]>
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 | 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" |
7bdb32b9 | 38 | #include "tm_p.h" |
0e5921e8 | 39 | #include "splay-tree.h" |
7f905405 | 40 | #include "debug.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 | |
373e2177 NB |
57 | /* The input filename as understood by CPP, where "" represents stdin. */ |
58 | static const char *cpp_filename; | |
3aac38d7 | 59 | |
67821e3a | 60 | /* The current line map. */ |
47d89cf3 | 61 | static const struct line_map *map; |
67821e3a | 62 | |
97293897 NB |
63 | /* The line used to refresh the lineno global variable after each token. */ |
64 | static unsigned int src_lineno; | |
65 | ||
0e5921e8 ZW |
66 | /* We may keep statistics about how long which files took to compile. */ |
67 | static int header_time, body_time; | |
68 | static splay_tree file_info_tree; | |
3ab6dd7c | 69 | |
e8bbfc4e RK |
70 | /* Cause the `yydebug' variable to be defined. */ |
71 | #define YYDEBUG 1 | |
72 | ||
e8bbfc4e RK |
73 | /* File used for outputting assembler code. */ |
74 | extern FILE *asm_out_file; | |
75 | ||
12a39b12 JM |
76 | #undef WCHAR_TYPE_SIZE |
77 | #define WCHAR_TYPE_SIZE TYPE_PRECISION (wchar_type_node) | |
e8bbfc4e RK |
78 | |
79 | /* Number of bytes in a wide character. */ | |
80 | #define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT) | |
81 | ||
ec5c56db | 82 | int indent_level; /* Number of { minus number of }. */ |
0e5921e8 ZW |
83 | int pending_lang_change; /* If we need to switch languages - C++ only */ |
84 | int c_header_level; /* depth in C headers - C++ only */ | |
fbb18613 JM |
85 | |
86 | /* Nonzero tells yylex to ignore \ in string constants. */ | |
87 | static int ignore_escape_flag; | |
e9a25f70 | 88 | |
0e5921e8 ZW |
89 | static void parse_float PARAMS ((PTR)); |
90 | static tree lex_number PARAMS ((const char *, unsigned int)); | |
91 | static tree lex_string PARAMS ((const char *, unsigned int, int)); | |
c8a96070 | 92 | static tree lex_charconst PARAMS ((const cpp_token *)); |
0e5921e8 ZW |
93 | static void update_header_times PARAMS ((const char *)); |
94 | static int dump_one_header PARAMS ((splay_tree_node, void *)); | |
97293897 | 95 | static void cb_line_change PARAMS ((cpp_reader *, const cpp_token *, int)); |
8bbbef34 NB |
96 | static void cb_ident PARAMS ((cpp_reader *, unsigned int, |
97 | const cpp_string *)); | |
47d89cf3 | 98 | static void cb_file_change PARAMS ((cpp_reader *, const struct line_map *)); |
8bbbef34 NB |
99 | static void cb_def_pragma PARAMS ((cpp_reader *, unsigned int)); |
100 | static void cb_define PARAMS ((cpp_reader *, unsigned int, | |
101 | cpp_hashnode *)); | |
102 | static void cb_undef PARAMS ((cpp_reader *, unsigned int, | |
103 | cpp_hashnode *)); | |
e31c7eec | 104 | \f |
3b304f5b | 105 | const char * |
0e5921e8 | 106 | init_c_lex (filename) |
3b304f5b | 107 | const char *filename; |
e3d1fd32 | 108 | { |
b61c5ed0 | 109 | struct cpp_callbacks *cb; |
0e5921e8 ZW |
110 | struct c_fileinfo *toplevel; |
111 | ||
112 | /* Set up filename timing. Must happen before cpp_start_read. */ | |
113 | file_info_tree = splay_tree_new ((splay_tree_compare_fn)strcmp, | |
114 | 0, | |
115 | (splay_tree_delete_value_fn)free); | |
a8a05998 | 116 | toplevel = get_fileinfo ("<top level>"); |
0e5921e8 ZW |
117 | if (flag_detailed_statistics) |
118 | { | |
119 | header_time = 0; | |
120 | body_time = get_run_time (); | |
121 | toplevel->time = body_time; | |
122 | } | |
123 | ||
124 | #ifdef MULTIBYTE_CHARS | |
125 | /* Change to the native locale for multibyte conversions. */ | |
126 | setlocale (LC_CTYPE, ""); | |
127 | GET_ENVIRONMENT (literal_codeset, "LANG"); | |
128 | #endif | |
129 | ||
b61c5ed0 NB |
130 | cb = cpp_get_callbacks (parse_in); |
131 | ||
97293897 | 132 | cb->line_change = cb_line_change; |
b61c5ed0 NB |
133 | cb->ident = cb_ident; |
134 | cb->file_change = cb_file_change; | |
135 | cb->def_pragma = cb_def_pragma; | |
0e5921e8 | 136 | |
65289a3a NB |
137 | /* Set the debug callbacks if we can use them. */ |
138 | if (debug_info_level == DINFO_LEVEL_VERBOSE | |
139 | && (write_symbols == DWARF_DEBUG || write_symbols == DWARF2_DEBUG)) | |
140 | { | |
b61c5ed0 NB |
141 | cb->define = cb_define; |
142 | cb->undef = cb_undef; | |
65289a3a NB |
143 | } |
144 | ||
373e2177 | 145 | |
b12da25e | 146 | if (filename == 0 || !strcmp (filename, "-")) |
373e2177 NB |
147 | filename = "stdin", cpp_filename = ""; |
148 | else | |
149 | cpp_filename = filename; | |
5c60e5c0 | 150 | |
851ef036 NB |
151 | /* Start it at 0. */ |
152 | lineno = 0; | |
