK_HASH_FUNC(h
, c
);
2396 /* fast case : no stray found, so we have the full token
2397 and we have already hashed it */
2399 h
&= (TOK_HASH_SIZE
- 1);
2400 pts
= &hash_ident
[h
];
2405 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2407 pts
= &(ts
->hash_next
);
2409 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2413 cstr_reset(&tokcstr
);
2416 cstr_ccat(&tokcstr
, *p1
);
2422 while (isidnum_table
[c
-CH_EOF
]
2423 || (tcc_state
->dollars_in_identifiers
? (c
== '$') : 0)) {
2424 cstr_ccat(&tokcstr
, c
);
2427 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2433 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2435 goto parse_ident_fast
;
2438 if (c
== '\'' || c
== '\"') {
2442 cstr_reset(&tokcstr
);
2443 cstr_ccat(&tokcstr
, 'L');
2444 goto parse_ident_slow
;
2448 case '0': case '1': case '2': case '3':
2449 case '4': case '5': case '6': case '7':
2452 cstr_reset(&tokcstr
);
2453 /* after the first digit, accept digits, alpha, '.' or sign if
2454 prefixed by 'eEpP' */
2458 cstr_ccat(&tokcstr
, c
);
2460 if (!(isnum(c
) || isid(c
) || c
== '.' ||
2461 ((c
== '+' || c
== '-') &&
2462 (t
== 'e' || t
== 'E' || t
== 'p' || t
== 'P'))))
2465 /* We add a trailing '\0' to ease parsing */
2466 cstr_ccat(&tokcstr
, '\0');
2467 tokc
.cstr
= &tokcstr
;
2471 /* special dot handling because it can also start a number */
2474 cstr_reset(&tokcstr
);
2475 cstr_ccat(&tokcstr
, '.');
2477 } else if (c
== '.') {
2480 if ((parse_flags
& PARSE_FLAG_ASM_FILE
) == 0)
2501 /* parse the string */
2503 p
= parse_pp_string(p
, sep
, &str
);
2504 cstr_ccat(&str
, '\0');
2506 /* eval the escape (should be done as TOK_PPNUM) */
2507 cstr_reset(&tokcstr
);
2508 parse_escape_string(&tokcstr
, str
.data
, is_long
);
2513 /* XXX: make it portable */
2517 char_size
= sizeof(nwchar_t
);
2518 if (tokcstr
.size
<= char_size
)
2519 tcc_error("empty character constant");
2520 if (tokcstr
.size
> 2 * char_size
)
2521 tcc_warning("multi-character character constant");
2523 tokc
.i
= *(int8_t *)tokcstr
.data
;
2526 tokc
.i
= *(nwchar_t
*)tokcstr
.data
;
2530 tokc
.cstr
= &tokcstr
;
2544 } else if (c
== '<') {
2562 } else if (c
== '>') {
2580 } else if (c
== '=') {
2593 } else if (c
== '=') {
2606 } else if (c
== '=') {
2619 } else if (c
== '=') {
2622 } else if (c
== '>') {
2630 PARSE2('!', '!', '=', TOK_NE
)
2631 PARSE2('=', '=', '=', TOK_EQ
)
2632 PARSE2('*', '*', '=', TOK_A_MUL
)
2633 PARSE2('%', '%', '=', TOK_A_MOD
)
2634 PARSE2('^', '^', '=', TOK_A_XOR
)
2636 /* comments or operator */
2640 p
= parse_comment(p
);
2641 /* comments replaced by a blank */
2643 goto keep_tok_flags
;
2644 } else if (c
== '/') {
2645 p
= parse_line_comment(p
);
2647 goto keep_tok_flags
;
2648 } else if (c
== '=') {
2668 case '@': /* only used in assembler */
2674 if ((parse_flags
& PARSE_FLAG_ASM_FILE
) == 0)
2675 tcc_error("unrecognized character \\x%02x", c
);
2685 #if defined(PARSE_DEBUG)
2686 printf("token = %s\n", get_tok_str(tok
, &tokc
));
2690 /* return next token without macro substitution. Can read input from
2692 static void next_nomacro_spc(void)
2698 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2699 if (tok
