]>
Commit | Line | Data |
---|---|---|
d45cf215 | 1 | /* Lexical analyzer for C and Objective C. |
517cbe13 JL |
2 | Copyright (C) 1987, 1988, 1989, 1992, 1994, 1995, 1996, 1997 |
3 | 1998, 1999, 2000 Free Software Foundation, Inc. | |
e8bbfc4e RK |
4 | |
5 | This file is part of GNU CC. | |
6 | ||
7 | GNU CC is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
9 | the Free Software Foundation; either version 2, or (at your option) | |
10 | any later version. | |
11 | ||
12 | GNU CC is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with GNU CC; see the file COPYING. If not, write to | |
940d9d63 RK |
19 | the Free Software Foundation, 59 Temple Place - Suite 330, |
20 | Boston, MA 02111-1307, USA. */ | |
e8bbfc4e | 21 | |
e9a25f70 | 22 | #include "config.h" |
670ee920 | 23 | #include "system.h" |
e8bbfc4e | 24 | |
e8bbfc4e | 25 | #include "rtl.h" |
8f17b5c5 | 26 | #include "expr.h" |
e8bbfc4e RK |
27 | #include "tree.h" |
28 | #include "input.h" | |
d6f4ec51 | 29 | #include "output.h" |
e8bbfc4e RK |
30 | #include "c-lex.h" |
31 | #include "c-tree.h" | |
32 | #include "flags.h" | |
0e5921e8 | 33 | #include "timevar.h" |
8b97c5f8 | 34 | #include "cpplib.h" |
3d6f7931 | 35 | #include "c-pragma.h" |
5f6da302 | 36 | #include "toplev.h" |
ab87f8c8 | 37 | #include "intl.h" |
1526a060 | 38 | #include "ggc.h" |
7bdb32b9 | 39 | #include "tm_p.h" |
0e5921e8 | 40 | #include "splay-tree.h" |
ab87f8c8 | 41 | |
ab87f8c8 JL |
42 | /* MULTIBYTE_CHARS support only works for native compilers. |
43 | ??? Ideally what we want is to model widechar support after | |
44 | the current floating point support. */ | |
45 | #ifdef CROSS_COMPILE | |
46 | #undef MULTIBYTE_CHARS | |
47 | #endif | |
e8bbfc4e RK |
48 | |
49 | #ifdef MULTIBYTE_CHARS | |
56f48ce9 | 50 | #include "mbchar.h" |
e8bbfc4e | 51 | #include <locale.h> |
56f48ce9 | 52 | #endif /* MULTIBYTE_CHARS */ |
c5c76735 JL |
53 | #ifndef GET_ENVIRONMENT |
54 | #define GET_ENVIRONMENT(ENV_VALUE,ENV_NAME) ((ENV_VALUE) = getenv (ENV_NAME)) | |
55 | #endif | |
e8bbfc4e | 56 | |
0e5921e8 | 57 | #if USE_CPPLIB |
c8724862 | 58 | extern cpp_reader parse_in; |
e56e519d DB |
59 | #else |
60 | /* Stream for reading from the input file. */ | |
61 | FILE *finput; | |
e3d1fd32 PB |
62 | #endif |
63 | ||
0e5921e8 ZW |
64 | /* Private idea of the line number. See discussion in c_lex(). */ |
65 | static int lex_lineno; | |
66 | ||
67 | /* We may keep statistics about how long which files took to compile. */ | |
68 | static int header_time, body_time; | |
69 | static splay_tree file_info_tree; | |
3ab6dd7c | 70 | |
e8bbfc4e RK |
71 | /* Cause the `yydebug' variable to be defined. */ |
72 | #define YYDEBUG 1 | |
73 | ||
0e5921e8 | 74 | #if !USE_CPPLIB |
505e0385 | 75 | |
0e5921e8 ZW |
76 | struct putback_buffer |
77 | { | |
69f21756 | 78 | unsigned char *buffer; |
505e0385 DB |
79 | int buffer_size; |
80 | int index; | |
81 | }; | |
82 | ||
83 | static struct putback_buffer putback = {NULL, 0, -1}; | |
84 | ||
6e090c76 | 85 | static inline int getch PARAMS ((void)); |
93868d11 | 86 | |
505e0385 DB |
87 | static inline int |
88 | getch () | |
89 | { | |
90 | if (putback.index != -1) | |
91 | { | |
92 | int ch = putback.buffer[putback.index]; | |
93 | --putback.index; | |
94 | return ch; | |
95 | } | |
96 | return getc (finput); | |
97 | } | |
98 | ||
6e090c76 | 99 | static inline void put_back PARAMS ((int)); |
93868d11 | 100 | |
505e0385 DB |
101 | static inline void |
102 | put_back (ch) | |
103 | int ch; | |
104 | { | |
105 | if (ch != EOF) | |
106 | { | |
107 | if (putback.index == putback.buffer_size - 1) | |
108 | { | |
109 | putback.buffer_size += 16; | |
110 | putback.buffer = xrealloc (putback.buffer, putback.buffer_size); | |
111 | } | |
112 | putback.buffer[++putback.index] = ch; | |
113 | } | |
114 | } | |
e3d1fd32 | 115 | |
fbb18613 JM |
116 | int linemode; |
117 | ||
0e5921e8 | 118 | #endif |
e8bbfc4e RK |
119 | |
120 | /* File used for outputting assembler code. */ | |
121 | extern FILE *asm_out_file; | |
122 | ||
12a39b12 JM |
123 | #undef WCHAR_TYPE_SIZE |
124 | #define WCHAR_TYPE_SIZE TYPE_PRECISION (wchar_type_node) | |
e8bbfc4e RK |
125 | |
126 | /* Number of bytes in a wide character. */ | |
127 | #define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT) | |
128 | ||
0e5921e8 | 129 | #if !USE_CPPLIB |
e8bbfc4e | 130 | static int maxtoken; /* Current nominal length of token buffer. */ |
0e5921e8 ZW |
131 | static char *token_buffer; /* Pointer to token buffer. |
132 | Actual allocated length is maxtoken + 2. */ | |
133 | #endif | |
e8bbfc4e | 134 | |
0e5921e8 ZW |
135 | int indent_level; /* Number of { minus number of }. */ |
136 | int pending_lang_change; /* If we need to switch languages - C++ only */ | |
137 | int c_header_level; /* depth in C headers - C++ only */ | |
fbb18613 JM |
138 | |
139 | /* Nonzero tells yylex to ignore \ in string constants. */ | |
140 | static int ignore_escape_flag; | |
e9a25f70 | 141 | |
0e5921e8 ZW |
142 | static const char *readescape PARAMS ((const char *, const char *, |
143 | unsigned int *)); | |
144 | static const char *read_ucs PARAMS ((const char *, const char *, | |
145 | unsigned int *, int)); | |
146 | static void parse_float PARAMS ((PTR)); | |
147 | static tree lex_number PARAMS ((const char *, unsigned int)); | |
148 | static tree lex_string PARAMS ((const char *, unsigned int, int)); | |
149 | static tree lex_charconst PARAMS ((const char *, unsigned int, int)); | |
150 | static void update_header_times PARAMS ((const char *)); | |
151 | static int dump_one_header PARAMS ((splay_tree_node, void *)); | |
e8bbfc4e | 152 | |
0e5921e8 | 153 | #if !USE_CPPLIB |
6e090c76 KG |
154 | static int skip_white_space PARAMS ((int)); |
155 | static char *extend_token_buffer PARAMS ((const char *)); | |
6e090c76 KG |
156 | static void extend_token_buffer_to PARAMS ((int)); |
157 | static int read_line_number PARAMS ((int *)); | |
0e5921e8 ZW |
158 | static void process_directive PARAMS ((void)); |
159 | #else | |
160 | static void cb_ident PARAMS ((cpp_reader *, const unsigned char *, | |
161 | unsigned int)); | |
162 | static void cb_enter_file PARAMS ((cpp_reader *)); | |
163 | static void cb_leave_file PARAMS ((cpp_reader *)); | |
164 | static void cb_rename_file PARAMS ((cpp_reader *)); | |
8b97c5f8 | 165 | static void cb_def_pragma PARAMS ((cpp_reader *)); |
0e5921e8 | 166 | #endif |
e31c7eec | 167 | |
e31c7eec | 168 | \f |
3b304f5b | 169 | const char * |
0e5921e8 | 170 | init_c_lex (filename) |
3b304f5b | 171 | const char *filename; |
e3d1fd32 | 172 | { |
0e5921e8 ZW |
173 | struct c_fileinfo *toplevel; |
174 | ||
175 | /* Set up filename timing. Must happen before cpp_start_read. */ | |
176 | file_info_tree = splay_tree_new ((splay_tree_compare_fn)strcmp, | |
177 | 0, | |
178 | (splay_tree_delete_value_fn)free); | |
179 | toplevel = get_fileinfo ("<top level>"); | |
180 | if (flag_detailed_statistics) | |
181 | { | |
182 | header_time = 0; | |
183 | body_time = get_run_time (); | |
184 | toplevel->time = body_time; | |
185 | } | |
186 | ||
187 | #ifdef MULTIBYTE_CHARS | |
188 | /* Change to the native locale for multibyte conversions. */ | |
189 | setlocale (LC_CTYPE, ""); | |
190 | GET_ENVIRONMENT (literal_codeset, "LANG"); | |
191 | #endif | |
192 | ||
e56e519d DB |
193 | #if !USE_CPPLIB |
194 | /* Open input file. */ | |
195 | if (filename == 0 || !strcmp (filename, "-")) | |
196 | { | |
197 | finput = stdin; | |
198 | filename = "stdin"; | |
199 | } | |
200 | else | |
201 | finput = fopen (filename, "r"); | |
202 | if (finput == 0) | |
203 | pfatal_with_name (filename); | |
204 | ||
205 | #ifdef IO_BUFFER_SIZE | |
206 | setvbuf (finput, (char *) xmalloc (IO_BUFFER_SIZE), _IOFBF, IO_BUFFER_SIZE); | |
207 | #endif | |
b12da25e | 208 | #else /* !USE_CPPLIB */ |
0e5921e8 ZW |
209 | |
210 | parse_in.cb.ident = cb_ident; | |
211 | parse_in.cb.enter_file = cb_enter_file; | |
212 | parse_in.cb.leave_file = cb_leave_file; | |
213 | parse_in.cb.rename_file = cb_rename_file; | |
8b97c5f8 | 214 | parse_in.cb.def_pragma = cb_def_pragma; |
0e5921e8 ZW |
215 | |
216 | /* Make sure parse_in.digraphs matches flag_digraphs. */ | |
217 | CPP_OPTION (&parse_in, digraphs) = flag_digraphs; | |
218 | ||
219 | if (! cpp_start_read (&parse_in, 0 /* no printer */, filename)) | |
e3d1fd32 | 220 | abort (); |
add7091b | 221 | |
b12da25e ZW |
222 | if (filename == 0 || !strcmp (filename, "-")) |
223 | filename = "stdin"; | |
e56e519d | 224 | #endif |
5c60e5c0 | 225 | |
0e5921e8 ZW |
226 | #if !USE_CPPLIB |
227 | maxtoken = 40; | |
228 | token_buffer = (char *) xmalloc (maxtoken + 2); | |
229 | #endif | |
230 | /* Start it at 0, because check_newline is called at the very beginning | |
231 | and will increment it to 1. */ | |
232 | lineno = lex_lineno = 0; | |
b12da25e | 233 | |
5c60e5c0 | 234 | return filename; |
e3d1fd32 PB |
235 | } |
236 | ||
0e5921e8 ZW |
237 | struct c_fileinfo * |
238 | get_fileinfo (name) | |
239 | const char *name; | |
e3d1fd32 | 240 | { |
0e5921e8 ZW |
241 | splay_tree_node n; |
242 | struct c_fileinfo *fi; | |
243 | ||
244 | n = splay_tree_lookup (file_info_tree, (splay_tree_key) name); | |
245 | if (n) | |
246 | return (struct c_fileinfo *) n->value; | |
247 | ||
248 | fi = (struct c_fileinfo *) xmalloc (sizeof (struct c_fileinfo)); | |
249 | fi->time = 0; | |
250 | fi->interface_only = 0; | |
251 | fi->interface_unknown = 1; | |
252 | splay_tree_insert (file_info_tree, (splay_tree_key) name, | |
253 | (splay_tree_value) fi); | |
254 | return fi; | |
e56e519d | 255 | } |
e3d1fd32 | 256 | |
0e5921e8 ZW |
257 | static void |
258 | update_header_times (name) | |
259 | const char *name; | |
e8bbfc4e | 260 | { |
0e5921e8 ZW |
261 | /* Changing files again. This means currently collected time |
262 | is charged against header time, and body time starts back at 0. */ | |
263 | if (flag_detailed_statistics) | |
e8bbfc4e | 264 | { |
0e5921e8 ZW |
265 | int this_time = get_run_time (); |
266 | struct c_fileinfo *file = get_fileinfo (name); | |
267 | header_time += this_time - body_time; | |
268 | file->time += this_time - body_time; | |
269 | body_time = this_time; | |
e8bbfc4e RK |
270 | } |
271 | } | |
272 | ||
0e5921e8 ZW |
273 | static int |
274 | dump_one_header (n, dummy) | |
275 | splay_tree_node n; | |
276 | void *dummy ATTRIBUTE_UNUSED; | |
e8bbfc4e | 277 | { |
0e5921e8 ZW |
278 | print_time ((const char *) n->key, |
279 | ((struct c_fileinfo *) n->value)->time); | |
280 | return 0; | |
e8bbfc4e | 281 | } |
e8bbfc4e RK |
282 | |
283 | void | |
0e5921e8 | 284 | dump_time_statistics () |
e8bbfc4e | 285 | { |
0e5921e8 ZW |
286 | struct c_fileinfo *file = get_fileinfo (input_filename); |
287 | int this_time = get_run_time (); | |
288 | file->time += this_time - body_time; | |
289 | ||
290 | fprintf (stderr, "\n******\n"); | |
291 | print_time ("header files (total)", header_time); | |
292 | print_time ("main file (total)", this_time - body_time); | |
293 | fprintf (stderr, "ratio = %g : 1\n", | |
294 | (double)header_time / (double)(this_time - body_time)); | |
295 | fprintf (stderr, "\n******\n"); | |
296 | ||
297 | splay_tree_foreach (file_info_tree, dump_one_header, 0); | |