b12da25e | 153 | |
5c60e5c0 | 154 | return filename; |
e3d1fd32 PB |
155 | } |
156 | ||
3aac38d7 RH |
157 | /* A thin wrapper around the real parser that initializes the |
158 | integrated preprocessor after debug output has been initialized. */ | |
159 | ||
160 | int | |
161 | yyparse() | |
162 | { | |
373e2177 | 163 | if (! cpp_start_read (parse_in, cpp_filename)) |
3aac38d7 RH |
164 | return 1; /* cpplib has emitted an error. */ |
165 | ||
166 | return yyparse_1(); | |
167 | } | |
168 | ||
0e5921e8 ZW |
169 | struct c_fileinfo * |
170 | get_fileinfo (name) | |
171 | const char *name; | |
e3d1fd32 | 172 | { |
0e5921e8 ZW |
173 | splay_tree_node n; |
174 | struct c_fileinfo *fi; | |
175 | ||
176 | n = splay_tree_lookup (file_info_tree, (splay_tree_key) name); | |
177 | if (n) | |
178 | return (struct c_fileinfo *) n->value; | |
179 | ||
180 | fi = (struct c_fileinfo *) xmalloc (sizeof (struct c_fileinfo)); | |
181 | fi->time = 0; | |
182 | fi->interface_only = 0; | |
183 | fi->interface_unknown = 1; | |
184 | splay_tree_insert (file_info_tree, (splay_tree_key) name, | |
185 | (splay_tree_value) fi); | |
186 | return fi; | |
e56e519d | 187 | } |
e3d1fd32 | 188 | |
0e5921e8 ZW |
189 | static void |
190 | update_header_times (name) | |
191 | const char *name; | |
e8bbfc4e | 192 | { |
0e5921e8 ZW |
193 | /* Changing files again. This means currently collected time |
194 | is charged against header time, and body time starts back at 0. */ | |
195 | if (flag_detailed_statistics) | |
e8bbfc4e | 196 | { |
0e5921e8 ZW |
197 | int this_time = get_run_time (); |
198 | struct c_fileinfo *file = get_fileinfo (name); | |
199 | header_time += this_time - body_time; | |
200 | file->time += this_time - body_time; | |
201 | body_time = this_time; | |
e8bbfc4e RK |
202 | } |
203 | } | |
204 | ||
0e5921e8 ZW |
205 | static int |
206 | dump_one_header (n, dummy) | |
207 | splay_tree_node n; | |
208 | void *dummy ATTRIBUTE_UNUSED; | |
e8bbfc4e | 209 | { |
0e5921e8 ZW |
210 | print_time ((const char *) n->key, |
211 | ((struct c_fileinfo *) n->value)->time); | |
212 | return 0; | |
e8bbfc4e | 213 | } |
e8bbfc4e RK |
214 | |
215 | void | |
0e5921e8 | 216 | dump_time_statistics () |
e8bbfc4e | 217 | { |
0e5921e8 ZW |
218 | struct c_fileinfo *file = get_fileinfo (input_filename); |
219 | int this_time = get_run_time (); | |
220 | file->time += this_time - body_time; | |
221 | ||
222 | fprintf (stderr, "\n******\n"); | |
223 | print_time ("header files (total)", header_time); | |
224 | print_time ("main file (total)", this_time - body_time); | |
225 | fprintf (stderr, "ratio = %g : 1\n", | |
226 | (double)header_time / (double)(this_time - body_time)); | |
227 | fprintf (stderr, "\n******\n"); | |
228 | ||
229 | splay_tree_foreach (file_info_tree, dump_one_header, 0); | |
e8bbfc4e | 230 | } |
a6124a42 | 231 | |
27e2564a NB |
232 | /* Not yet handled: #pragma, #define, #undef. |
233 | No need to deal with linemarkers under normal conditions. */ | |
234 | ||
0e5921e8 | 235 | static void |
8bbbef34 | 236 | cb_ident (pfile, line, str) |
27e2564a | 237 | cpp_reader *pfile ATTRIBUTE_UNUSED; |
8bbbef34 | 238 | unsigned int line ATTRIBUTE_UNUSED; |
f5720527 | 239 | const cpp_string *str ATTRIBUTE_UNUSED; |
0e5921e8 | 240 | { |
0e5921e8 | 241 | #ifdef ASM_OUTPUT_IDENT |
27e2564a | 242 | if (! flag_no_ident) |
0e5921e8 | 243 | { |
27e2564a NB |
244 | /* Convert escapes in the string. */ |
245 | tree value = lex_string ((const char *)str->text, str->len, 0); | |
246 | ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (value)); | |
0e5921e8 ZW |
247 | } |
248 | #endif | |
27e2564a NB |
249 | } |
250 | ||
97293897 NB |
251 | /* Called at the start of every non-empty line. TOKEN is the first |
252 | lexed token on the line. Used for diagnostic line numbers. */ | |
253 | static void | |
254 | cb_line_change (pfile, token, parsing_args) | |
255 | cpp_reader *pfile ATTRIBUTE_UNUSED; | |
256 | const cpp_token *token; | |
257 | int parsing_args ATTRIBUTE_UNUSED; | |
258 | { | |
259 | src_lineno = SOURCE_LINE (map, token->line); | |
260 | } | |
261 | ||
27e2564a | 262 | static void |
47d89cf3 | 263 | cb_file_change (pfile, new_map) |
27e2564a | 264 | cpp_reader *pfile ATTRIBUTE_UNUSED; |
47d89cf3 | 265 | const struct line_map *new_map; |
27e2564a | 266 | { |
47d89cf3 | 267 | unsigned int to_line = SOURCE_LINE (new_map, new_map->to_line); |
d82fc108 | 268 | |
47d89cf3 | 269 | if (new_map->reason == LC_ENTER) |
fbb18613 | 270 | { |
5bea1ccf JM |
271 | /* Don't stack the main buffer on the input stack; |
272 | we already did in compile_file. */ | |
47d89cf3 NB |
