]>
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 | 4 | |
1322177d | 5 | This file is part of GCC. |
e8bbfc4e | 6 | |
1322177d LB |
7 | GCC is free software; you can redistribute it and/or modify it under |
8 | the terms of the GNU General Public License as published by the Free | |
9 | Software Foundation; either version 2, or (at your option) any later | |
10 | version. | |
e8bbfc4e | 11 | |
1322177d LB |
12 | GCC is distributed in the hope that it will be useful, but WITHOUT ANY |
13 | WARRANTY; without even the implied warranty of MERCHANTABILITY or | |
14 | FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License | |
15 | for more details. | |
e8bbfc4e RK |
16 | |
17 | You should have received a copy of the GNU General Public License | |
1322177d LB |
18 | along with GCC; see the file COPYING. If not, write to the Free |
19 | Software Foundation, 59 Temple Place - Suite 330, Boston, MA | |
20 | 02111-1307, USA. */ | |
e8bbfc4e | 21 | |
e9a25f70 | 22 | #include "config.h" |
670ee920 | 23 | #include "system.h" |
e8bbfc4e | 24 | |
e8bbfc4e RK |
25 | #include "rtl.h" |
26 | #include "tree.h" | |
eb3aaa5b | 27 | #include "expr.h" |
e8bbfc4e | 28 | #include "input.h" |
d6f4ec51 | 29 | #include "output.h" |
e8bbfc4e RK |
30 | #include "c-lex.h" |
31 | #include "c-tree.h" | |
52dabb6c | 32 | #include "c-common.h" |
e8bbfc4e | 33 | #include "flags.h" |
0e5921e8 | 34 | #include "timevar.h" |
8b97c5f8 | 35 | #include "cpplib.h" |
3d6f7931 | 36 | #include "c-pragma.h" |
5f6da302 | 37 | #include "toplev.h" |
ab87f8c8 | 38 | #include "intl.h" |
7bdb32b9 | 39 | #include "tm_p.h" |
0e5921e8 | 40 | #include "splay-tree.h" |
7f905405 | 41 | #include "debug.h" |
ab87f8c8 | 42 | |
e8bbfc4e | 43 | #ifdef MULTIBYTE_CHARS |
56f48ce9 | 44 | #include "mbchar.h" |
e8bbfc4e | 45 | #include <locale.h> |
56f48ce9 | 46 | #endif /* MULTIBYTE_CHARS */ |
c5c76735 JL |
47 | #ifndef GET_ENVIRONMENT |
48 | #define GET_ENVIRONMENT(ENV_VALUE,ENV_NAME) ((ENV_VALUE) = getenv (ENV_NAME)) | |
49 | #endif | |
e8bbfc4e | 50 | |
67821e3a | 51 | /* The current line map. */ |
47d89cf3 | 52 | static const struct line_map *map; |
67821e3a | 53 | |
97293897 NB |
54 | /* The line used to refresh the lineno global variable after each token. */ |
55 | static unsigned int src_lineno; | |
56 | ||
0e5921e8 ZW |
57 | /* We may keep statistics about how long which files took to compile. */ |
58 | static int header_time, body_time; | |
59 | static splay_tree file_info_tree; | |
3ab6dd7c | 60 | |
e8bbfc4e RK |
61 | /* File used for outputting assembler code. */ |
62 | extern FILE *asm_out_file; | |
63 | ||
12a39b12 JM |
64 | #undef WCHAR_TYPE_SIZE |
65 | #define WCHAR_TYPE_SIZE TYPE_PRECISION (wchar_type_node) | |
e8bbfc4e RK |
66 | |
67 | /* Number of bytes in a wide character. */ | |
68 | #define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT) | |
69 | ||
ec5c56db | 70 | int indent_level; /* Number of { minus number of }. */ |
0e5921e8 ZW |
71 | int pending_lang_change; /* If we need to switch languages - C++ only */ |
72 | int c_header_level; /* depth in C headers - C++ only */ | |
fbb18613 JM |
73 | |
74 | /* Nonzero tells yylex to ignore \ in string constants. */ | |
75 | static int ignore_escape_flag; | |
e9a25f70 | 76 | |
0e5921e8 | 77 | static tree lex_number PARAMS ((const char *, unsigned int)); |
62ae2529 JJ |
78 | static tree lex_string PARAMS ((const unsigned char *, unsigned int, |
79 | int)); | |
c8a96070 | 80 | static tree lex_charconst PARAMS ((const cpp_token *)); |
0e5921e8 ZW |
81 | static void update_header_times PARAMS ((const char *)); |
82 | static int dump_one_header PARAMS ((splay_tree_node, void *)); | |
97293897 | 83 | static void cb_line_change PARAMS ((cpp_reader *, const cpp_token *, int)); |
8bbbef34 NB |
84 | static void cb_ident PARAMS ((cpp_reader *, unsigned int, |
85 | const cpp_string *)); | |
47d89cf3 | 86 | static void cb_file_change PARAMS ((cpp_reader *, const struct line_map *)); |
8bbbef34 NB |
87 | static void cb_def_pragma PARAMS ((cpp_reader *, unsigned int)); |
88 | static void cb_define PARAMS ((cpp_reader *, unsigned int, | |
89 | cpp_hashnode *)); | |
90 | static void cb_undef PARAMS ((cpp_reader *, unsigned int, | |
91 | cpp_hashnode *)); | |
e31c7eec | 92 | \f |
3b304f5b | 93 | const char * |
0e5921e8 | 94 | init_c_lex (filename) |
3b304f5b | 95 | const char *filename; |
e3d1fd32 | 96 | { |
b61c5ed0 | 97 | struct cpp_callbacks *cb; |
0e5921e8 ZW |
98 | struct c_fileinfo *toplevel; |
99 | ||
f5e99456 | 100 | /* Set up filename timing. Must happen before cpp_read_main_file. */ |
0e5921e8 ZW |
101 | file_info_tree = splay_tree_new ((splay_tree_compare_fn)strcmp, |
102 | 0, | |
103 | (splay_tree_delete_value_fn)free); | |
a8a05998 | 104 | toplevel = get_fileinfo ("<top level>"); |
0e5921e8 ZW |
105 | if (flag_detailed_statistics) |
106 | { | |
107 | header_time = 0; | |
108 | body_time = get_run_time (); | |
109 | toplevel->time = body_time; | |
110 | } | |
111 | ||
112 | #ifdef MULTIBYTE_CHARS | |
113 | /* Change to the native locale for multibyte conversions. */ | |
114 | setlocale (LC_CTYPE, ""); | |
115 | GET_ENVIRONMENT (literal_codeset, "LANG"); | |
116 | #endif | |
117 | ||
b61c5ed0 NB |
118 | cb = cpp_get_callbacks (parse_in); |
119 | ||
97293897 | 120 | cb->line_change = cb_line_change; |
b61c5ed0 NB |
121 | cb->ident = cb_ident; |
122 | cb->file_change = cb_file_change; | |
123 | cb->def_pragma = cb_def_pragma; | |
0e5921e8 | 124 | |
65289a3a NB |
125 | /* Set the debug callbacks if we can use them. */ |
126 | if (debug_info_level == DINFO_LEVEL_VERBOSE | |
7a0c8d71 DR |
127 | && (write_symbols == DWARF_DEBUG || write_symbols == DWARF2_DEBUG |
128 | || write_symbols == VMS_AND_DWARF2_DEBUG)) | |
65289a3a | 129 | { |
b61c5ed0 NB |
130 | cb->define = cb_define; |
131 | cb->undef = cb_undef; | |
65289a3a NB |
132 | } |
133 | ||
851ef036 NB |
134 | /* Start it at 0. */ |
135 | lineno = 0; | |
b12da25e | 136 | |
c65fd410 | 137 | if (filename == NULL || !strcmp (filename, "-")) |
f5e99456 NB |
138 | filename = ""; |
139 | ||
140 | return cpp_read_main_file (parse_in, filename, ident_hash); | |
e3d1fd32 PB |
141 | } |
142 | ||
3aac38d7 | 143 | /* A thin wrapper around the real parser that initializes the |
bebc7e8b ZW |
144 | integrated preprocessor after debug output has been initialized. |
145 | Also, make sure the start_source_file debug hook gets called for | |
146 | the primary source file. */ | |
3aac38d7 | 147 | |