== TOK_LINENUM
) {
2700 file
->line_num
= tokc
.i
;
2709 ST_FUNC
void next_nomacro(void)
2713 } while (is_space(tok
));
2716 /* substitute arguments in replacement lists in macro_str by the values in
2717 args (field d) and return allocated string */
2718 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
2720 int last_tok
, t
, spc
;
2730 TOK_GET(&t
, ¯o_str
, &cval
);
2735 TOK_GET(&t
, ¯o_str
, &cval
);
2738 s
= sym_find2(args
, t
);
2744 TOK_GET(&t
, &st
, &cval
);
2745 if (t
!= TOK_PLCHLDR
&& !check_space(t
, &spc
))
2746 cstr_cat(&cstr
, get_tok_str(t
, &cval
));
2749 cstr_ccat(&cstr
, '\0');
2751 printf("stringize: %s\n", (char *)cstr
.data
);
2755 tok_str_add2(&str
, TOK_STR
, &cval
);
2758 tok_str_add2(&str
, t
, &cval
);
2760 } else if (t
>= TOK_IDENT
) {
2761 s
= sym_find2(args
, t
);
2764 /* if '##' is present before or after, no arg substitution */
2765 if (*macro_str
== TOK_TWOSHARPS
|| last_tok
== TOK_TWOSHARPS
) {
2766 /* special case for var arg macros : ## eats the
2767 ',' if empty VA_ARGS variable. */
2768 /* XXX: test of the ',' is not 100%
2769 reliable. should fix it to avoid security
2771 if (gnu_ext
&& s
->type
.t
&&
2772 last_tok
== TOK_TWOSHARPS
&&
2773 str
.len
>= 2 && str
.str
[str
.len
- 2] == ',') {
2774 if (*st
== TOK_PLCHLDR
) {
2775 /* suppress ',' '##' */
2778 /* suppress '##' and add variable */
2786 TOK_GET(&t1
, &st
, &cval
);
2789 tok_str_add2(&str
, t1
, &cval
);
2792 } else if (*st
!= TOK_PLCHLDR
) {
2793 /* NOTE: the stream cannot be read when macro
2794 substituing an argument */
2795 macro_subst(&str
, nested_list
, st
, NULL
);
2798 tok_str_add(&str
, t
);
2801 tok_str_add2(&str
, t
, &cval
);
2805 tok_str_add(&str
, 0);
2809 static char const ab_month_name
[12][4] =
2811 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
2812 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
2815 /* do macro substitution of current token with macro 's' and add
2816 result to (tok_str,tok_len). 'nested_list' is the list of all
2817 macros we got inside to avoid recursing. Return non zero if no
2818 substitution needs to be done */
2819 static int macro_subst_tok(TokenString
*tok_str
,
2820 Sym
**nested_list
, Sym
*s
, struct macro_level
**can_read_stream
)
2822 Sym
*args
, *sa
, *sa1
;
2823 int mstr_allocated
, parlevel
, *mstr
, t
, t1
, spc
;
2831 /* if symbol is a macro, prepare substitution */
2832 /* special macros */
2833 if (tok
== TOK___LINE__
) {
2834 snprintf(buf
, sizeof(buf
), "%d", file
->line_num
);
2838 } else if (tok
== TOK___FILE__
) {
2839 cstrval
= file
->filename
;
2841 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
2846 tm
= localtime(&ti
);
2847 if (tok
== TOK___DATE__
) {
2848 snprintf(buf
, sizeof(buf
), "%s %2d %d",
2849 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
2851 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
2852 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
2859 cstr_cat(&cstr
, cstrval
);
2860 cstr_ccat(&cstr
, '\0');
2862 tok_str_add2(tok_str
, t1
, &cval
);
2867 if (s
->type
.t
== MACRO_FUNC
) {
2868 /* NOTE: we do not use next_nomacro to avoid eating the