273 | if (map == NULL) |
274 | main_input_filename = new_map->to_file; | |
d82fc108 | 275 | else |
0e5921e8 | 276 | { |
47d89cf3 NB |
277 | lineno = SOURCE_LINE (new_map - 1, new_map->from_line - 1); |
278 | push_srcloc (new_map->to_file, 1); | |
27e2564a | 279 | input_file_stack->indent_level = indent_level; |
47d89cf3 | 280 | (*debug_hooks->start_source_file) (lineno, new_map->to_file); |
27e2564a NB |
281 | #ifndef NO_IMPLICIT_EXTERN_C |
282 | if (c_header_level) | |
283 | ++c_header_level; | |
47d89cf3 | 284 | else if (new_map->sysp == 2) |
27e2564a NB |
285 | { |
286 | c_header_level = 1; | |
287 | ++pending_lang_change; | |
288 | } | |
0e5921e8 | 289 | #endif |
27e2564a | 290 | } |
fbb18613 | 291 | } |
47d89cf3 | 292 | else if (new_map->reason == LC_LEAVE) |
fbb18613 | 293 | { |
0e5921e8 | 294 | #ifndef NO_IMPLICIT_EXTERN_C |
47d89cf3 NB |
295 | if (c_header_level && --c_header_level == 0) |
296 | { | |
297 | if (new_map->sysp == 2) | |
298 | warning ("badly nested C headers from preprocessor"); | |
299 | --pending_lang_change; | |
300 | } | |
0e5921e8 ZW |
301 | #endif |
302 | #if 0 | |
47d89cf3 NB |
303 | if (indent_level != input_file_stack->indent_level) |
304 | { | |
305 | warning_with_file_and_line | |
306 | (input_filename, lineno, | |
307 | "This file contains more '%c's than '%c's.", | |
308 | indent_level > input_file_stack->indent_level ? '{' : '}', | |
309 | indent_level > input_file_stack->indent_level ? '}' : '{'); | |
0468bc75 | 310 | } |
47d89cf3 NB |
311 | #endif |
312 | pop_srcloc (); | |
313 | ||
314 | (*debug_hooks->end_source_file) (to_line); | |
e8bbfc4e | 315 | } |
fbb18613 | 316 | |
47d89cf3 NB |
317 | update_header_times (new_map->to_file); |
318 | in_system_header = new_map->sysp != 0; | |
319 | input_filename = new_map->to_file; | |
320 | lineno = to_line; | |
321 | map = new_map; | |
0e5921e8 | 322 | |
0e5921e8 ZW |
323 | /* Hook for C++. */ |
324 | extract_interface_info (); | |
325 | } | |
8b97c5f8 ZW |
326 | |
327 | static void | |
8bbbef34 | 328 | cb_def_pragma (pfile, line) |
8b97c5f8 | 329 | cpp_reader *pfile; |
67821e3a | 330 | unsigned int line; |
8b97c5f8 ZW |
331 | { |
332 | /* Issue a warning message if we have been asked to do so. Ignore | |
333 | unknown pragmas in system headers unless an explicit | |
ec5c56db | 334 | -Wunknown-pragmas has been given. */ |
8b97c5f8 ZW |
335 | if (warn_unknown_pragmas > in_system_header) |
336 | { | |
23356f93 | 337 | const unsigned char *space, *name = 0; |
4ed5bcfb | 338 | const cpp_token *s; |
23356f93 | 339 | |
4ed5bcfb NB |
340 | s = cpp_get_token (pfile); |
341 | space = cpp_token_as_text (pfile, s); | |
342 | s = cpp_get_token (pfile); | |
343 | if (s->type == CPP_NAME) | |
344 | name = cpp_token_as_text (pfile, s); | |
8b97c5f8 | 345 | |
67821e3a | 346 | lineno = SOURCE_LINE (map, line); |
8b97c5f8 ZW |
347 | if (name) |
348 | warning ("ignoring #pragma %s %s", space, name); | |
349 | else | |
350 | warning ("ignoring #pragma %s", space); | |
351 | } | |
352 | } | |
0e5921e8 | 353 | |
65289a3a NB |
354 | /* #define callback for DWARF and DWARF2 debug info. */ |
355 | static void | |
8bbbef34 | 356 | cb_define (pfile, line, node) |
65289a3a | 357 | cpp_reader *pfile; |
67821e3a | 358 | unsigned int line; |
65289a3a NB |
359 | cpp_hashnode *node; |
360 | { | |
67821e3a | 361 | (*debug_hooks->define) (SOURCE_LINE (map, line), |
7f905405 | 362 | (const char *) cpp_macro_definition (pfile, node)); |
65289a3a NB |
363 | } |
364 | ||
365 | /* #undef callback for DWARF and DWARF2 debug info. */ | |
366 | static void | |
8bbbef34 | 367 | cb_undef (pfile, line, node) |
67821e3a NB |
368 | cpp_reader *pfile ATTRIBUTE_UNUSED; |
369 | unsigned int line; | |
65289a3a NB |
370 | cpp_hashnode *node; |
371 | { | |
67821e3a | 372 | (*debug_hooks->undef) (SOURCE_LINE (map, line), |
7f905405 | 373 | (const char *) NODE_NAME (node)); |
65289a3a NB |
374 | } |
375 | ||
0e5921e8 ZW |
376 | #if 0 /* not yet */ |
377 | /* Returns nonzero if C is a universal-character-name. Give an error if it | |
378 | is not one which may appear in an identifier, as per [extendid]. | |
379 | ||
380 | Note that extended character support in identifiers has not yet been | |
381 | implemented. It is my personal opinion that this is not a desirable | |
382 | feature. Portable code cannot count on support for more than the basic | |
383 | identifier character set. */ | |
384 | ||
385 | static inline int | |
386 | is_extended_char (c) | |
387 | int c; | |
388 | { | |
389 | #ifdef TARGET_EBCDIC | |
390 | return 0; | |
391 | #else | |
392 | /* ASCII. */ | |