52dabb6c | 148 | void |
ff45c01e NB |
149 | c_common_parse_file (set_yydebug) |
150 | int set_yydebug ATTRIBUTE_UNUSED; | |
3aac38d7 | 151 | { |
ff45c01e NB |
152 | #if YYDEBUG != 0 |
153 | yydebug = set_yydebug; | |
154 | #else | |
155 | warning ("YYDEBUG not defined"); | |
156 | #endif | |
157 | ||
bebc7e8b | 158 | (*debug_hooks->start_source_file) (lineno, input_filename); |
f5e99456 | 159 | cpp_finish_options (parse_in); |
3aac38d7 | 160 | |
52dabb6c | 161 | yyparse (); |
3aac38d7 RH |
162 | } |
163 | ||
0e5921e8 ZW |
164 | struct c_fileinfo * |
165 | get_fileinfo (name) | |
166 | const char *name; | |
e3d1fd32 | 167 | { |
0e5921e8 ZW |
168 | splay_tree_node n; |
169 | struct c_fileinfo *fi; | |
170 | ||
171 | n = splay_tree_lookup (file_info_tree, (splay_tree_key) name); | |
172 | if (n) | |
173 | return (struct c_fileinfo *) n->value; | |
174 | ||
175 | fi = (struct c_fileinfo *) xmalloc (sizeof (struct c_fileinfo)); | |
176 | fi->time = 0; | |
177 | fi->interface_only = 0; | |
178 | fi->interface_unknown = 1; | |
179 | splay_tree_insert (file_info_tree, (splay_tree_key) name, | |
180 | (splay_tree_value) fi); | |
181 | return fi; | |
e56e519d | 182 | } |
e3d1fd32 | 183 | |
0e5921e8 ZW |
184 | static void |
185 | update_header_times (name) | |
186 | const char *name; | |
e8bbfc4e | 187 | { |
0e5921e8 ZW |
188 | /* Changing files again. This means currently collected time |
189 | is charged against header time, and body time starts back at 0. */ | |
190 | if (flag_detailed_statistics) | |
e8bbfc4e | 191 | { |
0e5921e8 ZW |
192 | int this_time = get_run_time (); |
193 | struct c_fileinfo *file = get_fileinfo (name); | |
194 | header_time += this_time - body_time; | |
195 | file->time += this_time - body_time; | |
196 | body_time = this_time; | |
e8bbfc4e RK |
197 | } |
198 | } | |
199 | ||
0e5921e8 ZW |
200 | static int |
201 | dump_one_header (n, dummy) | |
202 | splay_tree_node n; | |
203 | void *dummy ATTRIBUTE_UNUSED; | |
e8bbfc4e | 204 | { |
0e5921e8 ZW |
205 | print_time ((const char *) n->key, |
206 | ((struct c_fileinfo *) n->value)->time); | |
207 | return 0; | |
e8bbfc4e | 208 | } |
e8bbfc4e RK |
209 | |
210 | void | |
0e5921e8 | 211 | dump_time_statistics () |
e8bbfc4e | 212 | { |
0e5921e8 ZW |
213 | struct c_fileinfo *file = get_fileinfo (input_filename); |
214 | int this_time = get_run_time (); | |
215 | file->time += this_time - body_time; | |
216 | ||
217 | fprintf (stderr, "\n******\n"); | |
218 | print_time ("header files (total)", header_time); | |
219 | print_time ("main file (total)", this_time - body_time); | |
220 | fprintf (stderr, "ratio = %g : 1\n", | |
221 | (double)header_time / (double)(this_time - body_time)); | |
222 | fprintf (stderr, "\n******\n"); | |
223 | ||
224 | splay_tree_foreach (file_info_tree, dump_one_header, 0); | |
e8bbfc4e | 225 | } |
a6124a42 | 226 | |
27e2564a NB |
227 | /* Not yet handled: #pragma, #define, #undef. |
228 | No need to deal with linemarkers under normal conditions. */ | |
229 | ||
0e5921e8 | 230 | static void |
8bbbef34 | 231 | cb_ident (pfile, line, str) |
27e2564a | 232 | cpp_reader *pfile ATTRIBUTE_UNUSED; |
8bbbef34 | 233 | unsigned int line ATTRIBUTE_UNUSED; |
f5720527 | 234 | const cpp_string *str ATTRIBUTE_UNUSED; |
0e5921e8 | 235 | { |
0e5921e8 | 236 | #ifdef ASM_OUTPUT_IDENT |
27e2564a | 237 | if (! flag_no_ident) |
0e5921e8 | 238 | { |
27e2564a | 239 | /* Convert escapes in the string. */ |
62ae2529 | 240 | tree value = lex_string (str->text, str->len, 0); |
27e2564a | 241 | ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (value)); |
0e5921e8 ZW |
242 | } |
243 | #endif | |
27e2564a NB |
244 | } |
245 | ||
97293897 NB |
246 | /* Called at the start of every non-empty line. TOKEN is the first |
247 | lexed token on the line. Used for diagnostic line numbers. */ | |
248 | static void | |
249 | cb_line_change (pfile, token, parsing_args) | |
250 | cpp_reader *pfile ATTRIBUTE_UNUSED; | |
251 | const cpp_token *token; | |
252 | int parsing_args ATTRIBUTE_UNUSED; | |
253 | { | |
254 | src_lineno = SOURCE_LINE (map, token->line); | |
255 | } | |
256 | ||
27e2564a | 257 | static void |
47d89cf3 | 258 | cb_file_change (pfile, new_map) |
27e2564a | 259 | cpp_reader *pfile ATTRIBUTE_UNUSED; |
47d89cf3 | 260 | const struct line_map *new_map; |
27e2564a | 261 | { |
47d89cf3 | 262 | unsigned int to_line = SOURCE_LINE (new_map, new_map->to_line); |
d82fc108 | 263 | |
47d89cf3 | 264 | if (new_map->reason == LC_ENTER) |
fbb18613 | 265 | { |
5bea1ccf JM |
266 | /* Don't stack the main buffer on the input stack; |
267 | we already did in compile_file. */ | |
47d89cf3 NB |
268 | if (map == NULL) |
269 | main_input_filename = new_map->to_file; | |
d82fc108 | 270 | else |
0e5921e8 | 271 | { |
f78ce0b7 JB |
272 | int included_at = SOURCE_LINE (new_map - 1, new_map->from_line - 1); |
273 | ||
274 | lineno = included_at; | |
47d89cf3 | 275 | push_srcloc (new_map->to_file, 1); |
27e2564a | 276 | input_file_stack->indent_level = indent_level; |
f78ce0b7 | 277 | (*debug_hooks->start_source_file) (included_at, new_map->to_file); |
27e2564a NB |
278 | #ifndef NO_IMPLICIT_EXTERN_C |
279 | if (c_header_level) | |
280 | ++c_header_level; | |
47d89cf3 | 281 | else if (new_map->sysp == 2) |
27e2564a NB |
282 | { |
283 | c_header_level = 1; | |
284 | ++pending_lang_change; | |
285 | } | |
0e5921e8 | 286 | #endif |
27e2564a | 287 | } |
fbb18613 | 288 | } |
47d89cf3 | 289 | else if (new_map->reason == LC_LEAVE) |
fbb18613 | 290 | { |
0e5921e8 | 291 | #ifndef NO_IMPLICIT_EXTERN_C |
47d89cf3 NB |
292 | if (c_header_level && --c_header_level == 0) |
293 | { | |
294 | if (new_map->sysp == 2) | |
295 | warning ("badly nested C headers from preprocessor"); | |
296 | --pending_lang_change; | |
297 | } | |
0e5921e8 ZW |
298 | #endif |
299 | #if 0 | |
47d89cf3 NB |
300 | if (indent_level != input_file_stack->indent_level) |
301 | { | |
302 | warning_with_file_and_line | |
303 | (input_filename, lineno, | |
1f978f5f | 304 | "this file contains more '%c's than '%c's", |
47d89cf3 NB |
305 | indent_level > input_file_stack->indent_level ? '{' : '}', |
306 | indent_level > input_file_stack->indent_level ? '}' : '{'); | |
0468bc75 | 307 | } |
47d89cf3 NB |
308 | #endif |
309 | pop_srcloc (); | |
310 | ||
311 | (*debug_hooks->end_source_file) (to_line); | |
e8bbfc4e | 312 | } |
fbb18613 | 313 | |
47d89cf3 NB |
314 | update_header_times (new_map->to_file); |
315 | in_system_header = new_map->sysp != 0; | |