e8bbfc4e | 298 | } |
a6124a42 | 299 | |
0e5921e8 | 300 | #if !USE_CPPLIB |
a6124a42 | 301 | |
e8bbfc4e RK |
302 | /* If C is not whitespace, return C. |
303 | Otherwise skip whitespace and return first nonwhite char read. */ | |
304 | ||
305 | static int | |
306 | skip_white_space (c) | |
307 | register int c; | |
308 | { | |
e8bbfc4e RK |
309 | for (;;) |
310 | { | |
311 | switch (c) | |
312 | { | |
0e5921e8 ZW |
313 | /* There is no need to process comments, backslash-newline, |
314 | or \r here. None can occur in the output of cpp. */ | |
e8bbfc4e RK |
315 | |
316 | case '\n': | |
fbb18613 JM |
317 | if (linemode) |
318 | { | |
0e5921e8 | 319 | put_back (c); |
fbb18613 JM |
320 | return EOF; |
321 | } | |
e8bbfc4e RK |
322 | c = check_newline (); |
323 | break; | |
324 | ||
0e5921e8 | 325 | /* Per C99, horizontal whitespace is just these four characters. */ |
e8bbfc4e RK |
326 | case ' ': |
327 | case '\t': | |
328 | case '\f': | |
e8bbfc4e | 329 | case '\v': |
0e5921e8 | 330 | c = getch (); |
0dcd8cee RS |
331 | break; |
332 | ||
e8bbfc4e | 333 | case '\\': |
0e5921e8 ZW |
334 | error ("stray '\\' in program"); |
335 | c = getch (); | |
e8bbfc4e RK |
336 | break; |
337 | ||
338 | default: | |
339 | return (c); | |
340 | } | |
341 | } | |
342 | } | |
343 | ||
fbb18613 | 344 | /* Skips all of the white space at the current location in the input file. */ |
e8bbfc4e RK |
345 | |
346 | void | |
347 | position_after_white_space () | |
348 | { | |
349 | register int c; | |
350 | ||
0e5921e8 | 351 | c = getch (); |
e8bbfc4e | 352 | |
0e5921e8 | 353 | put_back (skip_white_space (c)); |
e8bbfc4e RK |
354 | } |
355 | ||
356 | /* Make the token buffer longer, preserving the data in it. | |
357 | P should point to just beyond the last valid character in the old buffer. | |
358 | The value we return is a pointer to the new buffer | |
359 | at a place corresponding to P. */ | |
360 | ||
fbb18613 JM |
361 | static void |
362 | extend_token_buffer_to (size) | |
363 | int size; | |
364 | { | |
365 | do | |
366 | maxtoken = maxtoken * 2 + 10; | |
367 | while (maxtoken < size); | |
368 | token_buffer = (char *) xrealloc (token_buffer, maxtoken + 2); | |
369 | } | |
370 | ||
e8bbfc4e RK |
371 | static char * |
372 | extend_token_buffer (p) | |
5d5993dd | 373 | const char *p; |
e8bbfc4e RK |
374 | { |
375 | int offset = p - token_buffer; | |
fbb18613 | 376 | extend_token_buffer_to (offset); |
e8bbfc4e RK |
377 | return token_buffer + offset; |
378 | } | |
e3d1fd32 | 379 | \f |
f09db6e0 | 380 | |
fbb18613 JM |
381 | static int |
382 | read_line_number (num) | |
383 | int *num; | |
384 | { | |
0e5921e8 ZW |
385 | tree value; |
386 | enum cpp_ttype token = c_lex (&value); | |
fbb18613 | 387 | |
0e5921e8 | 388 | if (token == CPP_NUMBER && TREE_CODE (value) == INTEGER_CST) |
fbb18613 | 389 | { |
0e5921e8 | 390 | *num = TREE_INT_CST_LOW (value); |
fbb18613 JM |
391 | return 1; |
392 | } | |
393 | else | |
394 | { | |
0e5921e8 | 395 | if (token != CPP_EOF) |
fbb18613 JM |
396 | error ("invalid #-line"); |
397 | return 0; | |
398 | } | |
399 | } | |
0e5921e8 | 400 | |
e8bbfc4e RK |
401 | /* At the beginning of a line, increment the line number |
402 | and process any #-directive on this line. | |
403 | If the line is a #-directive, read the entire line and return a newline. | |
0e5921e8 | 404 | Otherwise, return the line's first non-whitespace character. */ |
e8bbfc4e RK |
405 | |
406 | int | |
407 | check_newline () | |
408 | { | |
409 | register int c; | |
e8bbfc4e | 410 | |
0e5921e8 ZW |
411 | /* Loop till we get a nonblank, non-directive line. */ |
412 | for (;;) | |
e8bbfc4e | 413 | { |
0e5921e8 ZW |
414 | /* Read first nonwhite char on the line. */ |
415 | do | |
416 | c = getch (); | |
417 | while (c == ' ' || c == '\t'); | |
418 | ||
419 | lex_lineno++; | |
420 | if (c == '#') | |
421 | { | |
422 | process_directive (); | |
423 | return '\n'; | |
424 | } | |
fbb18613 | 425 | |
0e5921e8 ZW |
426 | else if (c != '\n') |
427 | break; | |
e8bbfc4e | 428 | } |
0e5921e8 ZW |
429 | return c; |
430 | } | |
e8bbfc4e | 431 | |
0e5921e8 ZW |
432 | static void |
433 | process_directive () | |
434 | { | |
435 | enum cpp_ttype token; | |
436 | tree value; | |
437 | int saw_line; | |
438 | enum { act_none, act_push, act_pop } action; | |
439 | int action_number, l; | |
440 | char *new_file; | |
441 | #ifndef NO_IMPLICIT_EXTERN_C | |
ae54392b | 442 | int entering_c_header = 0; |
0e5921e8 ZW |
443 | #endif |
444 | ||
fbb18613 JM |
445 | /* Don't read beyond this line. */ |
446 | saw_line = 0; | |
447 | linemode = 1; | |
448 | ||
0e5921e8 | 449 | token = c_lex (&value); |
e8bbfc4e | 450 | |
0e5921e8 | 451 | if (token == CPP_NAME) |
e8bbfc4e | 452 | { |
fbb18613 JM |
453 | /* If a letter follows, then if the word here is `line', skip |
454 | it and ignore it; otherwise, ignore the line, with an error | |
455 | if the word isn't `pragma'. */ | |
75cb8865 | 456 | |
0e5921e8 | 457 | const char *name = IDENTIFIER_POINTER (value); |
75cb8865 | 458 | |
fbb18613 JM |
459 | if (!strcmp (name, "pragma")) |
460 | { | |
0e5921e8 | 461 | dispatch_pragma (); |
30acbc3e | 462 | goto skipline; |
e8bbfc4e | 463 | } |
fbb18613 | 464 | else if (!strcmp (name, "define")) |
e8bbfc4e | 465 | { |
0e5921e8 | 466 | debug_define (lex_lineno, GET_DIRECTIVE_LINE ()); |
fbb18613 | 467 | goto skipline; |
e8bbfc4e | 468 | } |
fbb18613 | 469 | else if (!strcmp (name, "undef")) |
e8bbfc4e | 470 | { |
0e5921e8 | 471 | debug_undef (lex_lineno, GET_DIRECTIVE_LINE ()); |
fbb18613 | 472 | goto skipline; |
e8bbfc4e | 473 | } |
fbb18613 | 474 | else if (!strcmp (name, "line")) |
e8bbfc4e | 475 | { |
fbb18613 | 476 | saw_line = 1; |
0e5921e8 | 477 | token = c_lex (&value); |
fbb18613 | 478 | goto linenum; |
e8bbfc4e | 479 | } |
fbb18613 | 480 | else if (!strcmp (name, "ident")) |
e8bbfc4e | 481 | { |
0e5921e8 ZW |
482 | /* #ident. We expect a string constant here. |
483 | The pedantic warning and syntax error are now in cpp. */ | |
e8bbfc4e | 484 | |
0e5921e8 ZW |
485 | token = c_lex (&value); |
486 | if (token != CPP_STRING || TREE_CODE (value) != STRING_CST) | |
fbb18613 | 487 | goto skipline; |
e8bbfc4e | 488 | |
0e5921e8 | 489 | #ifdef ASM_OUTPUT_IDENT |
fbb18613 JM |
490 | if (! flag_no_ident) |
491 | { | |
0e5921e8 | 492 | ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (value)); |
e8bbfc4e | 493 | } |
0e5921e8 | 494 | #endif |
fbb18613 JM |
495 | |
496 | /* Skip the rest of this line. */ | |
497 | goto skipline; | |
e8bbfc4e RK |
498 | } |
499 | ||
fbb18613 | 500 | error ("undefined or invalid # directive `%s'", name); |
e8bbfc4e RK |
501 | goto skipline; |
502 | } | |
503 | ||
fbb18613 JM |
504 | /* If the # is the only nonwhite char on the line, |
505 | just ignore it. Check the new newline. */ | |
0e5921e8 | 506 | if (token == CPP_EOF) |
fbb18613 JM |
507 | goto skipline; |
508 | ||
e8bbfc4e RK |
509 | linenum: |
510 | /* Here we have either `#line' or `# <nonletter>'. | |
511 | In either case, it should be a line number; a digit should follow. */ | |
512 | ||
0e5921e8 | 513 | if (token != CPP_NUMBER || TREE_CODE (value) != INTEGER_CST) |
fbb18613 JM |
514 | { |
515 | error ("invalid #-line"); | |
516 | goto skipline; | |
517 | } | |
e8bbfc4e | 518 | |
fbb18613 JM |
519 | /* subtract one, because it is the following line that |
520 | gets the specified number */ | |
e8bbfc4e | 521 | |
0e5921e8 | 522 | l = TREE_INT_CST_LOW (value) - 1; |
e8bbfc4e | 523 | |
fbb18613 JM |
524 | /* More follows: it must be a string constant (filename). |
525 | It would be neat to use cpplib to quickly process the string, but | |
526 | (1) we don't have a handy tokenization of the string, and | |
527 | (2) I don't know how well that would work in the presense | |
528 | of filenames that contain wide characters. */ | |
e8bbfc4e | 529 | |
fbb18613 | 530 | if (saw_line) |
e8bbfc4e | 531 | { |
fbb18613 JM |
532 | /* Don't treat \ as special if we are processing #line 1 "...". |
533 | If you want it to be treated specially, use # 1 "...". */ | |
534 | ignore_escape_flag = 1; | |
535 | } | |
e8bbfc4e | 536 | |
fbb18613 | 537 | /* Read the string constant. */ |
0e5921e8 | 538 | token = c_lex (&value); |
e8bbfc4e | 539 | |
fbb18613 | 540 | ignore_escape_flag = 0; |
e8bbfc4e | 541 | |
0e5921e8 | 542 | if (token == CPP_EOF) |
fbb18613 JM |
543 | { |
544 | /* No more: store the line number and check following line. */ | |
0e5921e8 | 545 | lex_lineno = l; |
fbb18613 JM |
546 | goto skipline; |
547 | } | |
e8bbfc4e | 548 | |
0e5921e8 | 549 | if (token != CPP_STRING || TREE_CODE (value) != STRING_CST) |
fbb18613 JM |
550 | { |
551 | error ("invalid #line"); | |
552 | goto skipline; | |
553 | } | |
e8bbfc4e | 554 | |
0e5921e8 | 555 | new_file = TREE_STRING_POINTER (value); |
e8bbfc4e | 556 | |
fbb18613 | 557 | if (main_input_filename == 0) |
0e5921e8 | 558 | main_input_filename = new_file; |
e8bbfc4e | 559 | |
fbb18613 JM |
560 | action = act_none; |
561 | action_number = 0; | |
e8bbfc4e | 562 | |
fbb18613 JM |
563 | /* Each change of file name |
564 | reinitializes whether we are now in a system header. */ | |
565 | in_system_header = 0; | |
e8bbfc4e | 566 | |
fbb18613 JM |
567 | if (!read_line_number (&action_number)) |
568 | { | |
569 | /* Update the name in the top element of input_file_stack. */ | |
5b450ae5 RS |
570 | if (input_file_stack) |
571 | input_file_stack->name = input_filename; | |
fbb18613 | 572 | } |
5b450ae5 | 573 | |
fbb18613 JM |
574 | /* `1' after file name means entering new file. |
575 | `2' after file name means just left a file. */ | |
e8bbfc4e | 576 | |
fbb18613 JM |
577 | if (action_number == 1) |
578 | { | |
579 | action = act_push; | |
580 | read_line_number (&action_number); | |
581 | } | |
582 | else if (action_number == 2) | |
583 | { | |
584 | action = act_pop; | |
585 | read_line_number (&action_number); | |
586 | } | |
587 | if (action_number == 3) | |
588 | { | |
e8bbfc4e | 589 | /* `3' after file name means this is a system header file. */ |
fbb18613 JM |
590 | in_system_header = 1; |
591 | read_line_number (&action_number); | |
592 | } | |
0e5921e8 ZW |
593 | #ifndef NO_IMPLICIT_EXTERN_C |
594 | if (action_number == 4) | |
595 | { | |
596 | /* `4' after file name means this is a C header file. */ | |
597 | entering_c_header = 1; | |
598 | read_line_number (&action_number); | |
599 | } | |
600 | #endif | |
e8bbfc4e | 601 | |
fbb18613 | 602 | /* Do the actions implied by the preceding numbers. */ |
fbb18613 JM |
603 | if (action == act_push) |
604 | { | |
0e5921e8 ZW |
605 | lineno = lex_lineno; |
606 | push_srcloc (input_filename, 1); | |
607 | input_file_stack->indent_level = indent_level; | |
fbb18613 | 608 | debug_start_source_file (input_filename); |
0e5921e8 ZW |
609 | #ifndef NO_IMPLICIT_EXTERN_C |
610 | if (c_header_level) | |
611 | ++c_header_level; | |
612 | else if (entering_c_header) | |
613 | { | |
614 | c_header_level = 1; | |
615 | ++pending_lang_change; | |
616 | } | |
617 | #endif | |
fbb18613 JM |
618 | } |
619 | else if (action == act_pop) | |
620 | { | |
621 | /* Popping out of a file. */ | |
622 | if (input_file_stack->next) | |
0468bc75 | 623 | { |