393 | if (c < 0x7f) | |
394 | return 0; | |
395 | ||
396 | /* None of the valid chars are outside the Basic Multilingual Plane (the | |
397 | low 16 bits). */ | |
398 | if (c > 0xffff) | |
399 | { | |
400 | error ("universal-character-name '\\U%08x' not valid in identifier", c); | |
401 | return 1; | |
402 | } | |
403 | ||
404 | /* Latin */ | |
405 | if ((c >= 0x00c0 && c <= 0x00d6) | |
406 | || (c >= 0x00d8 && c <= 0x00f6) | |
407 | || (c >= 0x00f8 && c <= 0x01f5) | |
408 | || (c >= 0x01fa && c <= 0x0217) | |
409 | || (c >= 0x0250 && c <= 0x02a8) | |
410 | || (c >= 0x1e00 && c <= 0x1e9a) | |
411 | || (c >= 0x1ea0 && c <= 0x1ef9)) | |
412 | return 1; | |
413 | ||
414 | /* Greek */ | |
415 | if ((c == 0x0384) | |
416 | || (c >= 0x0388 && c <= 0x038a) | |
417 | || (c == 0x038c) | |
418 | || (c >= 0x038e && c <= 0x03a1) | |
419 | || (c >= 0x03a3 && c <= 0x03ce) | |
420 | || (c >= 0x03d0 && c <= 0x03d6) | |
421 | || (c == 0x03da) | |
422 | || (c == 0x03dc) | |
423 | || (c == 0x03de) | |
424 | || (c == 0x03e0) | |
425 | || (c >= 0x03e2 && c <= 0x03f3) | |
426 | || (c >= 0x1f00 && c <= 0x1f15) | |
427 | || (c >= 0x1f18 && c <= 0x1f1d) | |
428 | || (c >= 0x1f20 && c <= 0x1f45) | |
429 | || (c >= 0x1f48 && c <= 0x1f4d) | |
430 | || (c >= 0x1f50 && c <= 0x1f57) | |
431 | || (c == 0x1f59) | |
432 | || (c == 0x1f5b) | |
433 | || (c == 0x1f5d) | |
434 | || (c >= 0x1f5f && c <= 0x1f7d) | |
435 | || (c >= 0x1f80 && c <= 0x1fb4) | |
436 | || (c >= 0x1fb6 && c <= 0x1fbc) | |
437 | || (c >= 0x1fc2 && c <= 0x1fc4) | |
438 | || (c >= 0x1fc6 && c <= 0x1fcc) | |
439 | || (c >= 0x1fd0 && c <= 0x1fd3) | |
440 | || (c >= 0x1fd6 && c <= 0x1fdb) | |
441 | || (c >= 0x1fe0 && c <= 0x1fec) | |
442 | || (c >= 0x1ff2 && c <= 0x1ff4) | |
443 | || (c >= 0x1ff6 && c <= 0x1ffc)) | |
444 | return 1; | |
445 | ||
446 | /* Cyrillic */ | |
447 | if ((c >= 0x0401 && c <= 0x040d) | |
448 | || (c >= 0x040f && c <= 0x044f) | |
449 | || (c >= 0x0451 && c <= 0x045c) | |
450 | || (c >= 0x045e && c <= 0x0481) | |
451 | || (c >= 0x0490 && c <= 0x04c4) | |
452 | || (c >= 0x04c7 && c <= 0x04c8) | |
453 | || (c >= 0x04cb && c <= 0x04cc) | |
454 | || (c >= 0x04d0 && c <= 0x04eb) | |
455 | || (c >= 0x04ee && c <= 0x04f5) | |
456 | || (c >= 0x04f8 && c <= 0x04f9)) | |
457 | return 1; | |
458 | ||
459 | /* Armenian */ | |
460 | if ((c >= 0x0531 && c <= 0x0556) | |
461 | || (c >= 0x0561 && c <= 0x0587)) | |
462 | return 1; | |
463 | ||
464 | /* Hebrew */ | |
465 | if ((c >= 0x05d0 && c <= 0x05ea) | |
466 | || (c >= 0x05f0 && c <= 0x05f4)) | |
467 | return 1; | |
468 | ||
469 | /* Arabic */ | |
470 | if ((c >= 0x0621 && c <= 0x063a) | |
471 | || (c >= 0x0640 && c <= 0x0652) | |
472 | || (c >= 0x0670 && c <= 0x06b7) | |
473 | || (c >= 0x06ba && c <= 0x06be) | |
474 | || (c >= 0x06c0 && c <= 0x06ce) | |
475 | || (c >= 0x06e5 && c <= 0x06e7)) | |
476 | return 1; | |
477 | ||
478 | /* Devanagari */ | |
479 | if ((c >= 0x0905 && c <= 0x0939) | |
480 | || (c >= 0x0958 && c <= 0x0962)) | |
481 | return 1; | |
482 | ||
483 | /* Bengali */ | |
484 | if ((c >= 0x0985 && c <= 0x098c) | |
485 | || (c >= 0x098f && c <= 0x0990) | |
486 | || (c >= 0x0993 && c <= 0x09a8) | |
487 | || (c >= 0x09aa && c <= 0x09b0) | |
488 | || (c == 0x09b2) | |
489 | || (c >= 0x09b6 && c <= 0x09b9) | |
490 | || (c >= 0x09dc && c <= 0x09dd) | |
491 | || (c >= 0x09df && c <= 0x09e1) | |
492 | || (c >= 0x09f0 && c <= 0x09f1)) | |
493 | return 1; | |
494 | ||
495 | /* Gurmukhi */ | |
496 | if ((c >= 0x0a05 && c <= 0x0a0a) | |
497 | || (c >= 0x0a0f && c <= 0x0a10) | |
498 | || (c >= 0x0a13 && c <= 0x0a28) | |
499 | || (c >= 0x0a2a && c <= 0x0a30) | |
500 | || (c >= 0x0a32 && c <= 0x0a33) | |
501 | || (c >= 0x0a35 && c <= 0x0a36) | |
502 | || (c >= 0x0a38 && c <= 0x0a39) | |
503 | || (c >= 0x0a59 && c <= 0x0a5c) | |
504 | || (c == 0x0a5e)) | |
505 | return 1; | |
506 | ||
507 | /* Gujarati */ | |
508 | if ((c >= 0x0a85 && c <= 0x0a8b) | |
509 | || (c == 0x0a8d) | |
510 | || (c >= 0x0a8f && c <= 0x0a91) | |
511 | || (c >= 0x0a93 && c <= 0x0aa8) | |
512 | || (c >= 0x0aaa && c <= 0x0ab0) | |
513 | || (c >= 0x0ab2 && c <= 0x0ab3) | |
514 | || (c >= 0x0ab5 && c <= 0x0ab9) | |
515 | || (c == 0x0ae0)) | |
516 | return 1; | |
517 | ||
518 | /* Oriya */ | |
519 | if ((c >= 0x0b05 && c <= 0x0b0c) | |
520 | || (c >= 0x0b0f && c <= 0x0b10) | |
521 | || (c >= 0x0b13 && c <= 0x0b28) | |
522 | || (c >= 0x0b2a && c <= 0x0b30) | |
523 | || (c >= 0x0b32 && c <= 0x0b33) | |
524 | || (c >= 0x0b36 && c <= 0x0b39) | |
525 | || (c >= 0x0b5c && c <= 0x0b5d) | |
526 | || (c >= 0x0b5f && c <= 0x0b61)) | |
527 | return 1; | |