316 | input_filename = new_map->to_file; | |
317 | lineno = to_line; | |
318 | map = new_map; | |
0e5921e8 | 319 | |
0e5921e8 ZW |
320 | /* Hook for C++. */ |
321 | extract_interface_info (); | |
322 | } | |
8b97c5f8 ZW |
323 | |
324 | static void | |
8bbbef34 | 325 | cb_def_pragma (pfile, line) |
8b97c5f8 | 326 | cpp_reader *pfile; |
67821e3a | 327 | unsigned int line; |
8b97c5f8 ZW |
328 | { |
329 | /* Issue a warning message if we have been asked to do so. Ignore | |
330 | unknown pragmas in system headers unless an explicit | |
ec5c56db | 331 | -Wunknown-pragmas has been given. */ |
8b97c5f8 ZW |
332 | if (warn_unknown_pragmas > in_system_header) |
333 | { | |
23356f93 | 334 | const unsigned char *space, *name = 0; |
4ed5bcfb | 335 | const cpp_token *s; |
23356f93 | 336 | |
4ed5bcfb NB |
337 | s = cpp_get_token (pfile); |
338 | space = cpp_token_as_text (pfile, s); | |
339 | s = cpp_get_token (pfile); | |
340 | if (s->type == CPP_NAME) | |
341 | name = cpp_token_as_text (pfile, s); | |
8b97c5f8 | 342 | |
67821e3a | 343 | lineno = SOURCE_LINE (map, line); |
8b97c5f8 ZW |
344 | if (name) |
345 | warning ("ignoring #pragma %s %s", space, name); | |
346 | else | |
347 | warning ("ignoring #pragma %s", space); | |
348 | } | |
349 | } | |
0e5921e8 | 350 | |
65289a3a NB |
351 | /* #define callback for DWARF and DWARF2 debug info. */ |
352 | static void | |
8bbbef34 | 353 | cb_define (pfile, line, node) |
65289a3a | 354 | cpp_reader *pfile; |
67821e3a | 355 | unsigned int line; |
65289a3a NB |
356 | cpp_hashnode *node; |
357 | { | |
67821e3a | 358 | (*debug_hooks->define) (SOURCE_LINE (map, line), |
7f905405 | 359 | (const char *) cpp_macro_definition (pfile, node)); |
65289a3a NB |
360 | } |
361 | ||
362 | /* #undef callback for DWARF and DWARF2 debug info. */ | |
363 | static void | |
8bbbef34 | 364 | cb_undef (pfile, line, node) |
67821e3a NB |
365 | cpp_reader *pfile ATTRIBUTE_UNUSED; |
366 | unsigned int line; | |
65289a3a NB |
367 | cpp_hashnode *node; |
368 | { | |
67821e3a | 369 | (*debug_hooks->undef) (SOURCE_LINE (map, line), |
7f905405 | 370 | (const char *) NODE_NAME (node)); |
65289a3a NB |
371 | } |
372 | ||
0e5921e8 ZW |
373 | #if 0 /* not yet */ |
374 | /* Returns nonzero if C is a universal-character-name. Give an error if it | |
375 | is not one which may appear in an identifier, as per [extendid]. | |
376 | ||
377 | Note that extended character support in identifiers has not yet been | |
378 | implemented. It is my personal opinion that this is not a desirable | |
379 | feature. Portable code cannot count on support for more than the basic | |
380 | identifier character set. */ | |
381 | ||
382 | static inline int | |
383 | is_extended_char (c) | |
384 | int c; | |
385 | { | |
386 | #ifdef TARGET_EBCDIC | |
387 | return 0; | |
388 | #else | |
389 | /* ASCII. */ | |
390 | if (c < 0x7f) | |
391 | return 0; | |
392 | ||
393 | /* None of the valid chars are outside the Basic Multilingual Plane (the | |
394 | low 16 bits). */ | |
395 | if (c > 0xffff) | |
396 | { | |
397 | error ("universal-character-name '\\U%08x' not valid in identifier", c); | |
398 | return 1; | |
399 | } | |
400 | ||
401 | /* Latin */ | |
402 | if ((c >= 0x00c0 && c <= 0x00d6) | |
403 | || (c >= 0x00d8 && c <= 0x00f6) | |
404 | || (c >= 0x00f8 && c <= 0x01f5) | |
405 | || (c >= 0x01fa && c <= 0x0217) | |
406 | || (c >= 0x0250 && c <= 0x02a8) | |
407 | || (c >= 0x1e00 && c <= 0x1e9a) | |
408 | || (c >= 0x1ea0 && c <= 0x1ef9)) | |
409 | return 1; | |
410 | ||
411 | /* Greek */ | |
412 | if ((c == 0x0384) | |
413 | || (c >= 0x0388 && c <= 0x038a) | |
414 | || (c == 0x038c) | |
415 | || (c >= 0x038e && c <= 0x03a1) | |
416 | || (c >= 0x03a3 && c <= 0x03ce) | |
417 | || (c >= 0x03d0 && c <= 0x03d6) | |
418 | || (c == 0x03da) | |
419 | || (c == 0x03dc) | |
420 | || (c == 0x03de) | |
421 | || (c == 0x03e0) | |
422 | || (c >= 0x03e2 && c <= 0x03f3) | |
423 | || (c >= 0x1f00 && c <= 0x1f15) | |
424 | || (c >= 0x1f18 && c <= 0x1f1d) | |
425 | || (c >= 0x1f20 && c <= 0x1f45) | |
426 | || (c >= 0x1f48 && c <= 0x1f4d) | |
427 | || (c >= 0x1f50 && c <= 0x1f57) | |
428 | || (c == 0x1f59) | |
429 | || (c == 0x1f5b) | |
430 | || (c == 0x1f5d) | |
431 | || (c >= 0x1f5f && c <= 0x1f7d) | |
432 | || (c >= 0x1f80 && c <= 0x1fb4) | |
433 | || (c >= 0x1fb6 && c <= 0x1fbc) | |
434 | || (c >= 0x1fc2 && c <= 0x1fc4) | |
435 | || (c >= 0x1fc6 && c <= 0x1fcc) | |
436 | || (c >= 0x1fd0 && c <= 0x1fd3) | |
437 | || (c >= 0x1fd6 && c <= 0x1fdb) | |
438 | || (c >= 0x1fe0 && c <= 0x1fec) | |
439 | || (c >= 0x1ff2 && c <= 0x1ff4) | |
440 | || (c >= 0x1ff6 && c <= 0x1ffc)) | |
441 | return 1; | |
442 | ||
443 | /* Cyrillic */ | |
444 | if ((c >= 0x0401 && c <= 0x040d) | |
445 | || (c >= 0x040f && c <= 0x044f) | |
446 | || (c >= 0x0451 && c <= 0x045c) | |
447 | || (c >= 0x045e && c <= 0x0481) | |
448 | || (c >= 0x0490 && c <= 0x04c4) | |
449 | || (c >= 0x04c7 && c <= 0x04c8) | |
450 | || (c >= 0x04cb && c <= 0x04cc) | |
451 | || (c >= 0x04d0 && c <= 0x04eb) | |
452 | || (c >= 0x04ee && c <= 0x04f5) | |
453 | || (c >= 0x04f8 && c <= 0x04f9)) | |
454 | return 1; | |
455 | ||
456 | /* Armenian */ | |
457 | if ((c >= 0x0531 && c <= 0x0556) | |
458 | || (c >= 0x0561 && c <= 0x0587)) | |
459 | return 1; | |
460 | ||
461 | /* Hebrew */ | |
462 | if ((c >= 0x05d0 && c <= 0x05ea) | |
463 | || (c >= 0x05f0 && c <= 0x05f4)) | |
464 | return 1; | |
465 | ||
466 | /* Arabic */ | |
467 | if ((c >= 0x0621 && c <= 0x063a) | |
468 | || (c >= 0x0640 && c <= 0x0652) | |
469 | || (c >= 0x0670 && c <= 0x06b7) | |
470 | || (c >= 0x06ba && c <= 0x06be) | |
471 | || (c >= 0x06c0 && c <= 0x06ce) | |
472 | || (c >= 0x06e5 && c <= 0x06e7)) | |
473 | return 1; | |
474 | ||
475 | /* Devanagari */ | |
476 | if ((c >= 0x0905 && c <= 0x0939) | |
477 | || (c >= 0x0958 && c <= 0x0962)) | |
478 | return 1; | |
479 | ||
480 | /* Bengali */ | |
481 | if ((c >= 0x0985 && c <= 0x098c) | |
482 | || (c >= 0x098f && c <= 0x0990) | |
483 | || (c >= 0x0993 && c <= 0x09a8) | |
484 | || (c >= 0x09aa && c <= 0x09b0) | |
485 | || (c == 0x09b2) | |
486 | || (c >= 0x09b6 && c <= 0x09b9) | |
487 | || (c >= 0x09dc && c <= 0x09dd) | |
488 | || (c >= 0x09df && c <= 0x09e1) | |
489 | || (c >= 0x09f0 && c <= 0x09f1)) | |
490 | return 1; | |
491 | ||
492 | /* Gurmukhi */ | |
493 | if ((c >= 0x0a05 && c <= 0x0a0a) | |
494 | || (c >= 0x0a0f && c <= 0x0a10) | |
495 | || (c >= 0x0a13 && c <= 0x0a28) | |
496 | || (c >= 0x0a2a && c <= 0x0a30) | |