0e5921e8 ZW |
624 | #ifndef NO_IMPLICIT_EXTERN_C |
625 | if (c_header_level && --c_header_level == 0) | |
626 | { | |
627 | if (entering_c_header) | |
628 | warning ("badly nested C headers from preprocessor"); | |
629 | --pending_lang_change; | |
630 | } | |
631 | #endif | |
632 | #if 0 | |
633 | if (indent_level != input_file_stack->indent_level) | |
fbb18613 JM |
634 | { |
635 | warning_with_file_and_line | |
0e5921e8 ZW |
636 | (input_filename, lex_lineno, |
637 | "This file contains more '%c's than '%c's.", | |
638 | indent_level > input_file_stack->indent_level ? '{' : '}', | |
639 | indent_level > input_file_stack->indent_level ? '}' : '{'); | |
fbb18613 | 640 | } |
0e5921e8 ZW |
641 | #endif |
642 | pop_srcloc (); | |
fbb18613 | 643 | debug_end_source_file (input_file_stack->line); |
0468bc75 | 644 | } |
fbb18613 JM |
645 | else |
646 | error ("#-lines for entering and leaving files don't match"); | |
e8bbfc4e | 647 | } |
fbb18613 | 648 | |
0e5921e8 ZW |
649 | update_header_times (new_file); |
650 | ||
651 | input_filename = new_file; | |
652 | lex_lineno = l; | |
653 | ||
654 | /* Hook for C++. */ | |
655 | extract_interface_info (); | |
e8bbfc4e RK |
656 | |
657 | /* skip the rest of this line. */ | |
658 | skipline: | |
fbb18613 | 659 | linemode = 0; |
fbb18613 | 660 | |
0e5921e8 | 661 | while (getch () != '\n'); |
e8bbfc4e | 662 | } |
0e5921e8 | 663 | #else /* USE_CPPLIB */ |
ca5b800a | 664 | |
0e5921e8 ZW |
665 | /* Not yet handled: #pragma, #define, #undef. |
666 | No need to deal with linemarkers under normal conditions. */ | |
ca5b800a | 667 | |
0e5921e8 ZW |
668 | static void |
669 | cb_ident (pfile, str, len) | |
670 | cpp_reader *pfile ATTRIBUTE_UNUSED; | |
671 | const unsigned char *str; | |
672 | unsigned int len; | |
673 | { | |
674 | #ifdef ASM_OUTPUT_IDENT | |
675 | if (! flag_no_ident) | |
676 | { | |
677 | /* Convert escapes in the string. */ | |
678 | tree value = lex_string ((const char *)str, len, 0); | |
679 | ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (value)); | |
680 | } | |
681 | #endif | |
682 | } | |
ca5b800a | 683 | |
0e5921e8 ZW |
684 | static void |
685 | cb_enter_file (pfile) | |
686 | cpp_reader *pfile; | |
ca5b800a | 687 | { |
0e5921e8 ZW |
688 | cpp_buffer *ip = CPP_BUFFER (pfile); |
689 | /* Bleah, need a better interface to this. */ | |
690 | const char *flags = cpp_syshdr_flags (pfile, ip); | |
691 | ||
692 | /* Mustn't stack the main buffer on the input stack. (Ick.) */ | |
693 | if (ip->prev) | |
694 | { | |
695 | lex_lineno = lineno = ip->prev->lineno - 1; | |
696 | push_srcloc (ggc_alloc_string (ip->nominal_fname, -1), 1); | |
697 | input_file_stack->indent_level = indent_level; | |
698 | debug_start_source_file (ip->nominal_fname); | |
699 | } | |
700 | else | |
701 | lex_lineno = 1; | |
702 | ||
703 | update_header_times (ip->nominal_fname); | |
704 | ||
705 | /* Hook for C++. */ | |
706 | extract_interface_info (); | |
707 | ||
708 | in_system_header = (flags[0] != 0); | |
709 | #ifndef NO_IMPLICIT_EXTERN_C | |
710 | if (c_header_level) | |
711 | ++c_header_level; | |
712 | else if (flags[2] != 0) | |
713 | { | |
714 | c_header_level = 1; | |
715 | ++pending_lang_change; | |
716 | } | |
717 | #endif | |
718 | } | |
719 | ||
720 | static void | |
721 | cb_leave_file (pfile) | |
722 | cpp_reader *pfile; | |
723 | { | |
724 | /* Bleah, need a better interface to this. */ | |
725 | const char *flags = cpp_syshdr_flags (pfile, CPP_BUFFER (pfile)); | |
726 | #if 0 | |
727 | if (indent_level != input_file_stack->indent_level) | |
728 | { | |
729 | warning_with_file_and_line | |
730 | (input_filename, lex_lineno, | |
731 | "This file contains more '%c's than '%c's.", | |
732 | indent_level > input_file_stack->indent_level ? '{' : '}', | |
733 | indent_level > input_file_stack->indent_level ? '}' : '{'); | |
734 | } | |
735 | #endif | |
736 | /* We get called for the main buffer, but we mustn't pop it. */ | |
737 | if (input_file_stack->next) | |
738 | pop_srcloc (); | |
739 | in_system_header = (flags[0] != 0); | |
740 | #ifndef NO_IMPLICIT_EXTERN_C | |
741 | if (c_header_level && --c_header_level == 0) | |
742 | { | |
743 | if (flags[2] != 0) | |
744 | warning ("badly nested C headers from preprocessor"); | |
745 | --pending_lang_change; | |
746 | } | |
747 | #endif | |
748 | lex_lineno = CPP_BUFFER (pfile)->lineno; | |
749 | debug_end_source_file (input_file_stack->line); | |
750 | ||
751 | update_header_times (input_file_stack->name); | |
752 | /* Hook for C++. */ | |
753 | extract_interface_info (); | |
754 | } | |
755 | ||
756 | static void | |
757 | cb_rename_file (pfile) | |
758 | cpp_reader *pfile; | |
759 | { | |
760 | cpp_buffer *ip = CPP_BUFFER (pfile); | |
761 | /* Bleah, need a better interface to this. */ | |
762 | const char *flags = cpp_syshdr_flags (pfile, ip); | |
763 | input_filename = ggc_alloc_string (ip->nominal_fname, -1); | |
764 | lex_lineno = ip->lineno; | |
765 | in_system_header = (flags[0] != 0); | |
766 | ||
767 | update_header_times (ip->nominal_fname); | |
768 | /* Hook for C++. */ | |
769 | extract_interface_info (); | |
770 | } | |
8b97c5f8 ZW |
771 | |
772 | static void | |
773 | cb_def_pragma (pfile) | |
774 | cpp_reader *pfile; | |
775 | { | |
776 | /* Issue a warning message if we have been asked to do so. Ignore | |
777 | unknown pragmas in system headers unless an explicit | |
778 | -Wunknown-pragmas has been given. */ | |
779 | if (warn_unknown_pragmas > in_system_header) | |
780 | { | |
781 | const unsigned char *space, *name; | |
782 | const cpp_token *t = pfile->first_directive_token + 2; | |
783 | ||
784 | space = t[0].val.node->name; | |
785 | name = t[1].type == CPP_NAME ? t[1].val.node->name : 0; | |
786 | if (name) | |
787 | warning ("ignoring #pragma %s %s", space, name); | |
788 | else | |
789 | warning ("ignoring #pragma %s", space); | |
790 | } | |
791 | } | |
0e5921e8 ZW |
792 | #endif /* USE_CPPLIB */ |
793 | ||
794 | /* Parse a '\uNNNN' or '\UNNNNNNNN' sequence. | |
795 | ||
796 | [lex.charset]: The character designated by the universal-character-name | |
797 | \UNNNNNNNN is that character whose character short name in ISO/IEC 10646 | |
798 | is NNNNNNNN; the character designated by the universal-character-name | |
799 | \uNNNN is that character whose character short name in ISO/IEC 10646 is | |
800 | 0000NNNN. If the hexadecimal value for a universal character name is | |
801 | less than 0x20 or in the range 0x7F-0x9F (inclusive), or if the | |
802 | universal character name designates a character in the basic source | |
803 | character set, then the program is ill-formed. | |
804 | ||
805 | We assume that wchar_t is Unicode, so we don't need to do any | |
806 | mapping. Is this ever wrong? */ | |
807 | ||
808 | static const char * | |
809 | read_ucs (p, limit, cptr, length) | |
810 | const char *p; | |
811 | const char *limit; | |
812 | unsigned int *cptr; | |
813 | int length; | |
814 | { | |
815 | unsigned int code = 0; | |
816 | int c; | |
817 | ||
818 | for (; length; --length) | |
ca5b800a | 819 | { |
0e5921e8 | 820 | if (p >= limit) |
a3100298 | 821 | { |
0e5921e8 | 822 | error ("incomplete universal-character-name"); |
a3100298 | 823 | break; |
0e5921e8 | 824 | } |
fbb18613 | 825 | |
0e5921e8 ZW |
826 | c = *p++; |
827 | if (! ISXDIGIT (c)) | |
828 | { | |
829 | error ("non hex digit '%c' in universal-character-name", c); | |
830 | p--; | |
831 | break; | |
a3100298 | 832 | } |
f09db6e0 | 833 | |
0e5921e8 ZW |
834 | code <<= 4; |
835 | if (c >= 'a' && c <= 'f') | |
836 | code += c - 'a' + 10; | |
837 | if (c >= 'A' && c <= 'F') | |
838 | code += c - 'A' + 10; | |
839 | if (c >= '0' && c <= '9') | |
840 | code += c - '0'; | |
ca5b800a | 841 | } |
ca5b800a | 842 | |
0e5921e8 ZW |
843 | #ifdef TARGET_EBCDIC |
844 | sorry ("universal-character-name on EBCDIC target"); | |
845 | *cptr = 0x3f; /* EBCDIC invalid character */ | |
846 | return p; | |
847 | #endif | |
848 | ||
849 | if (code > 0x9f && !(code & 0x80000000)) | |
850 | /* True extended character, OK. */; | |
851 | else if (code >= 0x20 && code < 0x7f) | |
852 | { | |
853 | /* ASCII printable character. The C character set consists of all of | |
854 | these except $, @ and `. We use hex escapes so that this also | |
855 | works with EBCDIC hosts. */ | |
856 | if (code != 0x24 && code != 0x40 && code != 0x60) | |
857 | error ("universal-character-name used for '%c'", code); | |
858 | } | |
859 | else | |
860 | error ("invalid universal-character-name"); | |
861 | ||
862 | *cptr = code; | |
863 | return p; | |
864 | } | |
e8bbfc4e | 865 | |
0e5921e8 ZW |
866 | /* Read an escape sequence and write its character equivalent into *CPTR. |
867 | P is the input pointer, which is just after the backslash. LIMIT | |
868 | is how much text we have. | |
869 | Returns the updated input pointer. */ | |
e8bbfc4e | 870 | |
0e5921e8 ZW |
871 | static const char * |
872 | readescape (p, limit, cptr) | |
873 | const char *p; | |
874 | const char *limit; | |
875 | unsigned int *cptr; | |
e8bbfc4e | 876 | { |
0e5921e8 | 877 | unsigned int c, code, count; |
1c7b145e | 878 | unsigned firstdig = 0; |
8696da34 | 879 | int nonnull; |
e8bbfc4e | 880 | |
0e5921e8 ZW |
881 | if (p == limit) |
882 | { | |
883 | /* cpp has already issued an error for this. */ | |
884 | *cptr = 0; | |
885 | return p; | |
886 | } | |
887 | ||
888 | c = *p++; | |
889 | ||
e8bbfc4e RK |
890 | switch (c) |
891 | { | |
892 | case 'x': | |
cde6e684 | 893 | if (warn_traditional && !in_system_header) |
e8bbfc4e RK |
894 | warning ("the meaning of `\\x' varies with -traditional"); |
895 | ||
896 | if (flag_traditional) | |
0e5921e8 ZW |
897 | { |
898 | *cptr = 'x'; | |
899 | return p; | |
900 | } | |
e8bbfc4e RK |
901 | |
902 | code = 0; | |
903 | count = 0; | |
8696da34 | 904 | nonnull = 0; |
0e5921e8 | 905 | while (p < limit) |
e8bbfc4e | 906 | { |
0e5921e8 | 907 | c = *p++; |
fbb18613 | 908 | if (! ISXDIGIT (c)) |
e8bbfc4e | 909 | { |
0e5921e8 | 910 | p--; |
e8bbfc4e RK |
911 | break; |
912 | } | |
913 | code *= 16; | |
914 | if (c >= 'a' && c <= 'f') | |
915 | code += c - 'a' + 10; | |
916 | if (c >= 'A' && c <= 'F') | |
917 | code += c - 'A' + 10; | |
918 | if (c >= '0' && c <= '9') | |
919 | code += c - '0'; | |
8696da34 RS |
920 | if (code != 0 || count != 0) |
921 | { | |
922 | if (count == 0) | |
923 | firstdig = code; | |
924 | count++; | |
925 | } | |
926 | nonnull = 1; | |
e8bbfc4e | 927 | } |
8696da34 | 928 | if (! nonnull) |
4082292a AO |
929 | { |
930 | warning ("\\x used with no following hex digits"); | |
0e5921e8 ZW |
931 | *cptr = 'x'; |
932 | return p; | |
4082292a | 933 | } |
be63d912 RS |
934 | else if (count == 0) |
935 | /* Digits are all 0's. Ok. */ | |
936 | ; | |
e8bbfc4e RK |
937 | else if ((count - 1) * 4 >= TYPE_PRECISION (integer_type_node) |
938 | || (count > 1 | |
fbb18613 JM |
939 | && (((unsigned)1 |
940 | << (TYPE_PRECISION (integer_type_node) | |
941 | - (count - 1) * 4)) | |
e8bbfc4e RK |
942 | <= firstdig))) |
943 | pedwarn ("hex escape out of range"); | |
0e5921e8 ZW |
944 | *cptr = code; |
945 | return p; | |
e8bbfc4e RK |
946 | |
947 | case '0': case '1': case '2': case '3': case '4': | |