528 | ||
529 | /* Tamil */ | |
530 | if ((c >= 0x0b85 && c <= 0x0b8a) | |
531 | || (c >= 0x0b8e && c <= 0x0b90) | |
532 | || (c >= 0x0b92 && c <= 0x0b95) | |
533 | || (c >= 0x0b99 && c <= 0x0b9a) | |
534 | || (c == 0x0b9c) | |
535 | || (c >= 0x0b9e && c <= 0x0b9f) | |
536 | || (c >= 0x0ba3 && c <= 0x0ba4) | |
537 | || (c >= 0x0ba8 && c <= 0x0baa) | |
538 | || (c >= 0x0bae && c <= 0x0bb5) | |
539 | || (c >= 0x0bb7 && c <= 0x0bb9)) | |
540 | return 1; | |
541 | ||
542 | /* Telugu */ | |
543 | if ((c >= 0x0c05 && c <= 0x0c0c) | |
544 | || (c >= 0x0c0e && c <= 0x0c10) | |
545 | || (c >= 0x0c12 && c <= 0x0c28) | |
546 | || (c >= 0x0c2a && c <= 0x0c33) | |
547 | || (c >= 0x0c35 && c <= 0x0c39) | |
548 | || (c >= 0x0c60 && c <= 0x0c61)) | |
549 | return 1; | |
550 | ||
551 | /* Kannada */ | |
552 | if ((c >= 0x0c85 && c <= 0x0c8c) | |
553 | || (c >= 0x0c8e && c <= 0x0c90) | |
554 | || (c >= 0x0c92 && c <= 0x0ca8) | |
555 | || (c >= 0x0caa && c <= 0x0cb3) | |
556 | || (c >= 0x0cb5 && c <= 0x0cb9) | |
557 | || (c >= 0x0ce0 && c <= 0x0ce1)) | |
558 | return 1; | |
559 | ||
560 | /* Malayalam */ | |
561 | if ((c >= 0x0d05 && c <= 0x0d0c) | |
562 | || (c >= 0x0d0e && c <= 0x0d10) | |
563 | || (c >= 0x0d12 && c <= 0x0d28) | |
564 | || (c >= 0x0d2a && c <= 0x0d39) | |
565 | || (c >= 0x0d60 && c <= 0x0d61)) | |
566 | return 1; | |
567 | ||
568 | /* Thai */ | |
569 | if ((c >= 0x0e01 && c <= 0x0e30) | |
570 | || (c >= 0x0e32 && c <= 0x0e33) | |
571 | || (c >= 0x0e40 && c <= 0x0e46) | |
572 | || (c >= 0x0e4f && c <= 0x0e5b)) | |
573 | return 1; | |
574 | ||
575 | /* Lao */ | |
576 | if ((c >= 0x0e81 && c <= 0x0e82) | |
577 | || (c == 0x0e84) | |
578 | || (c == 0x0e87) | |
579 | || (c == 0x0e88) | |
580 | || (c == 0x0e8a) | |
581 | || (c == 0x0e0d) | |
582 | || (c >= 0x0e94 && c <= 0x0e97) | |
583 | || (c >= 0x0e99 && c <= 0x0e9f) | |
584 | || (c >= 0x0ea1 && c <= 0x0ea3) | |
585 | || (c == 0x0ea5) | |
586 | || (c == 0x0ea7) | |
587 | || (c == 0x0eaa) | |
588 | || (c == 0x0eab) | |
589 | || (c >= 0x0ead && c <= 0x0eb0) | |
590 | || (c == 0x0eb2) | |
591 | || (c == 0x0eb3) | |
592 | || (c == 0x0ebd) | |
593 | || (c >= 0x0ec0 && c <= 0x0ec4) | |
594 | || (c == 0x0ec6)) | |
595 | return 1; | |
596 | ||
597 | /* Georgian */ | |
598 | if ((c >= 0x10a0 && c <= 0x10c5) | |
599 | || (c >= 0x10d0 && c <= 0x10f6)) | |
600 | return 1; | |
601 | ||
602 | /* Hiragana */ | |
603 | if ((c >= 0x3041 && c <= 0x3094) | |
604 | || (c >= 0x309b && c <= 0x309e)) | |
605 | return 1; | |
606 | ||
607 | /* Katakana */ | |
608 | if ((c >= 0x30a1 && c <= 0x30fe)) | |
609 | return 1; | |
610 | ||
611 | /* Bopmofo */ | |
612 | if ((c >= 0x3105 && c <= 0x312c)) | |
613 | return 1; | |
614 | ||
615 | /* Hangul */ | |
616 | if ((c >= 0x1100 && c <= 0x1159) | |
617 | || (c >= 0x1161 && c <= 0x11a2) | |
618 | || (c >= 0x11a8 && c <= 0x11f9)) | |
619 | return 1; | |
620 | ||
621 | /* CJK Unified Ideographs */ | |
622 | if ((c >= 0xf900 && c <= 0xfa2d) | |
623 | || (c >= 0xfb1f && c <= 0xfb36) | |
624 | || (c >= 0xfb38 && c <= 0xfb3c) | |
625 | || (c == 0xfb3e) | |
626 | || (c >= 0xfb40 && c <= 0xfb41) | |
627 | || (c >= 0xfb42 && c <= 0xfb44) | |
628 | || (c >= 0xfb46 && c <= 0xfbb1) | |
629 | || (c >= 0xfbd3 && c <= 0xfd3f) | |
630 | || (c >= 0xfd50 && c <= 0xfd8f) | |
631 | || (c >= 0xfd92 && c <= 0xfdc7) | |
632 | || (c >= 0xfdf0 && c <= 0xfdfb) | |
633 | || (c >= 0xfe70 && c <= 0xfe72) | |
634 | || (c == 0xfe74) | |
635 | || (c >= 0xfe76 && c <= 0xfefc) | |
636 | || (c >= 0xff21 && c <= 0xff3a) | |
637 | || (c >= 0xff41 && c <= 0xff5a) | |
638 | || (c >= 0xff66 && c <= 0xffbe) | |
639 | || (c >= 0xffc2 && c <= 0xffc7) | |
640 | || (c >= 0xffca && c <= 0xffcf) | |
641 | || (c >= 0xffd2 && c <= 0xffd7) | |
642 | || (c >= 0xffda && c <= 0xffdc) | |
643 | || (c >= 0x4e00 && c <= 0x9fa5)) | |
644 | return 1; | |
645 | ||
646 | error ("universal-character-name '\\u%04x' not valid in identifier", c); | |
647 | return 1; | |
648 | #endif | |
649 | } | |
650 | ||
651 | /* Add the UTF-8 representation of C to the token_buffer. */ | |
652 | ||
653 | static void | |
654 | utf8_extend_token (c) | |
655 | int c; | |
e8bbfc4e | 656 | { |
0e5921e8 ZW |
657 | int shift, mask; |
658 | ||
659 | if (c <= 0x0000007f) | |
660 | { | |
661 | extend_token (c); | |
662 | return; | |
663 | } | |
664 | else if (c <= 0x000007ff) | |
665 | shift = 6, mask = 0xc0; | |
666 | else if (c <= 0x0000ffff) | |
667 | shift = 12, mask = 0xe0; | |
668 | else if (c <= 0x001fffff) | |
669 | shift = 18, mask = 0xf0; | |
670 | else if (c <= 0x03ffffff) | |
671 | shift = 24, mask = 0xf8; | |