497 | || (c >= 0x0a32 && c <= 0x0a33) | |
498 | || (c >= 0x0a35 && c <= 0x0a36) | |
499 | || (c >= 0x0a38 && c <= 0x0a39) | |
500 | || (c >= 0x0a59 && c <= 0x0a5c) | |
501 | || (c == 0x0a5e)) | |
502 | return 1; | |
503 | ||
504 | /* Gujarati */ | |
505 | if ((c >= 0x0a85 && c <= 0x0a8b) | |
506 | || (c == 0x0a8d) | |
507 | || (c >= 0x0a8f && c <= 0x0a91) | |
508 | || (c >= 0x0a93 && c <= 0x0aa8) | |
509 | || (c >= 0x0aaa && c <= 0x0ab0) | |
510 | || (c >= 0x0ab2 && c <= 0x0ab3) | |
511 | || (c >= 0x0ab5 && c <= 0x0ab9) | |
512 | || (c == 0x0ae0)) | |
513 | return 1; | |
514 | ||
515 | /* Oriya */ | |
516 | if ((c >= 0x0b05 && c <= 0x0b0c) | |
517 | || (c >= 0x0b0f && c <= 0x0b10) | |
518 | || (c >= 0x0b13 && c <= 0x0b28) | |
519 | || (c >= 0x0b2a && c <= 0x0b30) | |
520 | || (c >= 0x0b32 && c <= 0x0b33) | |
521 | || (c >= 0x0b36 && c <= 0x0b39) | |
522 | || (c >= 0x0b5c && c <= 0x0b5d) | |
523 | || (c >= 0x0b5f && c <= 0x0b61)) | |
524 | return 1; | |
525 | ||
526 | /* Tamil */ | |
527 | if ((c >= 0x0b85 && c <= 0x0b8a) | |
528 | || (c >= 0x0b8e && c <= 0x0b90) | |
529 | || (c >= 0x0b92 && c <= 0x0b95) | |
530 | || (c >= 0x0b99 && c <= 0x0b9a) | |
531 | || (c == 0x0b9c) | |
532 | || (c >= 0x0b9e && c <= 0x0b9f) | |
533 | || (c >= 0x0ba3 && c <= 0x0ba4) | |
534 | || (c >= 0x0ba8 && c <= 0x0baa) | |
535 | || (c >= 0x0bae && c <= 0x0bb5) | |
536 | || (c >= 0x0bb7 && c <= 0x0bb9)) | |
537 | return 1; | |
538 | ||
539 | /* Telugu */ | |
540 | if ((c >= 0x0c05 && c <= 0x0c0c) | |
541 | || (c >= 0x0c0e && c <= 0x0c10) | |
542 | || (c >= 0x0c12 && c <= 0x0c28) | |
543 | || (c >= 0x0c2a && c <= 0x0c33) | |
544 | || (c >= 0x0c35 && c <= 0x0c39) | |
545 | || (c >= 0x0c60 && c <= 0x0c61)) | |
546 | return 1; | |
547 | ||
548 | /* Kannada */ | |
549 | if ((c >= 0x0c85 && c <= 0x0c8c) | |
550 | || (c >= 0x0c8e && c <= 0x0c90) | |
551 | || (c >= 0x0c92 && c <= 0x0ca8) | |
552 | || (c >= 0x0caa && c <= 0x0cb3) | |
553 | || (c >= 0x0cb5 && c <= 0x0cb9) | |
554 | || (c >= 0x0ce0 && c <= 0x0ce1)) | |
555 | return 1; | |
556 | ||
557 | /* Malayalam */ | |
558 | if ((c >= 0x0d05 && c <= 0x0d0c) | |
559 | || (c >= 0x0d0e && c <= 0x0d10) | |
560 | || (c >= 0x0d12 && c <= 0x0d28) | |
561 | || (c >= 0x0d2a && c <= 0x0d39) | |
562 | || (c >= 0x0d60 && c <= 0x0d61)) | |
563 | return 1; | |
564 | ||
565 | /* Thai */ | |
566 | if ((c >= 0x0e01 && c <= 0x0e30) | |
567 | || (c >= 0x0e32 && c <= 0x0e33) | |
568 | || (c >= 0x0e40 && c <= 0x0e46) | |
569 | || (c >= 0x0e4f && c <= 0x0e5b)) | |
570 | return 1; | |
571 | ||
572 | /* Lao */ | |
573 | if ((c >= 0x0e81 && c <= 0x0e82) | |
574 | || (c == 0x0e84) | |
575 | || (c == 0x0e87) | |
576 | || (c == 0x0e88) | |
577 | || (c == 0x0e8a) | |
578 | || (c == 0x0e0d) | |
579 | || (c >= 0x0e94 && c <= 0x0e97) | |
580 | || (c >= 0x0e99 && c <= 0x0e9f) | |
581 | || (c >= 0x0ea1 && c <= 0x0ea3) | |
582 | || (c == 0x0ea5) | |
583 | || (c == 0x0ea7) | |
584 | || (c == 0x0eaa) | |
585 | || (c == 0x0eab) | |
586 | || (c >= 0x0ead && c <= 0x0eb0) | |
587 | || (c == 0x0eb2) | |
588 | || (c == 0x0eb3) | |
589 | || (c == 0x0ebd) | |
590 | || (c >= 0x0ec0 && c <= 0x0ec4) | |
591 | || (c == 0x0ec6)) | |
592 | return 1; | |
593 | ||
594 | /* Georgian */ | |
595 | if ((c >= 0x10a0 && c <= 0x10c5) | |
596 | || (c >= 0x10d0 && c <= 0x10f6)) | |
597 | return 1; | |
598 | ||
599 | /* Hiragana */ | |
600 | if ((c >= 0x3041 && c <= 0x3094) | |
601 | || (c >= 0x309b && c <= 0x309e)) | |
602 | return 1; | |
603 | ||
604 | /* Katakana */ | |
605 | if ((c >= 0x30a1 && c <= 0x30fe)) | |
606 | return 1; | |
607 | ||
608 | /* Bopmofo */ | |
609 | if ((c >= 0x3105 && c <= 0x312c)) | |
610 | return 1; | |
611 | ||
612 | /* Hangul */ | |
613 | if ((c >= 0x1100 && c <= 0x1159) | |
614 | || (c >= 0x1161 && c <= 0x11a2) | |
615 | || (c >= 0x11a8 && c <= 0x11f9)) | |
616 | return 1; | |
617 | ||
618 | /* CJK Unified Ideographs */ | |
619 | if ((c >= 0xf900 && c <= 0xfa2d) | |
620 | || (c >= 0xfb1f && c <= 0xfb36) | |
621 | || (c >= 0xfb38 && c <= 0xfb3c) | |
622 | || (c == 0xfb3e) | |
623 | || (c >= 0xfb40 && c <= 0xfb41) | |
624 | || (c >= 0xfb42 && c <= 0xfb44) | |
625 | || (c >= 0xfb46 && c <= 0xfbb1) | |
626 | || (c >= 0xfbd3 && c <= 0xfd3f) | |
627 | || (c >= 0xfd50 && c <= 0xfd8f) | |
628 | || (c >= 0xfd92 && c <= 0xfdc7) | |
629 | || (c >= 0xfdf0 && c <= 0xfdfb) | |
630 | || (c >= 0xfe70 && c <= 0xfe72) | |
631 | || (c == 0xfe74) | |
632 | || (c >= 0xfe76 && c <= 0xfefc) | |
633 | || (c >= 0xff21 && c <= 0xff3a) | |
634 | || (c >= 0xff41 && c <= 0xff5a) | |
635 | || (c >= 0xff66 && c <= 0xffbe) | |
636 | || (c >= 0xffc2 && c <= 0xffc7) | |
637 | || (c >= 0xffca && c <= 0xffcf) | |
638 | || (c >= 0xffd2 && c <= 0xffd7) | |
639 | || (c >= 0xffda && c <= 0xffdc) | |
640 | || (c >= 0x4e00 && c <= 0x9fa5)) | |
641 | return 1; | |
642 | ||
643 | error ("universal-character-name '\\u%04x' not valid in identifier", c); | |
644 | return 1; | |
645 | #endif | |
646 | } | |
647 | ||
648 | /* Add the UTF-8 representation of C to the token_buffer. */ | |
649 | ||
650 | static void | |
651 | utf8_extend_token (c) | |
652 | int c; | |
e8bbfc4e | 653 | { |
0e5921e8 ZW |
654 | int shift, mask; |
655 | ||
656 | if (c <= 0x0000007f) | |
657 | { | |
658 | extend_token (c); | |
659 | return; | |
660 | } | |
661 | else if (c <= 0x000007ff) | |
662 | shift = 6, mask = 0xc0; | |
663 | else if (c <= 0x0000ffff) | |
664 | shift = 12, mask = 0xe0; | |
665 | else if (c <= 0x001fffff) | |
666 | shift = 18, mask = 0xf0; | |
667 | else if (c <= 0x03ffffff) | |
668 | shift = 24, mask = 0xf8; | |
e8bbfc4e | 669 | else |
0e5921e8 ZW |
670 | shift = 30, mask = 0xfc; |
671 | ||
672 | extend_token (mask | (c >> shift)); | |
673 | do | |
674 | { | |
675 | shift -= 6; | |
676 | extend_token ((unsigned char) (0x80 | (c >> shift))); | |
677 | } | |
678 | while (shift); | |
e8bbfc4e | 679 | } |
0e5921e8 | 680 | #endif |
e8bbfc4e RK |
681 | |
682 | #if 0 | |
e8bbfc4e RK |
683 | struct try_type |
684 | { | |
8b60264b KG |
685 | tree *const node_var; |
686 | const char unsigned_flag; | |
687 | const char long_flag; | |
688 | const char long_long_flag; | |
e8bbfc4e RK |
689 | }; |
690 | ||
75cb8865 | 691 | struct try_type type_sequence[] = |
e8bbfc4e RK |
692 | { |
693 | { &integer_type_node, 0, 0, 0}, | |
694 | { &unsigned_type_node, 1, 0, 0}, | |
695 | { &long_integer_type_node, 0, 1, 0}, | |
696 | { &long_unsigned_type_node, 1, 1, 0}, | |
697 | { &long_long_integer_type_node, 0, 1, 1}, | |