948 | case '5': case '6': case '7': | |
949 | code = 0; | |
0e5921e8 | 950 | for (count = 0; count < 3; count++) |
e8bbfc4e | 951 | { |
0e5921e8 ZW |
952 | if (c < '0' || c > '7') |
953 | { | |
954 | p--; | |
955 | break; | |
956 | } | |
e8bbfc4e | 957 | code = (code * 8) + (c - '0'); |
0e5921e8 ZW |
958 | if (p == limit) |
959 | break; | |
960 | c = *p++; | |
e8bbfc4e | 961 | } |
e8bbfc4e | 962 | |
0e5921e8 ZW |
963 | if (count == 3) |
964 | p--; | |
e8bbfc4e | 965 | |
0e5921e8 ZW |
966 | *cptr = code; |
967 | return p; | |
e8bbfc4e | 968 | |
0e5921e8 ZW |
969 | case '\\': case '\'': case '"': case '?': |
970 | *cptr = c; | |
971 | return p; | |
e8bbfc4e | 972 | |
0e5921e8 ZW |
973 | case 'n': *cptr = TARGET_NEWLINE; return p; |
974 | case 't': *cptr = TARGET_TAB; return p; | |
975 | case 'r': *cptr = TARGET_CR; return p; | |
976 | case 'f': *cptr = TARGET_FF; return p; | |
977 | case 'b': *cptr = TARGET_BS; return p; | |
978 | case 'v': *cptr = TARGET_VT; return p; | |
979 | case 'a': | |
980 | if (warn_traditional && !in_system_header) | |
981 | warning ("the meaning of '\\a' varies with -traditional"); | |
982 | *cptr = flag_traditional ? c : TARGET_BELL; | |
983 | return p; | |
e8bbfc4e | 984 | |
0e5921e8 ZW |
985 | /* Warnings and support checks handled by read_ucs(). */ |
986 | case 'u': case 'U': | |
987 | if (c_language != clk_cplusplus && !flag_isoc99) | |
988 | break; | |
e8bbfc4e | 989 | |
cde6e684 | 990 | if (warn_traditional && !in_system_header) |
0e5921e8 | 991 | warning ("the meaning of '\\%c' varies with -traditional", c); |
e8bbfc4e | 992 | |
0e5921e8 ZW |
993 | return read_ucs (p, limit, cptr, c == 'u' ? 4 : 8); |
994 | ||
995 | case 'e': case 'E': | |
dad112ca | 996 | if (pedantic) |
0e5921e8 ZW |
997 | pedwarn ("non-ISO-standard escape sequence, '\\%c'", c); |
998 | *cptr = TARGET_ESC; return p; | |
e8bbfc4e | 999 | |
0e5921e8 ZW |
1000 | /* '\(', etc, are used at beginning of line to avoid confusing Emacs. |
1001 | '\%' is used to prevent SCCS from getting confused. */ | |
1002 | case '(': case '{': case '[': case '%': | |
e8bbfc4e | 1003 | if (pedantic) |
0e5921e8 ZW |
1004 | pedwarn ("unknown escape sequence '\\%c'", c); |
1005 | *cptr = c; | |
1006 | return p; | |
e8bbfc4e | 1007 | } |
0e5921e8 | 1008 | |
fbb18613 | 1009 | if (ISGRAPH (c)) |
0e5921e8 | 1010 | pedwarn ("unknown escape sequence '\\%c'", c); |
e8bbfc4e | 1011 | else |
0e5921e8 ZW |
1012 | pedwarn ("unknown escape sequence: '\\' followed by char 0x%.2x", c); |
1013 | ||
1014 | *cptr = c; | |
1015 | return p; | |
e8bbfc4e | 1016 | } |
0e5921e8 ZW |
1017 | |
1018 | #if 0 /* not yet */ | |
1019 | /* Returns nonzero if C is a universal-character-name. Give an error if it | |
1020 | is not one which may appear in an identifier, as per [extendid]. | |
1021 | ||
1022 | Note that extended character support in identifiers has not yet been | |
1023 | implemented. It is my personal opinion that this is not a desirable | |
1024 | feature. Portable code cannot count on support for more than the basic | |
1025 | identifier character set. */ | |
1026 | ||
1027 | static inline int | |
1028 | is_extended_char (c) | |
1029 | int c; | |
1030 | { | |
1031 | #ifdef TARGET_EBCDIC | |
1032 | return 0; | |
1033 | #else | |
1034 | /* ASCII. */ | |
1035 | if (c < 0x7f) | |
1036 | return 0; | |
1037 | ||
1038 | /* None of the valid chars are outside the Basic Multilingual Plane (the | |
1039 | low 16 bits). */ | |
1040 | if (c > 0xffff) | |
1041 | { | |
1042 | error ("universal-character-name '\\U%08x' not valid in identifier", c); | |
1043 | return 1; | |
1044 | } | |
1045 | ||
1046 | /* Latin */ | |
1047 | if ((c >= 0x00c0 && c <= 0x00d6) | |
1048 | || (c >= 0x00d8 && c <= 0x00f6) | |
1049 | || (c >= 0x00f8 && c <= 0x01f5) | |
1050 | || (c >= 0x01fa && c <= 0x0217) | |
1051 | || (c >= 0x0250 && c <= 0x02a8) | |
1052 | || (c >= 0x1e00 && c <= 0x1e9a) | |
1053 | || (c >= 0x1ea0 && c <= 0x1ef9)) | |
1054 | return 1; | |
1055 | ||
1056 | /* Greek */ | |
1057 | if ((c == 0x0384) | |
1058 | || (c >= 0x0388 && c <= 0x038a) | |
1059 | || (c == 0x038c) | |
1060 | || (c >= 0x038e && c <= 0x03a1) | |
1061 | || (c >= 0x03a3 && c <= 0x03ce) | |
1062 | || (c >= 0x03d0 && c <= 0x03d6) | |
1063 | || (c == 0x03da) | |
1064 | || (c == 0x03dc) | |
1065 | || (c == 0x03de) | |
1066 | || (c == 0x03e0) | |
1067 | || (c >= 0x03e2 && c <= 0x03f3) | |
1068 | || (c >= 0x1f00 && c <= 0x1f15) | |
1069 | || (c >= 0x1f18 && c <= 0x1f1d) | |
1070 | || (c >= 0x1f20 && c <= 0x1f45) | |
1071 | || (c >= 0x1f48 && c <= 0x1f4d) | |
1072 | || (c >= 0x1f50 && c <= 0x1f57) | |
1073 | || (c == 0x1f59) | |
1074 | || (c == 0x1f5b) | |
1075 | || (c == 0x1f5d) | |
1076 | || (c >= 0x1f5f && c <= 0x1f7d) | |
1077 | || (c >= 0x1f80 && c <= 0x1fb4) | |
1078 | || (c >= 0x1fb6 && c <= 0x1fbc) | |
1079 | || (c >= 0x1fc2 && c <= 0x1fc4) | |
1080 | || (c >= 0x1fc6 && c <= 0x1fcc) | |
1081 | || (c >= 0x1fd0 && c <= 0x1fd3) | |
1082 | || (c >= 0x1fd6 && c <= 0x1fdb) | |
1083 | || (c >= 0x1fe0 && c <= 0x1fec) | |
1084 | || (c >= 0x1ff2 && c <= 0x1ff4) | |
1085 | || (c >= 0x1ff6 && c <= 0x1ffc)) | |
1086 | return 1; | |
1087 | ||
1088 | /* Cyrillic */ | |
1089 | if ((c >= 0x0401 && c <= 0x040d) | |
1090 | || (c >= 0x040f && c <= 0x044f) | |
1091 | || (c >= 0x0451 && c <= 0x045c) | |
1092 | || (c >= 0x045e && c <= 0x0481) | |
1093 | || (c >= 0x0490 && c <= 0x04c4) | |
1094 | || (c >= 0x04c7 && c <= 0x04c8) | |
1095 | || (c >= 0x04cb && c <= 0x04cc) | |
1096 | || (c >= 0x04d0 && c <= 0x04eb) | |
1097 | || (c >= 0x04ee && c <= 0x04f5) | |
1098 | || (c >= 0x04f8 && c <= 0x04f9)) | |
1099 | return 1; | |
1100 | ||
1101 | /* Armenian */ | |
1102 | if ((c >= 0x0531 && c <= 0x0556) | |
1103 | || (c >= 0x0561 && c <= 0x0587)) | |
1104 | return 1; | |
1105 | ||
1106 | /* Hebrew */ | |
1107 | if ((c >= 0x05d0 && c <= 0x05ea) | |
1108 | || (c >= 0x05f0 && c <= 0x05f4)) | |
1109 | return 1; | |
1110 | ||
1111 | /* Arabic */ | |
1112 | if ((c >= 0x0621 && c <= 0x063a) | |
1113 | || (c >= 0x0640 && c <= 0x0652) | |
1114 | || (c >= 0x0670 && c <= 0x06b7) | |
1115 | || (c >= 0x06ba && c <= 0x06be) | |
1116 | || (c >= 0x06c0 && c <= 0x06ce) | |
1117 | || (c >= 0x06e5 && c <= 0x06e7)) | |
1118 | return 1; | |
1119 | ||
1120 | /* Devanagari */ | |
1121 | if ((c >= 0x0905 && c <= 0x0939) | |
1122 | || (c >= 0x0958 && c <= 0x0962)) | |
1123 | return 1; | |
1124 | ||
1125 | /* Bengali */ | |
1126 | if ((c >= 0x0985 && c <= 0x098c) | |
1127 | || (c >= 0x098f && c <= 0x0990) | |
1128 | || (c >= 0x0993 && c <= 0x09a8) | |
1129 | || (c >= 0x09aa && c <= 0x09b0) | |
1130 | || (c == 0x09b2) | |
1131 | || (c >= 0x09b6 && c <= 0x09b9) | |
1132 | || (c >= 0x09dc && c <= 0x09dd) | |
1133 | || (c >= 0x09df && c <= 0x09e1) | |
1134 | || (c >= 0x09f0 && c <= 0x09f1)) | |
1135 | return 1; | |
1136 | ||
1137 | /* Gurmukhi */ | |
1138 | if ((c >= 0x0a05 && c <= 0x0a0a) | |
1139 | || (c >= 0x0a0f && c <= 0x0a10) | |
1140 | || (c >= 0x0a13 && c <= 0x0a28) | |
1141 | || (c >= 0x0a2a && c <= 0x0a30) | |
1142 | || (c >= 0x0a32 && c <= 0x0a33) | |
1143 | || (c >= 0x0a35 && c <= 0x0a36) | |
1144 | || (c >= 0x0a38 && c <= 0x0a39) | |
1145 | || (c >= 0x0a59 && c <= 0x0a5c) | |
1146 | || (c == 0x0a5e)) | |
1147 | return 1; | |
1148 | ||
1149 | /* Gujarati */ | |
1150 | if ((c >= 0x0a85 && c <= 0x0a8b) | |
1151 | || (c == 0x0a8d) | |
1152 | || (c >= 0x0a8f && c <= 0x0a91) | |
1153 | || (c >= 0x0a93 && c <= 0x0aa8) | |
1154 | || (c >= 0x0aaa && c <= 0x0ab0) | |
1155 | || (c >= 0x0ab2 && c <= 0x0ab3) | |
1156 | || (c >= 0x0ab5 && c <= 0x0ab9) | |
1157 | || (c == 0x0ae0)) | |
1158 | return 1; | |
1159 | ||
1160 | /* Oriya */ | |
1161 | if ((c >= 0x0b05 && c <= 0x0b0c) | |
1162 | || (c >= 0x0b0f && c <= 0x0b10) | |
1163 | || (c >= 0x0b13 && c <= 0x0b28) | |
1164 | || (c >= 0x0b2a && c <= 0x0b30) | |
1165 | || (c >= 0x0b32 && c <= 0x0b33) | |
1166 | || (c >= 0x0b36 && c <= 0x0b39) | |
1167 | || (c >= 0x0b5c && c <= 0x0b5d) | |
1168 | || (c >= 0x0b5f && c <= 0x0b61)) | |
1169 | return 1; | |
1170 | ||
1171 | /* Tamil */ | |
1172 | if ((c >= 0x0b85 && c <= 0x0b8a) | |
1173 | || (c >= 0x0b8e && c <= 0x0b90) | |
1174 | || (c >= 0x0b92 && c <= 0x0b95) | |
1175 | || (c >= 0x0b99 && c <= 0x0b9a) | |
1176 | || (c == 0x0b9c) | |
1177 | || (c >= 0x0b9e && c <= 0x0b9f) | |
1178 | || (c >= 0x0ba3 && c <= 0x0ba4) | |
1179 | || (c >= 0x0ba8 && c <= 0x0baa) | |
1180 | || (c >= 0x0bae && c <= 0x0bb5) | |
1181 | || (c >= 0x0bb7 && c <= 0x0bb9)) | |
1182 | return 1; | |
1183 | ||
1184 | /* Telugu */ | |
1185 | if ((c >= 0x0c05 && c <= 0x0c0c) | |
1186 | || (c >= 0x0c0e && c <= 0x0c10) | |
1187 | || (c >= 0x0c12 && c <= 0x0c28) | |
1188 | || (c >= 0x0c2a && c <= 0x0c33) | |
1189 | || (c >= 0x0c35 && c <= 0x0c39) | |
1190 | || (c >= 0x0c60 && c <= 0x0c61)) | |
1191 | return 1; | |
1192 | ||
1193 | /* Kannada */ | |
1194 | if ((c >= 0x0c85 && c <= 0x0c8c) | |
1195 | || (c >= 0x0c8e && c <= 0x0c90) | |
1196 | || (c >= 0x0c92 && c <= 0x0ca8) | |
1197 | || (c >= 0x0caa && c <= 0x0cb3) | |
1198 | || (c >= 0x0cb5 && c <= 0x0cb9) | |
1199 | || (c >= 0x0ce0 && c <= 0x0ce1)) | |
1200 | return 1; | |
1201 | ||
1202 | /* Malayalam */ | |
1203 | if ((c >= 0x0d05 && c <= 0x0d0c) | |
1204 | || (c >= 0x0d0e && c <= 0x0d10) | |
1205 | || (c >= 0x0d12 && c <= 0x0d28) | |
1206 | || (c >= 0x0d2a && c <= 0x0d39) | |
1207 | || (c >= 0x0d60 && c <= 0x0d61)) | |
1208 | return 1; | |
1209 | ||
1210 | /* Thai */ | |
1211 | if ((c >= 0x0e01 && c <= 0x0e30) | |
1212 | || (c >= 0x0e32 && c <= 0x0e33) | |
1213 | || (c >= 0x0e40 && c <= 0x0e46) | |
1214 | || (c >= 0x0e4f && c <= 0x0e5b)) | |
1215 | return 1; | |
1216 | ||
1217 | /* Lao */ | |
1218 | if ((c >= 0x0e81 && c <= 0x0e82) | |
1219 | || (c == 0x0e84) | |
1220 | || (c == 0x0e87) | |
1221 | || (c == 0x0e88) | |
1222 | || (c == 0x0e8a) | |
1223 | || (c == 0x0e0d) | |
1224 | || (c >= 0x0e94 && c <= 0x0e97) | |
1225 | || (c >= 0x0e99 && c <= 0x0e9f) | |
1226 | || (c >= 0x0ea1 && c <= 0x0ea3) | |
1227 | || (c == 0x0ea5) | |
1228 | || (c == 0x0ea7) | |
1229 | || (c == 0x0eaa) | |
1230 | || (c == 0x0eab) | |
1231 | || (c >= 0x0ead && c <= 0x0eb0) | |
1232 | || (c == 0x0eb2) | |
1233 | || (c == 0x0eb3) | |
1234 | || (c == 0x0ebd) | |
1235 | || (c >= 0x0ec0 && c <= 0x0ec4) | |
1236 | || (c == 0x0ec6)) | |
1237 | return 1; | |
1238 | ||
1239 | /* Georgian */ | |
1240 | if ((c >= 0x10a0 && c <= 0x10c5) | |
1241 | || (c >= 0x10d0 && c <= 0x10f6)) | |
1242 | return 1; | |
1243 | ||
1244 | /* Hiragana */ | |
1245 | if ((c >= 0x3041 && c <= 0x3094) | |
1246 | || (c >= 0x309b && c <= 0x309e)) | |
1247 | return 1; | |
1248 | ||
1249 | /* Katakana */ | |
1250 | if ((c >= 0x30a1 && c <= 0x30fe)) | |
1251 | return 1; | |
1252 | ||
1253 | /* Bopmofo */ | |
1254 | if ((c >= 0x3105 && c <= 0x312c)) | |
1255 | return 1; | |
1256 | ||
1257 | /* Hangul */ | |
1258 | if ((c >= 0x1100 && c <= 0x1159) | |
1259 | || (c >= 0x1161 && c <= 0x11a2) | |
1260 | || (c >= 0x11a8 && c <= 0x11f9)) | |
1261 | return 1; | |
1262 | ||
1263 | /* CJK Unified Ideographs */ | |
1264 | if ((c >= 0xf900 && c <= 0xfa2d) | |
1265 | || (c >= 0xfb1f && c <= 0xfb36) | |
1266 | || (c >= 0xfb38 && c <= 0xfb3c) | |