e8bbfc4e | 672 | else |
0e5921e8 ZW |
673 | shift = 30, mask = 0xfc; |
674 | ||
675 | extend_token (mask | (c >> shift)); | |
676 | do | |
677 | { | |
678 | shift -= 6; | |
679 | extend_token ((unsigned char) (0x80 | (c >> shift))); | |
680 | } | |
681 | while (shift); | |
e8bbfc4e | 682 | } |
0e5921e8 | 683 | #endif |
e8bbfc4e RK |
684 | |
685 | #if 0 | |
e8bbfc4e RK |
686 | struct try_type |
687 | { | |
8b60264b KG |
688 | tree *const node_var; |
689 | const char unsigned_flag; | |
690 | const char long_flag; | |
691 | const char long_long_flag; | |
e8bbfc4e RK |
692 | }; |
693 | ||
75cb8865 | 694 | struct try_type type_sequence[] = |
e8bbfc4e RK |
695 | { |
696 | { &integer_type_node, 0, 0, 0}, | |
697 | { &unsigned_type_node, 1, 0, 0}, | |
698 | { &long_integer_type_node, 0, 1, 0}, | |
699 | { &long_unsigned_type_node, 1, 1, 0}, | |
700 | { &long_long_integer_type_node, 0, 1, 1}, | |
701 | { &long_long_unsigned_type_node, 1, 1, 1} | |
702 | }; | |
703 | #endif /* 0 */ | |
704 | \f | |
ed513abf KG |
705 | struct pf_args |
706 | { | |
707 | /* Input */ | |
0e5921e8 ZW |
708 | const char *str; |
709 | int fflag; | |
710 | int lflag; | |
ed513abf | 711 | int base; |
fbb18613 | 712 | /* Output */ |
ed513abf | 713 | int conversion_errno; |
ed513abf | 714 | REAL_VALUE_TYPE value; |
0e5921e8 | 715 | tree type; |
ed513abf KG |
716 | }; |
717 | ||
718 | static void | |
719 | parse_float (data) | |
720 | PTR data; | |
721 | { | |
722 | struct pf_args * args = (struct pf_args *) data; | |
0e5921e8 ZW |
723 | const char *typename; |
724 | ||
fbb18613 JM |
725 | args->conversion_errno = 0; |
726 | args->type = double_type_node; | |
0e5921e8 | 727 | typename = "double"; |
ed513abf KG |
728 | |
729 | /* The second argument, machine_mode, of REAL_VALUE_ATOF | |
730 | tells the desired precision of the binary result | |
731 | of decimal-to-binary conversion. */ | |
732 | ||
0e5921e8 | 733 | if (args->fflag) |
ed513abf | 734 | { |
0e5921e8 ZW |
735 | if (args->lflag) |
736 | error ("both 'f' and 'l' suffixes on floating constant"); | |
ed513abf KG |
737 | |
738 | args->type = float_type_node; | |
0e5921e8 | 739 | typename = "float"; |
ed513abf | 740 | } |
0e5921e8 | 741 | else if (args->lflag) |
ed513abf KG |
742 | { |
743 | args->type = long_double_type_node; | |
0e5921e8 | 744 | typename = "long double"; |
ed513abf | 745 | } |
0e5921e8 | 746 | else if (flag_single_precision_constant) |
ed513abf | 747 | { |
0e5921e8 ZW |
748 | args->type = float_type_node; |
749 | typename = "float"; | |
ed513abf | 750 | } |
0e5921e8 ZW |
751 | |
752 | errno = 0; | |
753 | if (args->base == 16) | |
754 | args->value = REAL_VALUE_HTOF (args->str, TYPE_MODE (args->type)); | |
755 | else | |
756 | args->value = REAL_VALUE_ATOF (args->str, TYPE_MODE (args->type)); | |
757 | ||
758 | args->conversion_errno = errno; | |
759 | /* A diagnostic is required here by some ISO C testsuites. | |
760 | This is not pedwarn, because some people don't want | |
761 | an error for this. */ | |
762 | if (REAL_VALUE_ISINF (args->value) && pedantic) | |
763 | warning ("floating point number exceeds range of '%s'", typename); | |
ed513abf KG |
764 | } |
765 | ||
0e5921e8 ZW |
766 | int |
767 | c_lex (value) | |
768 | tree *value; | |
fbb18613 | 769 | { |
4ed5bcfb | 770 | const cpp_token *tok; |
0e5921e8 ZW |
771 | |
772 | retry: | |
773 | timevar_push (TV_CPP); | |
4ed5bcfb NB |
774 | do |
775 | tok = cpp_get_token (parse_in); | |
776 | while (tok->type == CPP_PADDING); | |
0e5921e8 ZW |
777 | timevar_pop (TV_CPP); |
778 | ||
779 | /* The C++ front end does horrible things with the current line | |
780 | number. To ensure an accurate line number, we must reset it | |
23356f93 | 781 | every time we return a token. */ |
97293897 | 782 | lineno = src_lineno; |
0e5921e8 ZW |
783 | |
784 | *value = NULL_TREE; | |
4ed5bcfb | 785 | switch (tok->type) |
0e5921e8 ZW |
786 | { |
787 | case CPP_OPEN_BRACE: indent_level++; break; | |
788 | case CPP_CLOSE_BRACE: indent_level--; break; | |
789 | ||
4ed5bcfb | 790 | /* Issue this error here, where we can get at tok->val.c. */ |
0e5921e8 | 791 | case CPP_OTHER: |
4ed5bcfb NB |
792 | if (ISGRAPH (tok->val.c)) |
793 | error ("stray '%c' in program", tok->val.c); | |
0e5921e8 | 794 | else |
4ed5bcfb | 795 | error ("stray '\\%o' in program", tok->val.c); |
0e5921e8 ZW |
796 | goto retry; |
797 | ||
0e5921e8 | 798 | case CPP_NAME: |
4ed5bcfb | 799 | *value = HT_IDENT_TO_GCC_IDENT (HT_NODE (tok->val.node)); |
0e5921e8 | 800 | break; |
fbb18613 | 801 | |
0e5921e8 | 802 | case CPP_NUMBER: |
4ed5bcfb | 803 | *value = lex_number ((const char *)tok->val.str.text, tok->val.str.len); |