698 | { &long_long_unsigned_type_node, 1, 1, 1} | |
699 | }; | |
700 | #endif /* 0 */ | |
701 | \f | |
0e5921e8 ZW |
702 | int |
703 | c_lex (value) | |
704 | tree *value; | |
fbb18613 | 705 | { |
4ed5bcfb | 706 | const cpp_token *tok; |
0e5921e8 ZW |
707 | |
708 | retry: | |
709 | timevar_push (TV_CPP); | |
4ed5bcfb NB |
710 | do |
711 | tok = cpp_get_token (parse_in); | |
712 | while (tok->type == CPP_PADDING); | |
0e5921e8 ZW |
713 | timevar_pop (TV_CPP); |
714 | ||
715 | /* The C++ front end does horrible things with the current line | |
716 | number. To ensure an accurate line number, we must reset it | |
23356f93 | 717 | every time we return a token. */ |
97293897 | 718 | lineno = src_lineno; |
0e5921e8 ZW |
719 | |
720 | *value = NULL_TREE; | |
a23c9413 | 721 | switch (tok->type) |
0e5921e8 ZW |
722 | { |
723 | case CPP_OPEN_BRACE: indent_level++; break; | |
724 | case CPP_CLOSE_BRACE: indent_level--; break; | |
725 | ||
4ed5bcfb | 726 | /* Issue this error here, where we can get at tok->val.c. */ |
0e5921e8 | 727 | case CPP_OTHER: |
4ed5bcfb NB |
728 | if (ISGRAPH (tok->val.c)) |
729 | error ("stray '%c' in program", tok->val.c); | |
0e5921e8 | 730 | else |
4ed5bcfb | 731 | error ("stray '\\%o' in program", tok->val.c); |
0e5921e8 ZW |
732 | goto retry; |
733 | ||
0e5921e8 | 734 | case CPP_NAME: |
4ed5bcfb | 735 | *value = HT_IDENT_TO_GCC_IDENT (HT_NODE (tok->val.node)); |
0e5921e8 | 736 | break; |
fbb18613 | 737 | |
0e5921e8 | 738 | case CPP_NUMBER: |
4ed5bcfb | 739 | *value = lex_number ((const char *)tok->val.str.text, tok->val.str.len); |
0e5921e8 | 740 | break; |
93868d11 | 741 | |
0e5921e8 ZW |
742 | case CPP_CHAR: |
743 | case CPP_WCHAR: | |
4ed5bcfb | 744 | *value = lex_charconst (tok); |
0e5921e8 | 745 | break; |
fbb18613 | 746 | |
0e5921e8 ZW |
747 | case CPP_STRING: |
748 | case CPP_WSTRING: | |
62ae2529 JJ |
749 | *value = lex_string (tok->val.str.text, tok->val.str.len, |
750 | tok->type == CPP_WSTRING); | |
0e5921e8 | 751 | break; |
fbb18613 | 752 | |
0e5921e8 ZW |
753 | /* These tokens should not be visible outside cpplib. */ |
754 | case CPP_HEADER_NAME: | |
755 | case CPP_COMMENT: | |
756 | case CPP_MACRO_ARG: | |
0e5921e8 ZW |
757 | abort (); |
758 | ||
759 | default: break; | |
760 | } | |
761 | ||
a23c9413 | 762 | return tok->type; |
0e5921e8 | 763 | } |
8d9bfdc5 | 764 | |
0e5921e8 | 765 | #define ERROR(msgid) do { error(msgid); goto syntax_error; } while(0) |
75cb8865 | 766 | |
0e5921e8 ZW |
767 | static tree |
768 | lex_number (str, len) | |
769 | const char *str; | |
770 | unsigned int len; | |
771 | { | |
772 | int base = 10; | |
773 | int count = 0; | |
774 | int largest_digit = 0; | |
775 | int numdigits = 0; | |
776 | int overflow = 0; | |
777 | int c; | |
778 | tree value; | |
779 | const char *p; | |
780 | enum anon1 { NOT_FLOAT = 0, AFTER_POINT, AFTER_EXPON } floatflag = NOT_FLOAT; | |
781 | ||
782 | /* We actually store only HOST_BITS_PER_CHAR bits in each part. | |
783 | The code below which fills the parts array assumes that a host | |
784 | int is at least twice as wide as a host char, and that | |
785 | HOST_BITS_PER_WIDE_INT is an even multiple of HOST_BITS_PER_CHAR. | |
786 | Two HOST_WIDE_INTs is the largest int literal we can store. | |
787 | In order to detect overflow below, the number of parts (TOTAL_PARTS) | |
788 | must be exactly the number of parts needed to hold the bits | |
ec5c56db | 789 | of two HOST_WIDE_INTs. */ |
0e5921e8 ZW |
790 | #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2) |
791 | unsigned int parts[TOTAL_PARTS]; | |
792 | ||
793 | /* Optimize for most frequent case. */ | |
794 | if (len == 1) | |
795 | { | |
796 | if (*str == '0') | |
797 | return integer_zero_node; | |
798 | else if (*str == '1') | |
799 | return integer_one_node; | |
800 | else | |
801 | return build_int_2 (*str - '0', 0); | |
802 | } | |
e8bbfc4e | 803 | |
0e5921e8 ZW |
804 | for (count = 0; count < TOTAL_PARTS; count++) |
805 | parts[count] = 0; | |
e8bbfc4e | 806 | |
0e5921e8 ZW |
807 | /* len is known to be >1 at this point. */ |
808 | p = str; | |
e8bbfc4e | 809 | |
0e5921e8 ZW |
810 | if (len > 2 && str[0] == '0' && (str[1] == 'x' || str[1] == 'X')) |
811 | { | |
812 | base = 16; | |
813 | p = str + 2; | |
814 | } | |
815 | /* The ISDIGIT check is so we are not confused by a suffix on 0. */ | |
816 | else if (str[0] == '0' && ISDIGIT (str[1])) | |
817 | { | |
818 | base = 8; | |
819 | p = str + 1; | |
820 | } | |
e8bbfc4e | 821 | |
0e5921e8 ZW |
822 | do |
823 | { | |
824 | c = *p++; | |
88d92ca5 | 825 | |
0e5921e8 ZW |
826 | if (c == '.') |
827 | { | |
c215b29f | 828 | if (floatflag == AFTER_POINT) |
0e5921e8 ZW |
829 | ERROR ("too many decimal points in floating constant"); |
830 | else if (floatflag == AFTER_EXPON) | |
831 | ERROR ("decimal point in exponent - impossible!"); | |
832 | else | |
833 | floatflag = AFTER_POINT; | |
c832a30e | 834 | |
0e5921e8 ZW |
835 | if (base == 8) |
836 | base = 10; | |
837 | } | |
838 | else if (c == '_') | |
839 | /* Possible future extension: silently ignore _ in numbers, | |
840 | permitting cosmetic grouping - e.g. 0x8000_0000 == 0x80000000 | |
841 | but somewhat easier to read. Ada has this? */ | |
842 | ERROR ("underscore in number"); | |
843 | else | |
844 | { | |
845 | int n; | |
846 | /* It is not a decimal point. | |
847 | It should be a digit (perhaps a hex digit). */ | |
0e0fda0d | 848 | |
faf31866 KG |
849 | if (ISDIGIT (c) |
850 | || (base == 16 && ISXDIGIT (c))) | |
0e5921e8 | 851 | { |
faf31866 | 852 | n = hex_value (c); |
0e5921e8 ZW |
853 | } |
854 | else if (base <= 10 && (c == 'e' || c == 'E')) | |
855 | { | |
856 | base = 10; | |
857 | floatflag = AFTER_EXPON; | |
858 | break; | |
859 | } | |
860 | else if (base == 16 && (c == 'p' || c == 'P')) | |
861 | { | |
862 | floatflag = AFTER_EXPON; | |
863 | break; /* start of exponent */ | |
864 | } | |
0e5921e8 ZW |
865 | else |
866 | { | |
867 | p--; | |
868 | break; /* start of suffix */ | |
869 | } | |
0e0fda0d | 870 | |
0e5921e8 ZW |
871 | if (n >= largest_digit) |
872 | largest_digit = n; | |
873 | numdigits++; | |
0e0fda0d | 874 | |
0e5921e8 ZW |
875 | for (count = 0; count < TOTAL_PARTS; count++) |
876 | { | |
877 | parts[count] *= base; | |
878 | if (count) | |
879 | { | |
880 | parts[count] | |
881 | += (parts[count-1] >> HOST_BITS_PER_CHAR); | |
882 | parts[count-1] | |
883 | &= (1 << HOST_BITS_PER_CHAR) - 1; | |
884 | } | |
885 | else | |
886 | parts[0] += n; | |