1267 | || (c == 0xfb3e) | |
1268 | || (c >= 0xfb40 && c <= 0xfb41) | |
1269 | || (c >= 0xfb42 && c <= 0xfb44) | |
1270 | || (c >= 0xfb46 && c <= 0xfbb1) | |
1271 | || (c >= 0xfbd3 && c <= 0xfd3f) | |
1272 | || (c >= 0xfd50 && c <= 0xfd8f) | |
1273 | || (c >= 0xfd92 && c <= 0xfdc7) | |
1274 | || (c >= 0xfdf0 && c <= 0xfdfb) | |
1275 | || (c >= 0xfe70 && c <= 0xfe72) | |
1276 | || (c == 0xfe74) | |
1277 | || (c >= 0xfe76 && c <= 0xfefc) | |
1278 | || (c >= 0xff21 && c <= 0xff3a) | |
1279 | || (c >= 0xff41 && c <= 0xff5a) | |
1280 | || (c >= 0xff66 && c <= 0xffbe) | |
1281 | || (c >= 0xffc2 && c <= 0xffc7) | |
1282 | || (c >= 0xffca && c <= 0xffcf) | |
1283 | || (c >= 0xffd2 && c <= 0xffd7) | |
1284 | || (c >= 0xffda && c <= 0xffdc) | |
1285 | || (c >= 0x4e00 && c <= 0x9fa5)) | |
1286 | return 1; | |
1287 | ||
1288 | error ("universal-character-name '\\u%04x' not valid in identifier", c); | |
1289 | return 1; | |
1290 | #endif | |
1291 | } | |
1292 | ||
1293 | /* Add the UTF-8 representation of C to the token_buffer. */ | |
1294 | ||
1295 | static void | |
1296 | utf8_extend_token (c) | |
1297 | int c; | |
e8bbfc4e | 1298 | { |
0e5921e8 ZW |
1299 | int shift, mask; |
1300 | ||
1301 | if (c <= 0x0000007f) | |
1302 | { | |
1303 | extend_token (c); | |
1304 | return; | |
1305 | } | |
1306 | else if (c <= 0x000007ff) | |
1307 | shift = 6, mask = 0xc0; | |
1308 | else if (c <= 0x0000ffff) | |
1309 | shift = 12, mask = 0xe0; | |
1310 | else if (c <= 0x001fffff) | |
1311 | shift = 18, mask = 0xf0; | |
1312 | else if (c <= 0x03ffffff) | |
1313 | shift = 24, mask = 0xf8; | |
e8bbfc4e | 1314 | else |
0e5921e8 ZW |
1315 | shift = 30, mask = 0xfc; |
1316 | ||
1317 | extend_token (mask | (c >> shift)); | |
1318 | do | |
1319 | { | |
1320 | shift -= 6; | |
1321 | extend_token ((unsigned char) (0x80 | (c >> shift))); | |
1322 | } | |
1323 | while (shift); | |
e8bbfc4e | 1324 | } |
0e5921e8 | 1325 | #endif |
e8bbfc4e RK |
1326 | |
1327 | #if 0 | |
e8bbfc4e RK |
1328 | struct try_type |
1329 | { | |
1330 | tree *node_var; | |
1331 | char unsigned_flag; | |
1332 | char long_flag; | |
1333 | char long_long_flag; | |
1334 | }; | |
1335 | ||
75cb8865 | 1336 | struct try_type type_sequence[] = |
e8bbfc4e RK |
1337 | { |
1338 | { &integer_type_node, 0, 0, 0}, | |
1339 | { &unsigned_type_node, 1, 0, 0}, | |
1340 | { &long_integer_type_node, 0, 1, 0}, | |
1341 | { &long_unsigned_type_node, 1, 1, 0}, | |
1342 | { &long_long_integer_type_node, 0, 1, 1}, | |
1343 | { &long_long_unsigned_type_node, 1, 1, 1} | |
1344 | }; | |
1345 | #endif /* 0 */ | |
1346 | \f | |
ed513abf KG |
1347 | struct pf_args |
1348 | { | |
1349 | /* Input */ | |
0e5921e8 ZW |
1350 | const char *str; |
1351 | int fflag; | |
1352 | int lflag; | |
ed513abf | 1353 | int base; |
fbb18613 | 1354 | /* Output */ |
ed513abf | 1355 | int conversion_errno; |
ed513abf | 1356 | REAL_VALUE_TYPE value; |
0e5921e8 | 1357 | tree type; |
ed513abf KG |
1358 | }; |
1359 | ||
1360 | static void | |
1361 | parse_float (data) | |
1362 | PTR data; | |
1363 | { | |
1364 | struct pf_args * args = (struct pf_args *) data; | |
0e5921e8 ZW |
1365 | const char *typename; |
1366 | ||
fbb18613 JM |
1367 | args->conversion_errno = 0; |
1368 | args->type = double_type_node; | |
0e5921e8 | 1369 | typename = "double"; |
ed513abf KG |
1370 | |
1371 | /* The second argument, machine_mode, of REAL_VALUE_ATOF | |
1372 | tells the desired precision of the binary result | |
1373 | of decimal-to-binary conversion. */ | |
1374 | ||
0e5921e8 | 1375 | if (args->fflag) |
ed513abf | 1376 | { |
0e5921e8 ZW |
1377 | if (args->lflag) |
1378 | error ("both 'f' and 'l' suffixes on floating constant"); | |
ed513abf KG |
1379 | |
1380 | args->type = float_type_node; | |
0e5921e8 | 1381 | typename = "float"; |
ed513abf | 1382 | } |
0e5921e8 | 1383 | else if (args->lflag) |
ed513abf KG |
1384 | { |
1385 | args->type = long_double_type_node; | |
0e5921e8 | 1386 | typename = "long double"; |
ed513abf | 1387 | } |
0e5921e8 | 1388 | else if (flag_single_precision_constant) |
ed513abf | 1389 | { |
0e5921e8 ZW |
1390 | args->type = float_type_node; |
1391 | typename = "float"; | |
ed513abf | 1392 | } |
0e5921e8 ZW |
1393 | |
1394 | errno = 0; | |
1395 | if (args->base == 16) | |
1396 | args->value = REAL_VALUE_HTOF (args->str, TYPE_MODE (args->type)); | |
1397 | else | |
1398 | args->value = REAL_VALUE_ATOF (args->str, TYPE_MODE (args->type)); | |
1399 | ||
1400 | args->conversion_errno = errno; | |
1401 | /* A diagnostic is required here by some ISO C testsuites. | |
1402 | This is not pedwarn, because some people don't want | |
1403 | an error for this. */ | |
1404 | if (REAL_VALUE_ISINF (args->value) && pedantic) | |
1405 | warning ("floating point number exceeds range of '%s'", typename); | |
ed513abf KG |
1406 | } |
1407 | ||
0e5921e8 ZW |
1408 | int |
1409 | c_lex (value) | |
1410 | tree *value; | |
fbb18613 JM |
1411 | { |
1412 | #if USE_CPPLIB | |
0e5921e8 ZW |
1413 | const cpp_token *tok; |
1414 | enum cpp_ttype type; | |
1415 | ||
1416 | retry: | |
1417 | timevar_push (TV_CPP); | |
1418 | tok = cpp_get_token (&parse_in); | |
1419 | timevar_pop (TV_CPP); | |
1420 | ||
1421 | /* The C++ front end does horrible things with the current line | |
1422 | number. To ensure an accurate line number, we must reset it | |
1423 | every time we return a token. If we reset it from tok->line | |
1424 | every time, we'll get line numbers inside macros referring to the | |
1425 | macro definition; this is nice, but we don't want to change the | |
1426 | behavior until integrated mode is the only option. So we keep our | |
1427 | own idea of the line number, and reset it from tok->line at each | |
1428 | new line (which never happens inside a macro). */ | |
1429 | if (tok->flags & BOL) | |
1430 | lex_lineno = tok->line; | |
1431 | ||
1432 | *value = NULL_TREE; | |
1433 | lineno = lex_lineno; | |
1434 | type = tok->type; | |
1435 | switch (type) | |
1436 | { | |
1437 | case CPP_OPEN_BRACE: indent_level++; break; | |
1438 | case CPP_CLOSE_BRACE: indent_level--; break; | |
1439 | ||
1440 | /* Issue this error here, where we can get at tok->val.aux. */ | |
1441 | case CPP_OTHER: | |
1442 | if (ISGRAPH (tok->val.aux)) | |
1443 | error ("stray '%c' in program", tok->val.aux); | |
1444 | else | |
1445 | error ("stray '\\%#o' in program", tok->val.aux); | |
1446 | goto retry; | |
1447 | ||
1448 | case CPP_DEFINED: | |
1449 | type = CPP_NAME; | |
1450 | case CPP_NAME: | |
1451 | *value = get_identifier ((const char *)tok->val.node->name); | |
1452 | break; | |
fbb18613 | 1453 | |
0e5921e8 ZW |
1454 | case CPP_INT: |
1455 | case CPP_FLOAT: | |
1456 | case CPP_NUMBER: | |
1457 | *value = lex_number ((const char *)tok->val.str.text, tok->val.str.len); | |
1458 | break; | |
93868d11 | 1459 | |
0e5921e8 ZW |
1460 | case CPP_CHAR: |
1461 | case CPP_WCHAR: | |
1462 | *value = lex_charconst ((const char *)tok->val.str.text, | |
1463 | tok->val.str.len, tok->type == CPP_WCHAR); | |
1464 | break; | |
fbb18613 | 1465 | |
0e5921e8 ZW |
1466 | case CPP_STRING: |
1467 | case CPP_WSTRING: | |
1468 | case CPP_OSTRING: | |
1469 | *value = lex_string ((const char *)tok->val.str.text, | |
1470 | tok->val.str.len, tok->type == CPP_WSTRING); | |
1471 | break; | |
fbb18613 | 1472 | |
0e5921e8 ZW |
1473 | /* These tokens should not be visible outside cpplib. */ |
1474 | case CPP_HEADER_NAME: | |
1475 | case CPP_COMMENT: | |
1476 | case CPP_MACRO_ARG: | |
1477 | case CPP_PLACEMARKER: | |
1478 | abort (); | |
1479 | ||
1480 | default: break; | |
1481 | } | |
1482 | ||
1483 | return type; | |
1484 | ||
1485 | #else | |
1486 | int c; | |
1487 | char *p; | |
e8bbfc4e | 1488 | int wide_flag = 0; |
e31c7eec | 1489 | int objc_flag = 0; |
0e5921e8 ZW |
1490 | int charconst = 0; |
1491 | ||
1492 | *value = NULL_TREE; | |
e8bbfc4e | 1493 | |
0e5921e8 ZW |
1494 | retry: |
1495 | c = getch (); | |
e8bbfc4e RK |
1496 | |
1497 | /* Effectively do c = skip_white_space (c) | |
1498 | but do it faster in the usual cases. */ | |
1499 | while (1) | |
1500 | switch (c) | |
1501 | { | |
e8bbfc4e RK |
1502 | case ' ': |
1503 | case '\t': | |
1504 | case '\f': | |
1505 | case '\v': | |
0e5921e8 | 1506 | c = getch (); |
e8bbfc4e RK |
1507 | break; |
1508 | ||
1509 | case '\n': | |
e8bbfc4e RK |
1510 | c = skip_white_space (c); |
1511 | default: | |
1512 | goto found_nonwhite; | |
1513 | } | |
1514 | found_nonwhite: | |
1515 | ||
0e5921e8 | 1516 | lineno = lex_lineno; |
e8bbfc4e RK |
1517 | |
1518 | switch (c) | |
1519 | { | |
1520 | case EOF: | |
0e5921e8 | 1521 | return CPP_EOF; |
e8bbfc4e | 1522 | |
e8bbfc4e RK |
1523 | case 'L': |
1524 | /* Capital L may start a wide-string or wide-character constant. */ | |
1525 | { | |
0e5921e8 ZW |
1526 | register int c1 = getch(); |
1527 | if (c1 == '\'') | |
e8bbfc4e RK |
1528 | { |
1529 | wide_flag = 1; | |
1530 | goto char_constant; | |
1531 | } | |
0e5921e8 | 1532 | if (c1 == '"') |
e8bbfc4e RK |
1533 | { |
1534 | wide_flag = 1; | |
1535 | goto string_constant; | |
1536 | } | |
0e5921e8 | 1537 | put_back (c1); |
e8bbfc4e RK |
1538 | } |
1539 | goto letter; | |
1540 | ||
1541 | case '@': | |
1542 | if (!doing_objc_thang) | |
0e5921e8 | 1543 | goto straychar; |
e31c7eec | 1544 | else |
e8bbfc4e | 1545 | { |
e31c7eec | 1546 | /* '@' may start a constant string object. */ |
0e5921e8 ZW |
1547 | register int c1 = getch (); |
1548 | if (c1 == '"') | |
e31c7eec TW |
1549 | { |
1550 | objc_flag = 1; | |
1551 | goto string_constant; | |
1552 | } | |
0e5921e8 | 1553 | put_back (c1); |
ddd5a7c1 | 1554 | /* Fall through to treat '@' as the start of an identifier. */ |
e8bbfc4e RK |
1555 | } |
1556 | ||
e8bbfc4e RK |
1557 | case 'A': case 'B': case 'C': case 'D': case 'E': |
1558 | case 'F': case 'G': case 'H': case 'I': case 'J': | |
1559 | case 'K': case 'M': case 'N': case 'O': | |
1560 | case 'P': case 'Q': case 'R': case 'S': case 'T': | |
1561 | case 'U': case 'V': case 'W': case 'X': case 'Y': | |
1562 | case 'Z': | |
1563 | case 'a': case 'b': case 'c': case 'd': case 'e': | |
1564 | case 'f': case 'g': case 'h': case 'i': case 'j': | |
1565 | case 'k': case 'l': case 'm': case 'n': case 'o': | |
1566 | case 'p': case 'q': case 'r': case 's': case 't': | |
1567 | case 'u': case 'v': case 'w': case 'x': case 'y': | |
1568 | case 'z': | |
1569 | case '_': | |
f84cddb9 | 1570 | case '$': |
e8bbfc4e | 1571 | letter: |
0e5921e8 ZW |
1572 | p = token_buffer; |
1573 | while (ISALNUM (c) || c == '_' || c == '$' || c == '@') | |
e8bbfc4e | 1574 | { |
0e5921e8 ZW |
1575 | /* Make sure this char really belongs in an identifier. */ |
1576 | if (c == '$') | |
f84cddb9 | 1577 | { |
0e5921e8 ZW |
1578 | if (! dollars_in_ident) |
1579 | error ("'$' in identifier"); | |
1580 | else if (pedantic) | |
1581 | pedwarn ("'$' in identifier"); | |
fbb18613 | 1582 | } |
e8bbfc4e | 1583 | |
0e5921e8 ZW |
1584 | if (p >= token_buffer + maxtoken) |
1585 | p = extend_token_buffer (p); | |
e8bbfc4e | 1586 | |
0e5921e8 ZW |
1587 | *p++ = c; |
1588 | c = getch(); | |
1589 | } | |
75cb8865 | 1590 | |
0e5921e8 | 1591 | put_back (c); |
e8bbfc4e | 1592 | |
0e5921e8 ZW |
1593 | if (p >= token_buffer + maxtoken) |