0e5921e8 | 804 | break; |
93868d11 | 805 | |
0e5921e8 ZW |
806 | case CPP_CHAR: |
807 | case CPP_WCHAR: | |
4ed5bcfb | 808 | *value = lex_charconst (tok); |
0e5921e8 | 809 | break; |
fbb18613 | 810 | |
0e5921e8 ZW |
811 | case CPP_STRING: |
812 | case CPP_WSTRING: | |
4ed5bcfb NB |
813 | *value = lex_string ((const char *)tok->val.str.text, |
814 | tok->val.str.len, tok->type == CPP_WSTRING); | |
0e5921e8 | 815 | break; |
fbb18613 | 816 | |
0e5921e8 ZW |
817 | /* These tokens should not be visible outside cpplib. */ |
818 | case CPP_HEADER_NAME: | |
819 | case CPP_COMMENT: | |
820 | case CPP_MACRO_ARG: | |
0e5921e8 ZW |
821 | abort (); |
822 | ||
823 | default: break; | |
824 | } | |
825 | ||
4ed5bcfb | 826 | return tok->type; |
0e5921e8 | 827 | } |
8d9bfdc5 | 828 | |
0e5921e8 | 829 | #define ERROR(msgid) do { error(msgid); goto syntax_error; } while(0) |
75cb8865 | 830 | |
0e5921e8 ZW |
831 | static tree |
832 | lex_number (str, len) | |
833 | const char *str; | |
834 | unsigned int len; | |
835 | { | |
836 | int base = 10; | |
837 | int count = 0; | |
838 | int largest_digit = 0; | |
839 | int numdigits = 0; | |
840 | int overflow = 0; | |
841 | int c; | |
842 | tree value; | |
843 | const char *p; | |
844 | enum anon1 { NOT_FLOAT = 0, AFTER_POINT, AFTER_EXPON } floatflag = NOT_FLOAT; | |
845 | ||
846 | /* We actually store only HOST_BITS_PER_CHAR bits in each part. | |
847 | The code below which fills the parts array assumes that a host | |
848 | int is at least twice as wide as a host char, and that | |
849 | HOST_BITS_PER_WIDE_INT is an even multiple of HOST_BITS_PER_CHAR. | |
850 | Two HOST_WIDE_INTs is the largest int literal we can store. | |
851 | In order to detect overflow below, the number of parts (TOTAL_PARTS) | |
852 | must be exactly the number of parts needed to hold the bits | |
ec5c56db | 853 | of two HOST_WIDE_INTs. */ |
0e5921e8 ZW |
854 | #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2) |
855 | unsigned int parts[TOTAL_PARTS]; | |
856 | ||
857 | /* Optimize for most frequent case. */ | |
858 | if (len == 1) | |
859 | { | |
860 | if (*str == '0') | |
861 | return integer_zero_node; | |
862 | else if (*str == '1') | |
863 | return integer_one_node; | |
864 | else | |
865 | return build_int_2 (*str - '0', 0); | |
866 | } | |
e8bbfc4e | 867 | |
0e5921e8 ZW |
868 | for (count = 0; count < TOTAL_PARTS; count++) |
869 | parts[count] = 0; | |
e8bbfc4e | 870 | |
0e5921e8 ZW |
871 | /* len is known to be >1 at this point. */ |
872 | p = str; | |
e8bbfc4e | 873 | |
0e5921e8 ZW |
874 | if (len > 2 && str[0] == '0' && (str[1] == 'x' || str[1] == 'X')) |
875 | { | |
876 | base = 16; | |
877 | p = str + 2; | |
878 | } | |
879 | /* The ISDIGIT check is so we are not confused by a suffix on 0. */ | |
880 | else if (str[0] == '0' && ISDIGIT (str[1])) | |
881 | { | |
882 | base = 8; | |
883 | p = str + 1; | |
884 | } | |
e8bbfc4e | 885 | |
0e5921e8 ZW |
886 | do |
887 | { | |
888 | c = *p++; | |
88d92ca5 | 889 | |
0e5921e8 ZW |
890 | if (c == '.') |
891 | { | |
c215b29f | 892 | if (floatflag == AFTER_POINT) |
0e5921e8 ZW |
893 | ERROR ("too many decimal points in floating constant"); |
894 | else if (floatflag == AFTER_EXPON) | |
895 | ERROR ("decimal point in exponent - impossible!"); | |
896 | else | |
897 | floatflag = AFTER_POINT; | |
c832a30e | 898 | |
0e5921e8 ZW |
899 | if (base == 8) |
900 | base = 10; | |
901 | } | |
902 | else if (c == '_') | |
903 | /* Possible future extension: silently ignore _ in numbers, | |
904 | permitting cosmetic grouping - e.g. 0x8000_0000 == 0x80000000 | |
905 | but somewhat easier to read. Ada has this? */ | |
906 | ERROR ("underscore in number"); | |
907 | else | |
908 | { | |
909 | int n; | |
910 | /* It is not a decimal point. | |
911 | It should be a digit (perhaps a hex digit). */ | |
0e0fda0d | 912 | |
0e5921e8 ZW |
913 | if (ISDIGIT (c)) |
914 | { | |
915 | n = c - '0'; | |
916 | } | |
917 | else if (base <= 10 && (c == 'e' || c == 'E')) | |
918 | { | |
919 | base = 10; | |
920 | floatflag = AFTER_EXPON; | |
921 | break; | |
922 | } | |
923 | else if (base == 16 && (c == 'p' || c == 'P')) | |
924 | { | |
925 | floatflag = AFTER_EXPON; | |
926 | break; /* start of exponent */ | |
927 | } | |
928 | else if (base == 16 && c >= 'a' && c <= 'f') | |
929 | { | |
930 | n = c - 'a' + 10; | |
931 | } | |
932 | else if (base == 16 && c >= 'A' && c <= 'F') | |
933 | { | |
934 | n = c - 'A' + 10; | |
935 | } | |
936 | else | |
937 | { | |
938 | p--; | |
939 | break; /* start of suffix */ | |
940 | } | |
0e0fda0d | 941 | |