887 | } | |
e8bbfc4e | 888 | |
0e5921e8 ZW |
889 | /* If the highest-order part overflows (gets larger than |
890 | a host char will hold) then the whole number has | |
891 | overflowed. Record this and truncate the highest-order | |
ec5c56db | 892 | part. */ |
0e5921e8 ZW |
893 | if (parts[TOTAL_PARTS - 1] >> HOST_BITS_PER_CHAR) |
894 | { | |
895 | overflow = 1; | |
896 | parts[TOTAL_PARTS - 1] &= (1 << HOST_BITS_PER_CHAR) - 1; | |
897 | } | |
898 | } | |
899 | } | |
900 | while (p < str + len); | |
91b0989e | 901 | |
0e5921e8 ZW |
902 | /* This can happen on input like `int i = 0x;' */ |
903 | if (numdigits == 0) | |
904 | ERROR ("numeric constant with no digits"); | |
91b0989e | 905 | |
0e5921e8 ZW |
906 | if (largest_digit >= base) |
907 | ERROR ("numeric constant contains digits beyond the radix"); | |
e8bbfc4e | 908 | |
0e5921e8 ZW |
909 | if (floatflag != NOT_FLOAT) |
910 | { | |
911 | tree type; | |
15e5ad76 ZW |
912 | const char *typename; |
913 | int imag, fflag, lflag; | |
0e5921e8 | 914 | REAL_VALUE_TYPE real; |
0e5921e8 ZW |
915 | char *copy; |
916 | ||
917 | if (base == 16 && floatflag != AFTER_EXPON) | |
918 | ERROR ("hexadecimal floating constant has no exponent"); | |
919 | ||
920 | /* Read explicit exponent if any, and put it in tokenbuf. */ | |
921 | if ((base == 10 && ((c == 'e') || (c == 'E'))) | |
922 | || (base == 16 && (c == 'p' || c == 'P'))) | |
923 | { | |
924 | if (p < str + len) | |
925 | c = *p++; | |
926 | if (p < str + len && (c == '+' || c == '-')) | |
927 | c = *p++; | |
928 | /* Exponent is decimal, even if string is a hex float. */ | |
929 | if (! ISDIGIT (c)) | |
930 | ERROR ("floating constant exponent has no digits"); | |
931 | while (p < str + len && ISDIGIT (c)) | |
932 | c = *p++; | |
933 | if (! ISDIGIT (c)) | |
934 | p--; | |
935 | } | |
56f48ce9 | 936 | |
0e5921e8 ZW |
937 | /* Copy the float constant now; we don't want any suffixes in the |
938 | string passed to parse_float. */ | |
939 | copy = alloca (p - str + 1); | |
940 | memcpy (copy, str, p - str); | |
941 | copy[p - str] = '\0'; | |
e8bbfc4e | 942 | |
0e5921e8 ZW |
943 | /* Now parse suffixes. */ |
944 | fflag = lflag = imag = 0; | |
945 | while (p < str + len) | |
946 | switch (*p++) | |
e8bbfc4e | 947 | { |
0e5921e8 ZW |
948 | case 'f': case 'F': |
949 | if (fflag) | |
950 | ERROR ("more than one 'f' suffix on floating constant"); | |
7065e130 | 951 | else if (warn_traditional && !in_system_header |
c691145a | 952 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 | 953 | warning ("traditional C rejects the 'f' suffix"); |
e8bbfc4e | 954 | |
0e5921e8 ZW |
955 | fflag = 1; |
956 | break; | |
e8bbfc4e | 957 | |
0e5921e8 ZW |
958 | case 'l': case 'L': |
959 | if (lflag) | |
960 | ERROR ("more than one 'l' suffix on floating constant"); | |
7065e130 | 961 | else if (warn_traditional && !in_system_header |
c691145a | 962 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 | 963 | warning ("traditional C rejects the 'l' suffix"); |
56f48ce9 | 964 | |
0e5921e8 ZW |
965 | lflag = 1; |
966 | break; | |
e8bbfc4e | 967 | |
0e5921e8 ZW |
968 | case 'i': case 'I': |
969 | case 'j': case 'J': | |
970 | if (imag) | |
971 | ERROR ("more than one 'i' or 'j' suffix on floating constant"); | |
972 | else if (pedantic) | |
973 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
974 | imag = 1; | |
975 | break; | |
e8bbfc4e | 976 | |
0e5921e8 ZW |
977 | default: |
978 | ERROR ("invalid suffix on floating constant"); | |
e8bbfc4e RK |
979 | } |
980 | ||
15e5ad76 ZW |
981 | type = double_type_node; |
982 | typename = "double"; | |
983 | ||
984 | if (fflag) | |
985 | { | |
986 | if (lflag) | |
987 | ERROR ("both 'f' and 'l' suffixes on floating constant"); | |
e8bbfc4e | 988 | |
15e5ad76 ZW |
989 | type = float_type_node; |
990 | typename = "float"; | |
991 | } | |
992 | else if (lflag) | |
0e5921e8 | 993 | { |
15e5ad76 ZW |
994 | type = long_double_type_node; |
995 | typename = "long double"; | |
0e5921e8 | 996 | } |
15e5ad76 ZW |
997 | else if (flag_single_precision_constant) |
998 | { | |
999 | type = float_type_node; | |
1000 | typename = "float"; | |
1001 | } | |
1002 | ||
1003 | /* Warn about this only after we know we're not issuing an error. */ | |
1004 | if (base == 16 && pedantic && !flag_isoc99) | |
1005 | pedwarn ("hexadecimal floating constants are only valid in C99"); | |
1006 | ||
1007 | /* The second argument, machine_mode, of REAL_VALUE_ATOF | |
1008 | tells the desired precision of the binary result | |
1009 | of decimal-to-binary conversion. */ | |
1010 | if (base == 16) | |
1011 | real = REAL_VALUE_HTOF (copy, TYPE_MODE (type)); | |
0e5921e8 | 1012 | else |
15e5ad76 ZW |
1013 | real = REAL_VALUE_ATOF (copy, TYPE_MODE (type)); |
1014 | ||
1015 | /* A diagnostic is required here by some ISO C testsuites. | |
1016 | This is not pedwarn, because some people don't want | |
1017 | an error for this. */ | |
1018 | if (REAL_VALUE_ISINF (real) && pedantic) | |
0e5921e8 | 1019 | warning ("floating point number exceeds range of 'double'"); |
e8bbfc4e | 1020 | |
0e5921e8 ZW |
1021 | /* Create a node with determined type and value. */ |
1022 | if (imag) | |
1023 | value = build_complex (NULL_TREE, convert (type, integer_zero_node), | |
1024 | build_real (type, real)); | |
1025 | else | |
1026 | value = build_real (type, real); | |
1027 | } | |
1028 | else | |
1029 | { | |
f458d1d5 | 1030 | tree trad_type, type; |
0e5921e8 ZW |
1031 | HOST_WIDE_INT high, low; |
1032 | int spec_unsigned = 0; | |
1033 | int spec_long = 0; | |
1034 | int spec_long_long = 0; | |
1035 | int spec_imag = 0; | |
1036 | int suffix_lu = 0; | |
1037 | int warn = 0, i; | |
1038 | ||
f458d1d5 | 1039 | trad_type = type = NULL_TREE; |
0e5921e8 ZW |
1040 | while (p < str + len) |
1041 | { | |
1042 | c = *p++; | |
1043 | switch (c) | |
1044 | { | |
1045 | case 'u': case 'U': | |
1046 | if (spec_unsigned) | |
1047 | error ("two 'u' suffixes on integer constant"); | |
7065e130 | 1048 | else if (warn_traditional && !in_system_header |
c691145a | 1049 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 ZW |
1050 | warning ("traditional C rejects the 'u' suffix"); |
1051 | ||
1052 | spec_unsigned = 1; | |
1053 | if (spec_long) | |
1054 | suffix_lu = 1; | |
1055 | break; | |
e8bbfc4e | 1056 | |
0e5921e8 ZW |
1057 | case 'l': case 'L': |
1058 | if (spec_long) | |
1059 | { | |
1060 | if (spec_long_long) | |
1061 | error ("three 'l' suffixes on integer constant"); | |