1594 | p = extend_token_buffer (p); | |
1595 | *p = 0; | |
e8bbfc4e | 1596 | |
0e5921e8 ZW |
1597 | *value = get_identifier (token_buffer); |
1598 | return CPP_NAME; | |
e8bbfc4e | 1599 | |
fbb18613 | 1600 | case '.': |
fbb18613 JM |
1601 | { |
1602 | /* It's hard to preserve tokenization on '.' because | |
1603 | it could be a symbol by itself, or it could be the | |
1604 | start of a floating point number and cpp won't tell us. */ | |
0e5921e8 | 1605 | int c1 = getch (); |
fbb18613 JM |
1606 | if (c1 == '.') |
1607 | { | |
0e5921e8 ZW |
1608 | int c2 = getch (); |
1609 | if (c2 == '.') | |
1610 | return CPP_ELLIPSIS; | |
1611 | ||
1612 | put_back (c2); | |
1613 | error ("parse error at '..'"); | |
fbb18613 | 1614 | } |
0e5921e8 ZW |
1615 | else if (c1 == '*' && c_language == clk_cplusplus) |
1616 | return CPP_DOT_STAR; | |
1617 | ||
1618 | put_back (c1); | |
fbb18613 | 1619 | if (ISDIGIT (c1)) |
0e5921e8 | 1620 | goto number; |
fbb18613 | 1621 | } |
0e5921e8 | 1622 | return CPP_DOT; |
d669f5da | 1623 | |
0e5921e8 | 1624 | case '0': case '1': case '2': case '3': case '4': |
e8bbfc4e | 1625 | case '5': case '6': case '7': case '8': case '9': |
fbb18613 | 1626 | number: |
0e5921e8 ZW |
1627 | p = token_buffer; |
1628 | /* Scan the next preprocessing number. All C numeric constants | |
1629 | are preprocessing numbers, but not all preprocessing numbers | |
1630 | are valid numeric constants. Preprocessing numbers fit the | |
1631 | regular expression \.?[0-9]([0-9a-zA-Z_.]|[eEpP][+-])* | |
1632 | See C99 section 6.4.8. */ | |
1633 | for (;;) | |
1634 | { | |
1635 | if (p >= token_buffer + maxtoken) | |
1636 | p = extend_token_buffer (p); | |
cc144655 | 1637 | |
0e5921e8 ZW |
1638 | *p++ = c; |
1639 | c = getch(); | |
e8bbfc4e | 1640 | |
0e5921e8 ZW |
1641 | if (c == '+' || c == '-') |
1642 | { | |
1643 | int d = p[-1]; | |
1644 | if (d == 'e' || d == 'E' || d == 'p' || d == 'P') | |
1645 | continue; | |
1646 | } | |
1647 | if (ISALNUM (c) || c == '_' || c == '.') | |
1648 | continue; | |
1649 | break; | |
1650 | } | |
1651 | put_back (c); | |
e8bbfc4e | 1652 | |
0e5921e8 ZW |
1653 | *value = lex_number (token_buffer, p - token_buffer); |
1654 | return CPP_NUMBER; | |
e8bbfc4e | 1655 | |
0e5921e8 ZW |
1656 | case '\'': |
1657 | char_constant: | |
1658 | charconst = 1; | |
e8bbfc4e | 1659 | |
0e5921e8 ZW |
1660 | case '"': |
1661 | string_constant: | |
1662 | { | |
1663 | int delimiter = charconst ? '\'' : '"'; | |
1664 | #ifdef MULTIBYTE_CHARS | |
1665 | int longest_char = local_mb_cur_max (); | |
1666 | (void) local_mbtowc (NULL_PTR, NULL_PTR, 0); | |
1667 | #endif | |
1668 | c = getch (); | |
1669 | p = token_buffer + 1; | |
e8bbfc4e | 1670 | |
0e5921e8 | 1671 | while (c != delimiter && c != EOF) |
e8bbfc4e | 1672 | { |
0e5921e8 ZW |
1673 | if (p + 2 > token_buffer + maxtoken) |
1674 | p = extend_token_buffer (p); | |
1675 | ||
1676 | /* ignore_escape_flag is set for reading the filename in #line. */ | |
1677 | if (!ignore_escape_flag && c == '\\') | |
e8bbfc4e | 1678 | { |
0e5921e8 ZW |
1679 | *p++ = c; |
1680 | *p++ = getch (); /* escaped character */ | |
1681 | c = getch (); | |
1682 | continue; | |
e8bbfc4e RK |
1683 | } |
1684 | else | |
1685 | { | |
0e5921e8 ZW |
1686 | #ifdef MULTIBYTE_CHARS |
1687 | int i; | |
1688 | int char_len = -1; | |
1689 | for (i = 0; i < longest_char; ++i) | |
e8bbfc4e | 1690 | { |
0e5921e8 ZW |
1691 | if (p + i >= token_buffer + maxtoken) |
1692 | p = extend_token_buffer (p); | |
1693 | p[i] = c; | |
e8bbfc4e | 1694 | |
0e5921e8 ZW |
1695 | char_len = local_mblen (p, i + 1); |
1696 | if (char_len != -1) | |
1697 | break; | |
1698 | c = getch (); | |
e8bbfc4e | 1699 | } |
0e5921e8 | 1700 | if (char_len == -1) |
cc144655 | 1701 | { |
0e5921e8 ZW |
1702 | /* Replace all except the first byte. */ |
1703 | put_back (c); | |
1704 | for (--i; i > 0; --i) | |
1705 | put_back (p[i]); | |
1706 | char_len = 1; | |
cc144655 | 1707 | } |
0e5921e8 ZW |
1708 | /* mbtowc sometimes needs an extra char before accepting */ |
1709 | else if (char_len <= i) | |
1710 | put_back (c); | |
e8bbfc4e | 1711 | |
0e5921e8 ZW |
1712 | p += char_len; |
1713 | #else | |
1714 | *p++ = c; | |
1715 | #endif | |
1716 | c = getch (); | |
e8bbfc4e RK |
1717 | } |
1718 | } | |
0e5921e8 | 1719 | } |
e8bbfc4e | 1720 | |
0e5921e8 ZW |
1721 | if (charconst) |
1722 | { | |
1723 | *value = lex_charconst (token_buffer + 1, p - (token_buffer + 1), | |
1724 | wide_flag); | |
1725 | return wide_flag ? CPP_WCHAR : CPP_CHAR; | |
1726 | } | |
1727 | else | |
1728 | { | |
1729 | *value = lex_string (token_buffer + 1, p - (token_buffer + 1), | |
1730 | wide_flag); | |
1731 | return wide_flag ? CPP_WSTRING : objc_flag ? CPP_OSTRING : CPP_STRING; | |
1732 | } | |
e8bbfc4e | 1733 | |
0e5921e8 ZW |
1734 | case '+': |
1735 | case '-': | |
1736 | case '&': | |
1737 | case '|': | |
1738 | case ':': | |
1739 | case '<': | |
1740 | case '>': | |
1741 | case '*': | |
1742 | case '/': | |
1743 | case '%': | |
1744 | case '^': | |
1745 | case '!': | |
1746 | case '=': | |
1747 | { | |
1748 | int c1; | |
1749 | enum cpp_ttype type = CPP_EOF; | |
e8bbfc4e | 1750 | |
0e5921e8 | 1751 | switch (c) |
e8bbfc4e | 1752 | { |
0e5921e8 ZW |
1753 | case '+': type = CPP_PLUS; break; |
1754 | case '-': type = CPP_MINUS; break; | |
1755 | case '&': type = CPP_AND; break; | |
1756 | case '|': type = CPP_OR; break; | |
1757 | case ':': type = CPP_COLON; break; | |
1758 | case '<': type = CPP_LESS; break; | |
1759 | case '>': type = CPP_GREATER; break; | |
1760 | case '*': type = CPP_MULT; break; | |
1761 | case '/': type = CPP_DIV; break; | |
1762 | case '%': type = CPP_MOD; break; | |
1763 | case '^': type = CPP_XOR; break; | |
1764 | case '!': type = CPP_NOT; break; | |
1765 | case '=': type = CPP_EQ; break; | |
1766 | } | |
e8bbfc4e | 1767 | |
0e5921e8 | 1768 | c1 = getch (); |
e8bbfc4e | 1769 | |
0e5921e8 ZW |
1770 | if (c1 == '=' && type < CPP_LAST_EQ) |
1771 | return type + (CPP_EQ_EQ - CPP_EQ); | |
1772 | else if (c == c1) | |
1773 | switch (c) | |
1774 | { | |
1775 | case '+': return CPP_PLUS_PLUS; | |
1776 | case '-': return CPP_MINUS_MINUS; | |
1777 | case '&': return CPP_AND_AND; | |
1778 | case '|': return CPP_OR_OR; | |
1779 | case ':': | |
1780 | if (c_language == clk_cplusplus) | |
1781 | return CPP_SCOPE; | |
1782 | break; | |
e8bbfc4e | 1783 | |
0e5921e8 ZW |
1784 | case '<': type = CPP_LSHIFT; goto do_triad; |
1785 | case '>': type = CPP_RSHIFT; goto do_triad; | |
1786 | } | |
1787 | else | |
1788 | switch (c) | |
1789 | { | |
1790 | case '-': | |
1791 | if (c1 == '>') | |
1792 | { | |
1793 | if (c_language == clk_cplusplus) | |
1794 | { | |
1795 | c1 = getch (); | |
1796 | if (c1 == '*') | |
1797 | return CPP_DEREF_STAR; | |
1798 | put_back (c1); | |
1799 | } | |
1800 | return CPP_DEREF; | |
1801 | } | |
1802 | break; | |
ed513abf | 1803 | |
0e5921e8 ZW |
1804 | case '>': |
1805 | if (c1 == '?' && c_language == clk_cplusplus) | |
1806 | { type = CPP_MAX; goto do_triad; } | |
1807 | break; | |
e26ceb28 | 1808 | |
0e5921e8 ZW |
1809 | case '<': |
1810 | if (c1 == ':' && flag_digraphs) | |
1811 | return CPP_OPEN_SQUARE; | |
1812 | if (c1 == '%' && flag_digraphs) | |
1813 | { indent_level++; return CPP_OPEN_BRACE; } | |
1814 | if (c1 == '?' && c_language == clk_cplusplus) | |
1815 | { type = CPP_MIN; goto do_triad; } | |
1816 | break; | |
e8bbfc4e | 1817 | |
0e5921e8 ZW |
1818 | case ':': |
1819 | if (c1 == '>' && flag_digraphs) | |
1820 | return CPP_CLOSE_SQUARE; | |
1821 | break; | |
1822 | case '%': | |
1823 | if (c1 == '>' && flag_digraphs) | |
1824 | { indent_level--; return CPP_CLOSE_BRACE; } | |
1825 | break; | |
1826 | } | |
a47a0ed5 | 1827 | |
0e5921e8 ZW |
1828 | put_back (c1); |
1829 | return type; | |
e8bbfc4e | 1830 | |
0e5921e8 ZW |
1831 | do_triad: |
1832 | c1 = getch (); | |
1833 | if (c1 == '=') | |
1834 | type += (CPP_EQ_EQ - CPP_EQ); | |
1835 | else | |
1836 | put_back (c1); | |
1837 | return type; | |
1838 | } | |
e8bbfc4e | 1839 | |
0e5921e8 ZW |
1840 | case '~': return CPP_COMPL; |
1841 | case '?': return CPP_QUERY; | |
1842 | case ',': return CPP_COMMA; | |
1843 | case '(': return CPP_OPEN_PAREN; | |
1844 | case ')': return CPP_CLOSE_PAREN; | |
1845 | case '[': return CPP_OPEN_SQUARE; | |
1846 | case ']': return CPP_CLOSE_SQUARE; | |
1847 | case '{': indent_level++; return CPP_OPEN_BRACE; | |
1848 | case '}': indent_level--; return CPP_CLOSE_BRACE; | |
1849 | case ';': return CPP_SEMICOLON; | |
1850 | ||
1851 | straychar: | |
1852 | default: | |
1853 | if (ISGRAPH (c)) | |
1854 | error ("stray '%c' in program", c); | |
1855 | else | |
1856 | error ("stray '\\%#o' in program", c); | |
1857 | goto retry; | |
1858 | } | |
1859 | /* NOTREACHED */ | |
1860 | #endif | |
1861 | } | |
8d9bfdc5 | 1862 | |
8d9bfdc5 | 1863 | |
0e5921e8 | 1864 | #define ERROR(msgid) do { error(msgid); goto syntax_error; } while(0) |
75cb8865 | 1865 | |
0e5921e8 ZW |
1866 | static tree |
1867 | lex_number (str, len) | |
1868 | const char *str; | |
1869 | unsigned int len; | |
1870 | { | |
1871 | int base = 10; | |
1872 | int count = 0; | |
1873 | int largest_digit = 0; | |
1874 | int numdigits = 0; | |
1875 | int overflow = 0; | |
1876 | int c; | |
1877 | tree value; | |
1878 | const char *p; | |
1879 | enum anon1 { NOT_FLOAT = 0, AFTER_POINT, AFTER_EXPON } floatflag = NOT_FLOAT; | |
1880 | ||
1881 | /* We actually store only HOST_BITS_PER_CHAR bits in each part. | |
1882 | The code below which fills the parts array assumes that a host | |
1883 | int is at least twice as wide as a host char, and that | |
1884 | HOST_BITS_PER_WIDE_INT is an even multiple of HOST_BITS_PER_CHAR. | |
1885 | Two HOST_WIDE_INTs is the largest int literal we can store. | |
1886 | In order to detect overflow below, the number of parts (TOTAL_PARTS) | |
1887 | must be exactly the number of parts needed to hold the bits | |
1888 | of two HOST_WIDE_INTs. */ | |
1889 | #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2) | |
1890 | unsigned int parts[TOTAL_PARTS]; | |
1891 | ||
1892 | /* Optimize for most frequent case. */ | |
1893 | if (len == 1) | |
1894 | { | |
1895 | if (*str == '0') | |
1896 | return integer_zero_node; | |
1897 | else if (*str == '1') | |
1898 | return integer_one_node; | |
1899 | else | |
1900 | return build_int_2 (*str - '0', 0); | |
1901 | } | |
e8bbfc4e | 1902 | |
0e5921e8 ZW |
1903 | for (count = 0; count < TOTAL_PARTS; count++) |
1904 | parts[count] = 0; | |
e8bbfc4e | 1905 | |
0e5921e8 ZW |
1906 | /* len is known to be >1 at this point. */ |
1907 | p = str; | |
e8bbfc4e | 1908 | |
0e5921e8 ZW |
1909 | if (len > 2 && str[0] == '0' && (str[1] == 'x' || str[1] == 'X')) |
1910 | { | |
1911 | base = 16; | |
1912 | p = str + 2; | |
1913 | } | |
1914 | /* The ISDIGIT check is so we are not confused by a suffix on 0. */ | |
1915 | else if (str[0] == '0' && ISDIGIT (str[1])) | |
1916 | { | |
1917 | base = 8; | |
1918 | p = str + 1; | |
1919 | } | |
e8bbfc4e | 1920 | |
0e5921e8 ZW |
1921 | do |
1922 | { | |
1923 | c = *p++; | |
88d92ca5 | 1924 | |
0e5921e8 ZW |
1925 | if (c == '.') |
1926 | { | |
1927 | if (base == 16 && pedantic && !flag_isoc99) | |