2869 next token. XXX: find better solution */
2873 while (is_space(t
= *p
) || TOK_LINEFEED
== t
)
2875 if (t
== 0 && can_read_stream
) {
2876 /* end of macro stream: we must look at the token
2877 after in the file */
2878 struct macro_level
*ml
= *can_read_stream
;
2884 *can_read_stream
= ml
-> prev
;
2886 /* also, end of scope for nested defined symbol */
2887 (*nested_list
)->v
= -1;
2891 ch
= tcc_peekc_slow(file
);
2892 while (is_space(ch
) || ch
== '\n' || ch
== '/')
2897 uint8_t *p
= file
->buf_ptr
;
2900 p
= parse_comment(p
);
2901 file
->buf_ptr
= p
- 1;
2902 } else if (c
== '/') {
2903 p
= parse_line_comment(p
);
2904 file
->buf_ptr
= p
- 1;
2912 if (t
!= '(') /* no macro subst */
2915 /* argument macro */
2920 /* NOTE: empty args are allowed, except if no args */
2922 /* handle '()' case */
2923 if (!args
&& !sa
&& tok
== ')')
2926 tcc_error("macro '%s' used with too many args",
2927 get_tok_str(s
->v
, 0));
2930 /* NOTE: non zero sa->t indicates VA_ARGS */
2931 while ((parlevel
> 0 ||
2933 (tok
!= ',' || sa
->type
.t
)))) {
2934 if (tok
== TOK_EOF
|| tok
== 0)
2938 else if (tok
== ')')
2940 if (tok
== TOK_LINEFEED
)
2942 if (!check_space(tok
, &spc
))
2943 tok_str_add2(&str
, tok
, &tokc
);
2947 tok_str_add(&str
, TOK_PLCHLDR
);
2949 tok_str_add(&str
, 0);
2950 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
2954 /* special case for gcc var args: add an empty
2955 var arg argument if it is omitted */
2956 if (sa
&& sa
->type
.t
&& gnu_ext
)
2966 tcc_error("macro '%s' used with too few args",
2967 get_tok_str(s
->v
, 0));
2970 /* now subst each arg */
2971 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
2976 tok_str_free(sa
->d
);
2982 sym_push2(nested_list
, s
->v
, 0, 0);
2983 macro_subst(tok_str
, nested_list
, mstr
, can_read_stream
);
2984 /* pop nested defined symbol */
2986 *nested_list
= sa1
->prev
;
2994 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
2995 return the resulting string (which must be freed). */
2996 static inline int *macro_twosharps(const int *macro_str
)
3000 TokenString macro_str1
;
3002 int n
, start_of_nosubsts
;
3004 /* we search the first '##' */
3005 for(ptr
= macro_str
;;) {
3007 TOK_GET(&t
, &ptr
, &cval
);
3008 if (t
== TOK_TWOSHARPS
)
3010 /* nothing more to do if end of string */
3015 /* we saw '##', so we need more processing to handle it */
3016 start_of_nosubsts
= -1;
3017 tok_str_new(¯o_str1
);
3018 for(ptr
= macro_str
;;) {
3019 TOK_GET(&tok
, &ptr
, &tokc
);
3022 if (tok
== TOK_TWOSHARPS
)
3024 if (tok
== TOK_NOSUBST
&& start_of_nosubsts
< 0)
3025 start_of_nosubsts
= macro_str1
.len
;
3026 while (*ptr
== TOK_TWOSHARPS
) {
3027 /* given 'a##b', remove nosubsts preceding 'a' */
3028 if (start_of_nosubsts
>= 0)
3029 macro_str1
.len
= start_of_nosubsts
;
3030 /* given 'a##b', skip '##' */
3032 /* given 'a##b', remove nosubsts preceding 'b' */
3033 while (t
== TOK_NOSUBST
)
3035 if (t
&& t
!= TOK_TWOSHARPS
) {
3037 TOK_GET(&t
, &ptr
, &cval
);
3038 /* We concatenate the two tokens */
3040 if (tok
!= TOK_PLCHLDR
)
3041 cstr_cat(&cstr
, get_tok_str(tok
, &tokc
));
3043 if (t