1062 | else if (suffix_lu) | |
1063 | error ("'lul' is not a valid integer suffix"); | |
1064 | else if (c != spec_long) | |
1065 | error ("'Ll' and 'lL' are not valid integer suffixes"); | |
1066 | else if (pedantic && ! flag_isoc99 | |
1067 | && ! in_system_header && warn_long_long) | |
1068 | pedwarn ("ISO C89 forbids long long integer constants"); | |
1069 | spec_long_long = 1; | |
1070 | } | |
1071 | spec_long = c; | |
1072 | break; | |
56f48ce9 | 1073 | |
0e5921e8 ZW |
1074 | case 'i': case 'I': case 'j': case 'J': |
1075 | if (spec_imag) | |
1076 | error ("more than one 'i' or 'j' suffix on integer constant"); | |
1077 | else if (pedantic) | |
1078 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
1079 | spec_imag = 1; | |
1080 | break; | |
56f48ce9 | 1081 | |
0e5921e8 ZW |
1082 | default: |
1083 | ERROR ("invalid suffix on integer constant"); | |
1084 | } | |
1085 | } | |
56f48ce9 | 1086 | |
ec5c56db | 1087 | /* If the literal overflowed, pedwarn about it now. */ |
0e5921e8 ZW |
1088 | if (overflow) |
1089 | { | |
1090 | warn = 1; | |
1091 | pedwarn ("integer constant is too large for this configuration of the compiler - truncated to %d bits", HOST_BITS_PER_WIDE_INT * 2); | |
1092 | } | |
e8bbfc4e | 1093 | |
0e5921e8 ZW |
1094 | /* This is simplified by the fact that our constant |
1095 | is always positive. */ | |
56f48ce9 | 1096 | |
0e5921e8 | 1097 | high = low = 0; |
e8bbfc4e | 1098 | |
0e5921e8 ZW |
1099 | for (i = 0; i < HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR; i++) |
1100 | { | |
1101 | high |= ((HOST_WIDE_INT) parts[i + (HOST_BITS_PER_WIDE_INT | |
1102 | / HOST_BITS_PER_CHAR)] | |
1103 | << (i * HOST_BITS_PER_CHAR)); | |
1104 | low |= (HOST_WIDE_INT) parts[i] << (i * HOST_BITS_PER_CHAR); | |
1105 | } | |
0468bc75 | 1106 | |
0e5921e8 ZW |
1107 | value = build_int_2 (low, high); |
1108 | TREE_TYPE (value) = long_long_unsigned_type_node; | |
e8bbfc4e | 1109 | |
0e5921e8 | 1110 | /* If warn_traditional, calculate both the ISO type and the |
f458d1d5 ZW |
1111 | traditional type, then see if they disagree. */ |
1112 | if (warn_traditional) | |
0e5921e8 | 1113 | { |
0e5921e8 ZW |
1114 | /* Traditionally, any constant is signed; but if unsigned is |
1115 | specified explicitly, obey that. Use the smallest size | |
1116 | with the right number of bits, except for one special | |
1117 | case with decimal constants. */ | |
1118 | if (! spec_long && base != 10 | |
1119 | && int_fits_type_p (value, unsigned_type_node)) | |
1120 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
1121 | /* A decimal constant must be long if it does not fit in | |
1122 | type int. I think this is independent of whether the | |
1123 | constant is signed. */ | |
1124 | else if (! spec_long && base == 10 | |
1125 | && int_fits_type_p (value, integer_type_node)) | |
1126 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
1127 | else if (! spec_long_long) | |
1128 | trad_type = (spec_unsigned | |
1129 | ? long_unsigned_type_node | |
1130 | : long_integer_type_node); | |
1131 | else if (int_fits_type_p (value, | |
1132 | spec_unsigned | |
1133 | ? long_long_unsigned_type_node | |
1134 | : long_long_integer_type_node)) | |
1135 | trad_type = (spec_unsigned | |
1136 | ? long_long_unsigned_type_node | |
1137 | : long_long_integer_type_node); | |
1138 | else | |
1139 | trad_type = (spec_unsigned | |
1140 | ? widest_unsigned_literal_type_node | |
1141 | : widest_integer_literal_type_node); | |
1142 | } | |
f458d1d5 ZW |
1143 | |
1144 | /* Calculate the ISO type. */ | |
1145 | if (! spec_long && ! spec_unsigned | |
1146 | && int_fits_type_p (value, integer_type_node)) | |
1147 | type = integer_type_node; | |
1148 | else if (! spec_long && (base != 10 || spec_unsigned) | |
1149 | && int_fits_type_p (value, unsigned_type_node)) | |
1150 | type = unsigned_type_node; | |
1151 | else if (! spec_unsigned && !spec_long_long | |
1152 | && int_fits_type_p (value, long_integer_type_node)) | |
1153 | type = long_integer_type_node; | |
1154 | else if (! spec_long_long | |
1155 | && int_fits_type_p (value, long_unsigned_type_node)) | |
1156 | type = long_unsigned_type_node; | |
1157 | else if (! spec_unsigned | |
1158 | && int_fits_type_p (value, long_long_integer_type_node)) | |
1159 | type = long_long_integer_type_node; | |
1160 | else if (int_fits_type_p (value, long_long_unsigned_type_node)) | |
1161 | type = long_long_unsigned_type_node; | |
1162 | else if (! spec_unsigned | |
1163 | && int_fits_type_p (value, widest_integer_literal_type_node)) | |
1164 | type = widest_integer_literal_type_node; | |
1165 | else | |
1166 | type = widest_unsigned_literal_type_node; | |
e8bbfc4e | 1167 | |
0e5921e8 ZW |
1168 | /* We assume that constants specified in a non-decimal |
1169 | base are bit patterns, and that the programmer really | |
1170 | meant what they wrote. */ | |
1171 | if (warn_traditional && !in_system_header | |
f458d1d5 | 1172 | && base == 10 && trad_type != type) |
0e5921e8 | 1173 | { |
f458d1d5 ZW |
1174 | if (TYPE_PRECISION (trad_type) != TYPE_PRECISION (type)) |
1175 | warning ("width of integer constant is different in traditional C"); | |
1176 | else if (TREE_UNSIGNED (trad_type) != TREE_UNSIGNED (type)) | |
1177 | warning ("integer constant is unsigned in ISO C, signed in traditional C"); | |
0e5921e8 | 1178 | else |
f458d1d5 | 1179 | warning ("width of integer constant may change on other systems in traditional C"); |
0e5921e8 | 1180 | } |
e8bbfc4e | 1181 | |
f458d1d5 | 1182 | if (pedantic && (flag_isoc99 || !spec_long_long) |
b15ad712 JM |
1183 | && !warn |
1184 | && ((flag_isoc99 | |
1185 | ? TYPE_PRECISION (long_long_integer_type_node) | |
1186 | : TYPE_PRECISION (long_integer_type_node)) < TYPE_PRECISION (type))) | |
0e5921e8 ZW |
1187 | { |
1188 | warn = 1; | |
b15ad712 JM |
1189 | pedwarn ("integer constant larger than the maximum value of %s", |
1190 | (flag_isoc99 | |
1191 | ? (TREE_UNSIGNED (type) | |
53fcdc76 PB |
1192 | ? _("an unsigned long long int") |
1193 | : _("a long long int")) | |
1194 | : _("an unsigned long int"))); | |
0e5921e8 | 1195 | } |
e8bbfc4e | 1196 | |
0e5921e8 ZW |
1197 | if (base == 10 && ! spec_unsigned && TREE_UNSIGNED (type)) |
1198 | warning ("decimal constant is so large that it is unsigned"); | |
e8bbfc4e | 1199 | |
0e5921e8 ZW |
1200 | if (spec_imag) |
1201 | { | |
1202 | if (TYPE_PRECISION (type) | |
1203 | <= TYPE_PRECISION (integer_type_node)) | |
1204 | value = build_complex (NULL_TREE, integer_zero_node, | |