1928 | pedwarn ("floating constant may not be in radix 16"); | |
1929 | else if (floatflag == AFTER_POINT) | |
1930 | ERROR ("too many decimal points in floating constant"); | |
1931 | else if (floatflag == AFTER_EXPON) | |
1932 | ERROR ("decimal point in exponent - impossible!"); | |
1933 | else | |
1934 | floatflag = AFTER_POINT; | |
c832a30e | 1935 | |
0e5921e8 ZW |
1936 | if (base == 8) |
1937 | base = 10; | |
1938 | } | |
1939 | else if (c == '_') | |
1940 | /* Possible future extension: silently ignore _ in numbers, | |
1941 | permitting cosmetic grouping - e.g. 0x8000_0000 == 0x80000000 | |
1942 | but somewhat easier to read. Ada has this? */ | |
1943 | ERROR ("underscore in number"); | |
1944 | else | |
1945 | { | |
1946 | int n; | |
1947 | /* It is not a decimal point. | |
1948 | It should be a digit (perhaps a hex digit). */ | |
0e0fda0d | 1949 | |
0e5921e8 ZW |
1950 | if (ISDIGIT (c)) |
1951 | { | |
1952 | n = c - '0'; | |
1953 | } | |
1954 | else if (base <= 10 && (c == 'e' || c == 'E')) | |
1955 | { | |
1956 | base = 10; | |
1957 | floatflag = AFTER_EXPON; | |
1958 | break; | |
1959 | } | |
1960 | else if (base == 16 && (c == 'p' || c == 'P')) | |
1961 | { | |
1962 | floatflag = AFTER_EXPON; | |
1963 | break; /* start of exponent */ | |
1964 | } | |
1965 | else if (base == 16 && c >= 'a' && c <= 'f') | |
1966 | { | |
1967 | n = c - 'a' + 10; | |
1968 | } | |
1969 | else if (base == 16 && c >= 'A' && c <= 'F') | |
1970 | { | |
1971 | n = c - 'A' + 10; | |
1972 | } | |
1973 | else | |
1974 | { | |
1975 | p--; | |
1976 | break; /* start of suffix */ | |
1977 | } | |
0e0fda0d | 1978 | |
0e5921e8 ZW |
1979 | if (n >= largest_digit) |
1980 | largest_digit = n; | |
1981 | numdigits++; | |
0e0fda0d | 1982 | |
0e5921e8 ZW |
1983 | for (count = 0; count < TOTAL_PARTS; count++) |
1984 | { | |
1985 | parts[count] *= base; | |
1986 | if (count) | |
1987 | { | |
1988 | parts[count] | |
1989 | += (parts[count-1] >> HOST_BITS_PER_CHAR); | |
1990 | parts[count-1] | |
1991 | &= (1 << HOST_BITS_PER_CHAR) - 1; | |
1992 | } | |
1993 | else | |
1994 | parts[0] += n; | |
1995 | } | |
e8bbfc4e | 1996 | |
0e5921e8 ZW |
1997 | /* If the highest-order part overflows (gets larger than |
1998 | a host char will hold) then the whole number has | |
1999 | overflowed. Record this and truncate the highest-order | |
2000 | part. */ | |
2001 | if (parts[TOTAL_PARTS - 1] >> HOST_BITS_PER_CHAR) | |
2002 | { | |
2003 | overflow = 1; | |
2004 | parts[TOTAL_PARTS - 1] &= (1 << HOST_BITS_PER_CHAR) - 1; | |
2005 | } | |
2006 | } | |
2007 | } | |
2008 | while (p < str + len); | |
91b0989e | 2009 | |
0e5921e8 ZW |
2010 | /* This can happen on input like `int i = 0x;' */ |
2011 | if (numdigits == 0) | |
2012 | ERROR ("numeric constant with no digits"); | |
91b0989e | 2013 | |
0e5921e8 ZW |
2014 | if (largest_digit >= base) |
2015 | ERROR ("numeric constant contains digits beyond the radix"); | |
e8bbfc4e | 2016 | |
0e5921e8 ZW |
2017 | if (floatflag != NOT_FLOAT) |
2018 | { | |
2019 | tree type; | |
2020 | int imag, fflag, lflag, conversion_errno; | |
2021 | REAL_VALUE_TYPE real; | |
2022 | struct pf_args args; | |
2023 | char *copy; | |
2024 | ||
2025 | if (base == 16 && floatflag != AFTER_EXPON) | |
2026 | ERROR ("hexadecimal floating constant has no exponent"); | |
2027 | ||
2028 | /* Read explicit exponent if any, and put it in tokenbuf. */ | |
2029 | if ((base == 10 && ((c == 'e') || (c == 'E'))) | |
2030 | || (base == 16 && (c == 'p' || c == 'P'))) | |
2031 | { | |
2032 | if (p < str + len) | |
2033 | c = *p++; | |
2034 | if (p < str + len && (c == '+' || c == '-')) | |
2035 | c = *p++; | |
2036 | /* Exponent is decimal, even if string is a hex float. */ | |
2037 | if (! ISDIGIT (c)) | |
2038 | ERROR ("floating constant exponent has no digits"); | |
2039 | while (p < str + len && ISDIGIT (c)) | |
2040 | c = *p++; | |
2041 | if (! ISDIGIT (c)) | |
2042 | p--; | |
2043 | } | |
56f48ce9 | 2044 | |
0e5921e8 ZW |
2045 | /* Copy the float constant now; we don't want any suffixes in the |
2046 | string passed to parse_float. */ | |
2047 | copy = alloca (p - str + 1); | |
2048 | memcpy (copy, str, p - str); | |
2049 | copy[p - str] = '\0'; | |
e8bbfc4e | 2050 | |
0e5921e8 ZW |
2051 | /* Now parse suffixes. */ |
2052 | fflag = lflag = imag = 0; | |
2053 | while (p < str + len) | |
2054 | switch (*p++) | |
e8bbfc4e | 2055 | { |
0e5921e8 ZW |
2056 | case 'f': case 'F': |
2057 | if (fflag) | |
2058 | ERROR ("more than one 'f' suffix on floating constant"); | |
2059 | else if (warn_traditional && !in_system_header) | |
2060 | warning ("traditional C rejects the 'f' suffix"); | |
e8bbfc4e | 2061 | |
0e5921e8 ZW |
2062 | fflag = 1; |
2063 | break; | |
e8bbfc4e | 2064 | |
0e5921e8 ZW |
2065 | case 'l': case 'L': |
2066 | if (lflag) | |
2067 | ERROR ("more than one 'l' suffix on floating constant"); | |
2068 | else if (warn_traditional && !in_system_header) | |
2069 | warning ("traditional C rejects the 'l' suffix"); | |
56f48ce9 | 2070 | |
0e5921e8 ZW |
2071 | lflag = 1; |
2072 | break; | |
e8bbfc4e | 2073 | |
0e5921e8 ZW |
2074 | case 'i': case 'I': |
2075 | case 'j': case 'J': | |
2076 | if (imag) | |
2077 | ERROR ("more than one 'i' or 'j' suffix on floating constant"); | |
2078 | else if (pedantic) | |
2079 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
2080 | imag = 1; | |
2081 | break; | |
e8bbfc4e | 2082 | |
0e5921e8 ZW |
2083 | default: |
2084 | ERROR ("invalid suffix on floating constant"); | |
e8bbfc4e RK |
2085 | } |
2086 | ||
0e5921e8 ZW |
2087 | /* Setup input for parse_float() */ |
2088 | args.str = copy; | |
2089 | args.fflag = fflag; | |
2090 | args.lflag = lflag; | |
2091 | args.base = base; | |
e8bbfc4e | 2092 | |
0e5921e8 ZW |
2093 | /* Convert string to a double, checking for overflow. */ |
2094 | if (do_float_handler (parse_float, (PTR) &args)) | |
2095 | { | |
2096 | /* Receive output from parse_float() */ | |
2097 | real = args.value; | |
2098 | } | |
2099 | else | |
2100 | /* We got an exception from parse_float() */ | |
2101 | ERROR ("floating constant out of range"); | |
e8bbfc4e | 2102 | |
0e5921e8 ZW |
2103 | /* Receive output from parse_float() */ |
2104 | conversion_errno = args.conversion_errno; | |
2105 | type = args.type; | |
2106 | ||
2107 | #ifdef ERANGE | |
2108 | /* ERANGE is also reported for underflow, | |
2109 | so test the value to distinguish overflow from that. */ | |
2110 | if (conversion_errno == ERANGE && !flag_traditional && pedantic | |
2111 | && (REAL_VALUES_LESS (dconst1, real) | |
2112 | || REAL_VALUES_LESS (real, dconstm1))) | |
2113 | warning ("floating point number exceeds range of 'double'"); | |
56f48ce9 | 2114 | #endif |
e8bbfc4e | 2115 | |
0e5921e8 ZW |
2116 | /* Create a node with determined type and value. */ |
2117 | if (imag) | |
2118 | value = build_complex (NULL_TREE, convert (type, integer_zero_node), | |
2119 | build_real (type, real)); | |
2120 | else | |
2121 | value = build_real (type, real); | |
2122 | } | |
2123 | else | |
2124 | { | |
2125 | tree trad_type, ansi_type, type; | |
2126 | HOST_WIDE_INT high, low; | |
2127 | int spec_unsigned = 0; | |
2128 | int spec_long = 0; | |
2129 | int spec_long_long = 0; | |
2130 | int spec_imag = 0; | |
2131 | int suffix_lu = 0; | |
2132 | int warn = 0, i; | |
2133 | ||
2134 | trad_type = ansi_type = type = NULL_TREE; | |
2135 | while (p < str + len) | |
2136 | { | |
2137 | c = *p++; | |
2138 | switch (c) | |
2139 | { | |
2140 | case 'u': case 'U': | |
2141 | if (spec_unsigned) | |
2142 | error ("two 'u' suffixes on integer constant"); | |
2143 | else if (warn_traditional && !in_system_header) | |
2144 | warning ("traditional C rejects the 'u' suffix"); | |
2145 | ||
2146 | spec_unsigned = 1; | |
2147 | if (spec_long) | |
2148 | suffix_lu = 1; | |
2149 | break; | |
e8bbfc4e | 2150 | |
0e5921e8 ZW |
2151 | case 'l': case 'L': |
2152 | if (spec_long) | |
2153 | { | |
2154 | if (spec_long_long) | |
2155 | error ("three 'l' suffixes on integer constant"); | |
2156 | else if (suffix_lu) | |
2157 | error ("'lul' is not a valid integer suffix"); | |
2158 | else if (c != spec_long) | |
2159 | error ("'Ll' and 'lL' are not valid integer suffixes"); | |
2160 | else if (pedantic && ! flag_isoc99 | |
2161 | && ! in_system_header && warn_long_long) | |
2162 | pedwarn ("ISO C89 forbids long long integer constants"); | |
2163 | spec_long_long = 1; | |
2164 | } | |
2165 | spec_long = c; | |
2166 | break; | |
56f48ce9 | 2167 | |
0e5921e8 ZW |
2168 | case 'i': case 'I': case 'j': case 'J': |
2169 | if (spec_imag) | |
2170 | error ("more than one 'i' or 'j' suffix on integer constant"); | |
2171 | else if (pedantic) | |
2172 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
2173 | spec_imag = 1; | |
2174 | break; | |
56f48ce9 | 2175 | |
0e5921e8 ZW |
2176 | default: |
2177 | ERROR ("invalid suffix on integer constant"); | |
2178 | } | |
2179 | } | |
56f48ce9 | 2180 | |
0e5921e8 ZW |
2181 | /* If the literal overflowed, pedwarn about it now. */ |
2182 | if (overflow) | |
2183 | { | |
2184 | warn = 1; | |
2185 | pedwarn ("integer constant is too large for this configuration of the compiler - truncated to %d bits", HOST_BITS_PER_WIDE_INT * 2); | |
2186 | } | |
e8bbfc4e | 2187 | |
0e5921e8 ZW |
2188 | /* This is simplified by the fact that our constant |
2189 | is always positive. */ | |
56f48ce9 | 2190 | |
0e5921e8 | 2191 | high = low = 0; |
e8bbfc4e | 2192 | |
0e5921e8 ZW |
2193 | for (i = 0; i < HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR; i++) |
2194 | { | |
2195 | high |= ((HOST_WIDE_INT) parts[i + (HOST_BITS_PER_WIDE_INT | |
2196 | / HOST_BITS_PER_CHAR)] | |
2197 | << (i * HOST_BITS_PER_CHAR)); | |
2198 | low |= (HOST_WIDE_INT) parts[i] << (i * HOST_BITS_PER_CHAR); | |
2199 | } | |
0468bc75 | 2200 | |
0e5921e8 ZW |
2201 | value = build_int_2 (low, high); |
2202 | TREE_TYPE (value) = long_long_unsigned_type_node; | |
e8bbfc4e | 2203 | |
0e5921e8 ZW |
2204 | /* If warn_traditional, calculate both the ISO type and the |
2205 | traditional type, then see if they disagree. | |
2206 | Otherwise, calculate only the type for the dialect in use. */ | |
2207 | if (warn_traditional || flag_traditional) | |
2208 | { | |
2209 | /* Calculate the traditional type. */ | |
2210 | /* Traditionally, any constant is signed; but if unsigned is | |
2211 | specified explicitly, obey that. Use the smallest size | |
2212 | with the right number of bits, except for one special | |
2213 | case with decimal constants. */ | |
2214 | if (! spec_long && base != 10 | |
2215 | && int_fits_type_p (value, unsigned_type_node)) | |
2216 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
2217 | /* A decimal constant must be long if it does not fit in | |
2218 | type int. I think this is independent of whether the | |
2219 | constant is signed. */ | |
2220 | else if (! spec_long && base == 10 | |
2221 | && int_fits_type_p (value, integer_type_node)) | |
2222 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
2223 | else if (! spec_long_long) | |
2224 | trad_type = (spec_unsigned | |
2225 | ? long_unsigned_type_node | |
2226 | : long_integer_type_node); | |
2227 | else if (int_fits_type_p (value, | |