!= TOK_PLCHLDR
|| tok
== TOK_PLCHLDR
)
3044 cstr_cat(&cstr
, get_tok_str(t
, &cval
));
3045 cstr_ccat(&cstr
, '\0');
3047 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3048 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3051 if (0 == *file
->buf_ptr
)
3053 tok_str_add2(¯o_str1
, tok
, &tokc
);
3054 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid preprocessing token",
3055 n
, cstr
.data
, (char*)cstr
.data
+ n
);
3061 if (tok
!= TOK_NOSUBST
) {
3062 tok_str_add2(¯o_str1
, tok
, &tokc
);
3064 start_of_nosubsts
= -1;
3066 tok_str_add2(¯o_str1
, tok
, &tokc
);
3068 tok_str_add(¯o_str1
, 0);
3069 return macro_str1
.str
;
3073 /* do macro substitution of macro_str and add result to
3074 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3075 inside to avoid recursing. */
3076 static void macro_subst(TokenString
*tok_str
, Sym
**nested_list
,
3077 const int *macro_str
, struct macro_level
** can_read_stream
)
3084 struct macro_level ml
;
3087 /* first scan for '##' operator handling */
3089 macro_str1
= macro_twosharps(ptr
);
3097 /* NOTE: ptr == NULL can only happen if tokens are read from
3098 file stream due to a macro function call */
3101 TOK_GET(&t
, &ptr
, &cval
);
3104 if (t
== TOK_NOSUBST
) {
3105 /* following token has already been subst'd. just copy it on */
3106 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3107 TOK_GET(&t
, &ptr
, &cval
);
3112 /* if nested substitution, do nothing */
3113 if (sym_find2(*nested_list
, t
)) {
3114 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3115 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3119 if (can_read_stream
)
3120 ml
.prev
= *can_read_stream
, *can_read_stream
= &ml
;
3121 macro_ptr
= (int *)ptr
;
3123 ret
= macro_subst_tok(tok_str
, nested_list
, s
, can_read_stream
);
3124 spc
= tok_str
->len
&& is_space(tok_str
->str
[tok_str
->len
-1]);
3125 ptr
= (int *)macro_ptr
;
3127 if (can_read_stream
&& *can_read_stream
== &ml
)
3128 *can_read_stream
= ml
.prev
;
3131 if (parse_flags
& PARSE_FLAG_SPACES
)
3136 tok_str_add(tok_str
, ' ');
3140 if (!check_space(t
, &spc
))
3141 tok_str_add2(tok_str
, t
, &cval
);
3145 tok_str_free(macro_str1
);
3148 /* return next token with macro substitution */
3149 ST_FUNC
void next(void)
3151 Sym
*nested_list
, *s
;
3153 struct macro_level
*ml
;
3156 if (parse_flags
& PARSE_FLAG_SPACES
)
3161 /* if not reading from macro substituted string, then try
3162 to substitute macros */
3163 if (tok
>= TOK_IDENT
&&
3164 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3165 s
= define_find(tok
);
3167 /* we have a macro: we try to substitute */
3171 if (macro_subst_tok(&str
, &nested_list
, s
, &ml
) == 0) {
3172 /* substitution done, NOTE: maybe empty */
3173 tok_str_add(&str
, 0);
3174 macro_ptr
= str
.str
;
3175 macro_ptr_allocated
= str
.str
;
3182 /* end of macro or end of unget buffer */
3183 if (unget_buffer_enabled
) {
3184 macro_ptr
= unget_saved_macro_ptr
;
3185 unget_buffer_enabled
= 0;
3187 /* end of macro string: free it */
3188 tok_str_free(macro_ptr_allocated
);
3189 macro_ptr_allocated
= NULL
;
3193 } else if (tok
== TOK_NOSUBST
) {