1205 | convert (integer_type_node, value)); | |
1206 | else | |
1207 | ERROR ("complex integer constant is too wide for 'complex int'"); | |
1208 | } | |
0e5921e8 ZW |
1209 | else |
1210 | TREE_TYPE (value) = type; | |
e8bbfc4e | 1211 | |
0e5921e8 | 1212 | /* If it's still an integer (not a complex), and it doesn't |
ec5c56db | 1213 | fit in the type we choose for it, then pedwarn. */ |
fbb18613 | 1214 | |
0e5921e8 ZW |
1215 | if (! warn |
1216 | && TREE_CODE (TREE_TYPE (value)) == INTEGER_TYPE | |
1217 | && ! int_fits_type_p (value, TREE_TYPE (value))) | |
1218 | pedwarn ("integer constant is larger than the maximum value for its type"); | |
1219 | } | |
fbb18613 | 1220 | |
0e5921e8 ZW |
1221 | if (p < str + len) |
1222 | error ("missing white space after number '%.*s'", (int) (p - str), str); | |
e8bbfc4e | 1223 | |
0e5921e8 | 1224 | return value; |
e8bbfc4e | 1225 | |
0e5921e8 ZW |
1226 | syntax_error: |
1227 | return integer_zero_node; | |
1228 | } | |
e8bbfc4e | 1229 | |
0e5921e8 ZW |
1230 | static tree |
1231 | lex_string (str, len, wide) | |
62ae2529 | 1232 | const unsigned char *str; |
0e5921e8 ZW |
1233 | unsigned int len; |
1234 | int wide; | |
1235 | { | |
1236 | tree value; | |
1237 | char *buf = alloca ((len + 1) * (wide ? WCHAR_BYTES : 1)); | |
1238 | char *q = buf; | |
62ae2529 | 1239 | const unsigned char *p = str, *limit = str + len; |
0e5921e8 ZW |
1240 | unsigned int c; |
1241 | unsigned width = wide ? WCHAR_TYPE_SIZE | |
1242 | : TYPE_PRECISION (char_type_node); | |
e9a25f70 | 1243 | |
0e5921e8 ZW |
1244 | #ifdef MULTIBYTE_CHARS |
1245 | /* Reset multibyte conversion state. */ | |
9714cf43 | 1246 | (void) local_mbtowc (NULL, NULL, 0); |
0e5921e8 | 1247 | #endif |
e9a25f70 | 1248 | |
0e5921e8 ZW |
1249 | while (p < limit) |
1250 | { | |
1251 | #ifdef MULTIBYTE_CHARS | |
1252 | wchar_t wc; | |
1253 | int char_len; | |
1254 | ||
62ae2529 | 1255 | char_len = local_mbtowc (&wc, (const char *) p, limit - p); |
0e5921e8 ZW |
1256 | if (char_len == -1) |
1257 | { | |
53fcdc76 | 1258 | warning ("ignoring invalid multibyte character"); |
0e5921e8 ZW |
1259 | char_len = 1; |
1260 | c = *p++; | |
1261 | } | |
1262 | else | |
1263 | { | |
1264 | p += char_len; | |
1265 | c = wc; | |
1266 | } | |
1267 | #else | |
1268 | c = *p++; | |
1269 | #endif | |
1270 | ||
1271 | if (c == '\\' && !ignore_escape_flag) | |
1272 | { | |
62729350 NB |
1273 | unsigned int mask; |
1274 | ||
1275 | if (width < HOST_BITS_PER_INT) | |
1276 | mask = ((unsigned int) 1 << width) - 1; | |
1277 | else | |
1278 | mask = ~0; | |
62ae2529 | 1279 | c = cpp_parse_escape (parse_in, &p, limit, mask); |
0e5921e8 ZW |
1280 | } |
1281 | ||
64cdc383 MH |
1282 | /* Add this single character into the buffer either as a wchar_t, |
1283 | a multibyte sequence, or as a single byte. */ | |
0e5921e8 ZW |
1284 | if (wide) |
1285 | { | |
1286 | unsigned charwidth = TYPE_PRECISION (char_type_node); | |
5c80f6e6 | 1287 | unsigned bytemask = (1 << charwidth) - 1; |
0e5921e8 ZW |
1288 | int byte; |
1289 | ||
1290 | for (byte = 0; byte < WCHAR_BYTES; ++byte) | |
1291 | { | |
1292 | int n; | |
1293 | if (byte >= (int) sizeof (c)) | |
1294 | n = 0; | |
1295 | else | |
1296 | n = (c >> (byte * charwidth)) & bytemask; | |
1297 | if (BYTES_BIG_ENDIAN) | |
1298 | q[WCHAR_BYTES - byte - 1] = n; | |
1299 | else | |
1300 | q[byte] = n; | |
1301 | } | |
1302 | q += WCHAR_BYTES; | |
1303 | } | |
64cdc383 MH |
1304 | #ifdef MULTIBYTE_CHARS |
1305 | else if (char_len > 1) | |
1306 | { | |
1307 | /* We're dealing with a multibyte character. */ | |
1308 | for ( ; char_len >0; --char_len) | |
1309 | { | |
1310 | *q++ = *(p - char_len); | |
1311 | } | |
1312 | } | |
1313 | #endif | |
0e5921e8 ZW |
1314 | else |
1315 | { | |
1316 | *q++ = c; | |
1317 | } | |
e8bbfc4e RK |
1318 | } |
1319 | ||
0e5921e8 ZW |
1320 | /* Terminate the string value, either with a single byte zero |
1321 | or with a wide zero. */ | |
e8bbfc4e | 1322 | |
0e5921e8 ZW |
1323 | if (wide) |
1324 | { | |
1325 | memset (q, 0, WCHAR_BYTES); | |
1326 | q += WCHAR_BYTES; | |
1327 | } | |
1328 | else | |
1329 | { | |
1330 | *q++ = '\0'; | |
1331 | } | |
1332 | ||
1333 | value = build_string (q - buf, buf); | |
1334 | ||
1335 | if (wide) | |
1336 | TREE_TYPE (value) = wchar_array_type_node; | |
1337 | else | |
1338 | TREE_TYPE (value) = char_array_type_node; | |
e8bbfc4e RK |
1339 | return value; |
1340 | } | |
1341 | ||
c8a96070 | 1342 | /* Converts a (possibly wide) character constant token into a tree. */ |
0e5921e8 | 1343 | static tree |
c8a96070 NB |
1344 | lex_charconst (token) |
1345 | const cpp_token *token; | |
e8bbfc4e | 1346 | { |
c8a96070 | 1347 | HOST_WIDE_INT result; |
9340544b | 1348 | tree type, value; |
c8a96070 NB |
1349 | unsigned int chars_seen; |
1350 | ||
1351 | result = cpp_interpret_charconst (parse_in, token, warn_multichar, | |
f458d1d5 | 1352 | &chars_seen); |
c8a96070 | 1353 | if (token->type == CPP_WCHAR) |
0e5921e8 | 1354 | { |
c8a96070 | 1355 | value = build_int_2 (result, 0); |
9340544b | 1356 | type = wchar_type_node; |
0e5921e8 | 1357 | } |
c8a96070 | 1358 | else |
0e5921e8 | 1359 | { |
c8a96070 NB |
1360 | if (result < 0) |
1361 | value = build_int_2 (result, -1); | |
0e5921e8 | 1362 | else |
c8a96070 NB |
1363 | value = build_int_2 (result, 0); |
1364 | ||
1365 | /* In C, a character constant has type 'int'. | |
1366 | In C++ 'char', but multi-char charconsts have type 'int'. */ | |
1367 | if (c_language == clk_cplusplus && chars_seen <= 1) | |
9340544b | 1368 | type = char_type_node; |
0e5921e8 | 1369 | else |
9340544b | 1370 | type = integer_type_node; |
0e5921e8 | 1371 | } |
9340544b ZW |
1372 | |
1373 | /* cpp_interpret_charconst issues a warning if the constant | |
1374 | overflows, but if the number fits in HOST_WIDE_INT anyway, it | |
1375 | will return it un-truncated, which may cause problems down the | |
1376 | line. So set the type to widest_integer_literal_type, call | |
1377 | convert to truncate it to the proper type, then clear | |
1378 | TREE_OVERFLOW so we don't get a second warning. | |
1379 | ||
1380 | FIXME: cpplib's assessment of overflow may not be accurate on a | |
1381 | platform where the final type can change at (compiler's) runtime. */ | |
1382 | ||
1383 | TREE_TYPE (value) = widest_integer_literal_type_node; | |
1384 | value = convert (type, value); | |
1385 | TREE_OVERFLOW (value) = 0; | |
1386 | ||
0e5921e8 | 1387 | return value; |
e8bbfc4e | 1388 | } |