2228 | spec_unsigned | |
2229 | ? long_long_unsigned_type_node | |
2230 | : long_long_integer_type_node)) | |
2231 | trad_type = (spec_unsigned | |
2232 | ? long_long_unsigned_type_node | |
2233 | : long_long_integer_type_node); | |
2234 | else | |
2235 | trad_type = (spec_unsigned | |
2236 | ? widest_unsigned_literal_type_node | |
2237 | : widest_integer_literal_type_node); | |
2238 | } | |
2239 | if (warn_traditional || ! flag_traditional) | |
2240 | { | |
2241 | /* Calculate the ISO type. */ | |
2242 | if (! spec_long && ! spec_unsigned | |
2243 | && int_fits_type_p (value, integer_type_node)) | |
2244 | ansi_type = integer_type_node; | |
2245 | else if (! spec_long && (base != 10 || spec_unsigned) | |
2246 | && int_fits_type_p (value, unsigned_type_node)) | |
2247 | ansi_type = unsigned_type_node; | |
2248 | else if (! spec_unsigned && !spec_long_long | |
2249 | && int_fits_type_p (value, long_integer_type_node)) | |
2250 | ansi_type = long_integer_type_node; | |
2251 | else if (! spec_long_long | |
2252 | && int_fits_type_p (value, long_unsigned_type_node)) | |
2253 | ansi_type = long_unsigned_type_node; | |
2254 | else if (! spec_unsigned | |
2255 | && int_fits_type_p (value, long_long_integer_type_node)) | |
2256 | ansi_type = long_long_integer_type_node; | |
2257 | else if (int_fits_type_p (value, long_long_unsigned_type_node)) | |
2258 | ansi_type = long_long_unsigned_type_node; | |
2259 | else if (! spec_unsigned | |
2260 | && int_fits_type_p (value, widest_integer_literal_type_node)) | |
2261 | ansi_type = widest_integer_literal_type_node; | |
2262 | else | |
2263 | ansi_type = widest_unsigned_literal_type_node; | |
2264 | } | |
e8bbfc4e | 2265 | |
0e5921e8 | 2266 | type = flag_traditional ? trad_type : ansi_type; |
e8bbfc4e | 2267 | |
0e5921e8 ZW |
2268 | /* We assume that constants specified in a non-decimal |
2269 | base are bit patterns, and that the programmer really | |
2270 | meant what they wrote. */ | |
2271 | if (warn_traditional && !in_system_header | |
2272 | && base == 10 && trad_type != ansi_type) | |
2273 | { | |
2274 | if (TYPE_PRECISION (trad_type) != TYPE_PRECISION (ansi_type)) | |
2275 | warning ("width of integer constant changes with -traditional"); | |
2276 | else if (TREE_UNSIGNED (trad_type) != TREE_UNSIGNED (ansi_type)) | |
2277 | warning ("integer constant is unsigned in ISO C, signed with -traditional"); | |
2278 | else | |
2279 | warning ("width of integer constant may change on other systems with -traditional"); | |
2280 | } | |
e8bbfc4e | 2281 | |
0e5921e8 ZW |
2282 | if (pedantic && !flag_traditional && !spec_long_long && !warn |
2283 | && (TYPE_PRECISION (long_integer_type_node) < TYPE_PRECISION (type))) | |
2284 | { | |
2285 | warn = 1; | |
2286 | pedwarn ("integer constant larger than the maximum value of an unsigned long int"); | |
2287 | } | |
e8bbfc4e | 2288 | |
0e5921e8 ZW |
2289 | if (base == 10 && ! spec_unsigned && TREE_UNSIGNED (type)) |
2290 | warning ("decimal constant is so large that it is unsigned"); | |
e8bbfc4e | 2291 | |
0e5921e8 ZW |
2292 | if (spec_imag) |
2293 | { | |
2294 | if (TYPE_PRECISION (type) | |
2295 | <= TYPE_PRECISION (integer_type_node)) | |
2296 | value = build_complex (NULL_TREE, integer_zero_node, | |
2297 | convert (integer_type_node, value)); | |
2298 | else | |
2299 | ERROR ("complex integer constant is too wide for 'complex int'"); | |
2300 | } | |
2301 | else if (flag_traditional && !int_fits_type_p (value, type)) | |
2302 | /* The traditional constant 0x80000000 is signed | |
2303 | but doesn't fit in the range of int. | |
2304 | This will change it to -0x80000000, which does fit. */ | |
2305 | { | |
2306 | TREE_TYPE (value) = unsigned_type (type); | |
2307 | value = convert (type, value); | |
2308 | TREE_OVERFLOW (value) = TREE_CONSTANT_OVERFLOW (value) = 0; | |
2309 | } | |
2310 | else | |
2311 | TREE_TYPE (value) = type; | |
e8bbfc4e | 2312 | |
0e5921e8 ZW |
2313 | /* If it's still an integer (not a complex), and it doesn't |
2314 | fit in the type we choose for it, then pedwarn. */ | |
fbb18613 | 2315 | |
0e5921e8 ZW |
2316 | if (! warn |
2317 | && TREE_CODE (TREE_TYPE (value)) == INTEGER_TYPE | |
2318 | && ! int_fits_type_p (value, TREE_TYPE (value))) | |
2319 | pedwarn ("integer constant is larger than the maximum value for its type"); | |
2320 | } | |
fbb18613 | 2321 | |
0e5921e8 ZW |
2322 | if (p < str + len) |
2323 | error ("missing white space after number '%.*s'", (int) (p - str), str); | |
e8bbfc4e | 2324 | |
0e5921e8 | 2325 | return value; |
e8bbfc4e | 2326 | |
0e5921e8 ZW |
2327 | syntax_error: |
2328 | return integer_zero_node; | |
2329 | } | |
e8bbfc4e | 2330 | |
0e5921e8 ZW |
2331 | static tree |
2332 | lex_string (str, len, wide) | |
2333 | const char *str; | |
2334 | unsigned int len; | |
2335 | int wide; | |
2336 | { | |
2337 | tree value; | |
2338 | char *buf = alloca ((len + 1) * (wide ? WCHAR_BYTES : 1)); | |
2339 | char *q = buf; | |
2340 | const char *p = str, *limit = str + len; | |
2341 | unsigned int c; | |
2342 | unsigned width = wide ? WCHAR_TYPE_SIZE | |
2343 | : TYPE_PRECISION (char_type_node); | |
e9a25f70 | 2344 | |
0e5921e8 ZW |
2345 | #ifdef MULTIBYTE_CHARS |
2346 | /* Reset multibyte conversion state. */ | |
2347 | (void) local_mbtowc (NULL_PTR, NULL_PTR, 0); | |
2348 | #endif | |
e9a25f70 | 2349 | |
0e5921e8 ZW |
2350 | while (p < limit) |
2351 | { | |
2352 | #ifdef MULTIBYTE_CHARS | |
2353 | wchar_t wc; | |
2354 | int char_len; | |
2355 | ||
2356 | char_len = local_mbtowc (&wc, p, limit - p); | |
2357 | if (char_len == -1) | |
2358 | { | |
2359 | warning ("Ignoring invalid multibyte character"); | |
2360 | char_len = 1; | |
2361 | c = *p++; | |
2362 | } | |
2363 | else | |
2364 | { | |
2365 | p += char_len; | |
2366 | c = wc; | |
2367 | } | |
2368 | #else | |
2369 | c = *p++; | |
2370 | #endif | |
2371 | ||
2372 | if (c == '\\' && !ignore_escape_flag) | |
2373 | { | |
2374 | p = readescape (p, limit, &c); | |
2375 | if (width < HOST_BITS_PER_INT | |
2376 | && (unsigned) c >= ((unsigned)1 << width)) | |
2377 | pedwarn ("escape sequence out of range for character"); | |
2378 | } | |
2379 | ||
2380 | /* Add this single character into the buffer either as a wchar_t | |
2381 | or as a single byte. */ | |
2382 | if (wide) | |
2383 | { | |
2384 | unsigned charwidth = TYPE_PRECISION (char_type_node); | |
5c80f6e6 | 2385 | unsigned bytemask = (1 << charwidth) - 1; |
0e5921e8 ZW |
2386 | int byte; |
2387 | ||
2388 | for (byte = 0; byte < WCHAR_BYTES; ++byte) | |
2389 | { | |
2390 | int n; | |
2391 | if (byte >= (int) sizeof (c)) | |
2392 | n = 0; | |
2393 | else | |
2394 | n = (c >> (byte * charwidth)) & bytemask; | |
2395 | if (BYTES_BIG_ENDIAN) | |
2396 | q[WCHAR_BYTES - byte - 1] = n; | |
2397 | else | |
2398 | q[byte] = n; | |
2399 | } | |
2400 | q += WCHAR_BYTES; | |
2401 | } | |
2402 | else | |
2403 | { | |
2404 | *q++ = c; | |
2405 | } | |
e8bbfc4e RK |
2406 | } |
2407 | ||
0e5921e8 ZW |
2408 | /* Terminate the string value, either with a single byte zero |
2409 | or with a wide zero. */ | |
e8bbfc4e | 2410 | |
0e5921e8 ZW |
2411 | if (wide) |
2412 | { | |
2413 | memset (q, 0, WCHAR_BYTES); | |
2414 | q += WCHAR_BYTES; | |
2415 | } | |
2416 | else | |
2417 | { | |
2418 | *q++ = '\0'; | |
2419 | } | |
2420 | ||
2421 | value = build_string (q - buf, buf); | |
2422 | ||
2423 | if (wide) | |
2424 | TREE_TYPE (value) = wchar_array_type_node; | |
2425 | else | |
2426 | TREE_TYPE (value) = char_array_type_node; | |
e8bbfc4e RK |
2427 | return value; |
2428 | } | |
2429 | ||
0e5921e8 ZW |
2430 | static tree |
2431 | lex_charconst (str, len, wide) | |
2432 | const char *str; | |
2433 | unsigned int len; | |
2434 | int wide; | |
e8bbfc4e | 2435 | { |
0e5921e8 ZW |
2436 | const char *limit = str + len; |
2437 | int result = 0; | |
2438 | int num_chars = 0; | |
2439 | int chars_seen = 0; | |
2440 | unsigned width = TYPE_PRECISION (char_type_node); | |
2441 | int max_chars; | |
2442 | unsigned int c; | |
2443 | tree value; | |
2444 | ||
2445 | #ifdef MULTIBYTE_CHARS | |
2446 | int longest_char = local_mb_cur_max (); | |
2447 | (void) local_mbtowc (NULL_PTR, NULL_PTR, 0); | |
2448 | #endif | |
2449 | ||
2450 | max_chars = TYPE_PRECISION (integer_type_node) / width; | |
2451 | if (wide) | |
2452 | width = WCHAR_TYPE_SIZE; | |
2453 | ||
2454 | while (str < limit) | |
2455 | { | |
2456 | #ifdef MULTIBYTE_CHARS | |
2457 | wchar_t wc; | |
2458 | int char_len; | |
2459 | ||
2460 | char_len = local_mbtowc (&wc, str, limit - str); | |
2461 | if (char_len == -1) | |
2462 | { | |
2463 | warning ("Ignoring invalid multibyte character"); | |
2464 | char_len = 1; | |
2465 | c = *str++; | |
2466 | } | |
2467 | else | |
2468 | { | |
2469 | p += char_len; | |
2470 | c = wc; | |
2471 | } | |
e8bbfc4e | 2472 | #else |
0e5921e8 ZW |
2473 | c = *str++; |
2474 | #endif | |
2475 | ||
2476 | ++chars_seen; | |
2477 | if (c == '\\') | |
2478 | { | |
2479 | str = readescape (str, limit, &c); | |
2480 | if (width < HOST_BITS_PER_INT | |
2481 | && (unsigned) c >= ((unsigned)1 << width)) | |
2482 | pedwarn ("escape sequence out of range for character"); | |
2483 | } | |
2484 | #ifdef MAP_CHARACTER | |
2485 | if (ISPRINT (c)) | |
2486 | c = MAP_CHARACTER (c); | |
e8bbfc4e | 2487 | #endif |
0e5921e8 ZW |
2488 | |
2489 | /* Merge character into result; ignore excess chars. */ | |
2490 | num_chars += (width / TYPE_PRECISION (char_type_node)); | |
2491 | if (num_chars < max_chars + 1) | |
2492 | { | |
2493 | if (width < HOST_BITS_PER_INT) | |
2494 | result = (result << width) | (c & ((1 << width) - 1)); | |
2495 | else | |
2496 | result = c; | |
2497 | } | |
2498 | } | |
2499 | ||
2500 | if (chars_seen == 0) | |
2501 | error ("empty character constant"); | |
2502 | else if (num_chars > max_chars) | |
2503 | { | |
2504 | num_chars = max_chars; | |
2505 | error ("character constant too long"); | |
2506 | } | |
2507 | else if (chars_seen != 1 && ! flag_traditional && warn_multichar) | |
2508 | warning ("multi-character character constant"); | |
2509 | ||
2510 | /* If char type is signed, sign-extend the constant. */ | |
2511 | if (! wide) | |
2512 | { | |
2513 | int num_bits = num_chars * width; | |
2514 | if (num_bits == 0) | |
2515 | /* We already got an error; avoid invalid shift. */ | |
2516 | value = build_int_2 (0, 0); | |
2517 | else if (TREE_UNSIGNED (char_type_node) | |
2518 | || ((result >> (num_bits - 1)) & 1) == 0) | |
2519 | value = build_int_2 (result & (~(unsigned HOST_WIDE_INT) 0 | |
2520 | >> (HOST_BITS_PER_WIDE_INT - num_bits)), | |
2521 | 0); | |
2522 | else | |
2523 | value = build_int_2 (result | ~(~(unsigned HOST_WIDE_INT) 0 | |
2524 | >> (HOST_BITS_PER_WIDE_INT - num_bits)), | |
2525 | -1); | |
2526 | /* In C, a character constant has type 'int'; in C++, 'char'. */ | |
2527 | if (chars_seen <= 1 && c_language == clk_cplusplus) | |
2528 | TREE_TYPE (value) = char_type_node; | |
2529 | else | |
2530 | TREE_TYPE (value) = integer_type_node; | |
2531 | } | |
2532 | else | |
2533 | { | |
2534 | value = build_int_2 (result, 0); | |
2535 | TREE_TYPE (value) = wchar_type_node; | |
2536 | } | |
2537 | ||
2538 | return value; | |
e8bbfc4e | 2539 | } |