3194 /* discard preprocessor's nosubst markers */
3199 /* convert preprocessor tokens into C tokens */
3200 if (tok
== TOK_PPNUM
&&
3201 (parse_flags
& PARSE_FLAG_TOK_NUM
)) {
3202 parse_number((char *)tokc
.cstr
->data
);
3206 /* push back current token and set current token to 'last_tok'. Only
3207 identifier case handled for labels. */
3208 ST_INLN
void unget_tok(int last_tok
)
3212 if (unget_buffer_enabled
)
3214 /* assert(macro_ptr == unget_saved_buffer + 1);
3215 assert(*macro_ptr == 0); */
3219 unget_saved_macro_ptr
= macro_ptr
;
3220 unget_buffer_enabled
= 1;
3222 q
= unget_saved_buffer
;
3225 n
= tok_ext_size(tok
) - 1;
3228 *q
= 0; /* end of token string */
3233 /* better than nothing, but needs extension to handle '-E' option
3235 ST_FUNC
void preprocess_init(TCCState
*s1
)
3237 s1
->include_stack_ptr
= s1
->include_stack
;
3238 /* XXX: move that before to avoid having to initialize
3239 file->ifdef_stack_ptr ? */
3240 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3241 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3244 s1
->pack_stack
[0] = 0;
3245 s1
->pack_stack_ptr
= s1
->pack_stack
;
3248 ST_FUNC
void preprocess_new(void)
3253 /* init isid table */
3255 for(i
=CH_EOF
;i
<256;i
++)
3256 isidnum_table
[i
-CH_EOF
] = isid(i
) || isnum(i
);
3258 /* add all tokens */
3260 tcc_free (table_ident
);
3263 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3265 tok_ident
= TOK_IDENT
;
3274 tok_alloc(p
, r
- p
- 1);
3279 static void line_macro_output(BufferedFile
*f
, const char *s
, TCCState
*s1
)
3281 switch (s1
->Pflag
) {
3282 case LINE_MACRO_OUTPUT_FORMAT_STD
:
3283 /* "tcc -E -P1" case */
3284 fprintf(s1
->ppfp
, "# line %d \"%s\"\n", f
->line_num
, f
->filename
);
3287 case LINE_MACRO_OUTPUT_FORMAT_NONE
:
3288 /* "tcc -E -P" case: don't output a line directive */
3291 case LINE_MACRO_OUTPUT_FORMAT_GCC
:
3293 /* "tcc -E" case: a gcc standard by default */
3294 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
, s
);
3299 /* Preprocess the current file */
3300 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3302 BufferedFile
*file_ref
, **iptr
, **iptr_new
;
3306 preprocess_init(s1
);
3307 ch
= file
->buf_ptr
[0];
3308 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3309 parse_flags
&= PARSE_FLAG_ASM_FILE
;
3310 parse_flags
|= PARSE_FLAG_PREPROCESS
| PARSE_FLAG_LINEFEED
| PARSE_FLAG_SPACES
;
3314 iptr
= s1
->include_stack_ptr
;
3318 if (tok
== TOK_EOF
) {
3320 } else if (file
!= file_ref
) {
3322 line_macro_output(file_ref
, "", s1
);
3324 } else if (tok
== TOK_LINEFEED
) {
3329 } else if (!token_seen
) {
3330 d
= file
->line_num
- file
->line_ref
;
3331 if (file
!= file_ref
|| d
>= 8) {
3334 if (tcc_state
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_GCC
) {
3335 iptr_new
= s1
->include_stack_ptr
;
3336 s
= iptr_new
> iptr
? " 1"
3337 : iptr_new
< iptr
? " 2"
3338 : iptr_new
> s1
->include_stack
? " 3"
3342 line_macro_output(file
, s
, s1
);
3345 fputs("\n", s1
->ppfp
), --d
;
3347 file
->line_ref
= (file_ref
= file
)->line_num
;
3348 token_seen
= tok
!= TOK_LINEFEED
;
3352 fputs(get_tok_str(tok
, &tokc
), s1
->ppfp
);