0e5921e8 ZW |
942 | if (n >= largest_digit) |
943 | largest_digit = n; | |
944 | numdigits++; | |
0e0fda0d | 945 | |
0e5921e8 ZW |
946 | for (count = 0; count < TOTAL_PARTS; count++) |
947 | { | |
948 | parts[count] *= base; | |
949 | if (count) | |
950 | { | |
951 | parts[count] | |
952 | += (parts[count-1] >> HOST_BITS_PER_CHAR); | |
953 | parts[count-1] | |
954 | &= (1 << HOST_BITS_PER_CHAR) - 1; | |
955 | } | |
956 | else | |
957 | parts[0] += n; | |
958 | } | |
e8bbfc4e | 959 | |
0e5921e8 ZW |
960 | /* If the highest-order part overflows (gets larger than |
961 | a host char will hold) then the whole number has | |
962 | overflowed. Record this and truncate the highest-order | |
ec5c56db | 963 | part. */ |
0e5921e8 ZW |
964 | if (parts[TOTAL_PARTS - 1] >> HOST_BITS_PER_CHAR) |
965 | { | |
966 | overflow = 1; | |
967 | parts[TOTAL_PARTS - 1] &= (1 << HOST_BITS_PER_CHAR) - 1; | |
968 | } | |
969 | } | |
970 | } | |
971 | while (p < str + len); | |
91b0989e | 972 | |
0e5921e8 ZW |
973 | /* This can happen on input like `int i = 0x;' */ |
974 | if (numdigits == 0) | |
975 | ERROR ("numeric constant with no digits"); | |
91b0989e | 976 | |
0e5921e8 ZW |
977 | if (largest_digit >= base) |
978 | ERROR ("numeric constant contains digits beyond the radix"); | |
e8bbfc4e | 979 | |
0e5921e8 ZW |
980 | if (floatflag != NOT_FLOAT) |
981 | { | |
982 | tree type; | |
983 | int imag, fflag, lflag, conversion_errno; | |
984 | REAL_VALUE_TYPE real; | |
985 | struct pf_args args; | |
986 | char *copy; | |
987 | ||
c215b29f JM |
988 | if (base == 16 && pedantic && !flag_isoc99) |
989 | pedwarn ("floating constant may not be in radix 16"); | |
990 | ||
0e5921e8 ZW |
991 | if (base == 16 && floatflag != AFTER_EXPON) |
992 | ERROR ("hexadecimal floating constant has no exponent"); | |
993 | ||
994 | /* Read explicit exponent if any, and put it in tokenbuf. */ | |
995 | if ((base == 10 && ((c == 'e') || (c == 'E'))) | |
996 | || (base == 16 && (c == 'p' || c == 'P'))) | |
997 | { | |
998 | if (p < str + len) | |
999 | c = *p++; | |
1000 | if (p < str + len && (c == '+' || c == '-')) | |
1001 | c = *p++; | |
1002 | /* Exponent is decimal, even if string is a hex float. */ | |
1003 | if (! ISDIGIT (c)) | |
1004 | ERROR ("floating constant exponent has no digits"); | |
1005 | while (p < str + len && ISDIGIT (c)) | |
1006 | c = *p++; | |
1007 | if (! ISDIGIT (c)) | |
1008 | p--; | |
1009 | } | |
56f48ce9 | 1010 | |
0e5921e8 ZW |
1011 | /* Copy the float constant now; we don't want any suffixes in the |
1012 | string passed to parse_float. */ | |
1013 | copy = alloca (p - str + 1); | |
1014 | memcpy (copy, str, p - str); | |
1015 | copy[p - str] = '\0'; | |
e8bbfc4e | 1016 | |
0e5921e8 ZW |
1017 | /* Now parse suffixes. */ |
1018 | fflag = lflag = imag = 0; | |
1019 | while (p < str + len) | |
1020 | switch (*p++) | |
e8bbfc4e | 1021 | { |
0e5921e8 ZW |
1022 | case 'f': case 'F': |
1023 | if (fflag) | |
1024 | ERROR ("more than one 'f' suffix on floating constant"); | |
7065e130 | 1025 | else if (warn_traditional && !in_system_header |
c691145a | 1026 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 | 1027 | warning ("traditional C rejects the 'f' suffix"); |
e8bbfc4e | 1028 | |
0e5921e8 ZW |
1029 | fflag = 1; |
1030 | break; | |
e8bbfc4e | 1031 | |
0e5921e8 ZW |
1032 | case 'l': case 'L': |
1033 | if (lflag) | |
1034 | ERROR ("more than one 'l' suffix on floating constant"); | |
7065e130 | 1035 | else if (warn_traditional && !in_system_header |
c691145a | 1036 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 | 1037 | warning ("traditional C rejects the 'l' suffix"); |
56f48ce9 | 1038 | |
0e5921e8 ZW |
1039 | lflag = 1; |
1040 | break; | |
e8bbfc4e | 1041 | |
0e5921e8 ZW |
1042 | case 'i': case 'I': |
1043 | case 'j': case 'J': | |
1044 | if (imag) | |
1045 | ERROR ("more than one 'i' or 'j' suffix on floating constant"); | |
1046 | else if (pedantic) | |
1047 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
1048 | imag = 1; | |
1049 | break; | |
e8bbfc4e | 1050 | |
0e5921e8 ZW |
1051 | default: |
1052 | ERROR ("invalid suffix on floating constant"); | |
e8bbfc4e RK |
1053 | } |
1054 | ||
0e5921e8 ZW |
1055 | /* Setup input for parse_float() */ |
1056 | args.str = copy; | |
1057 | args.fflag = fflag; | |
1058 | args.lflag = lflag; | |
1059 | args.base = base; | |
e8bbfc4e | 1060 | |
0e5921e8 ZW |
1061 | /* Convert string to a double, checking for overflow. */ |
1062 | if (do_float_handler (parse_float, (PTR) &args)) | |
1063 | { | |
1064 | /* Receive output from parse_float() */ | |
1065 | real = args.value; | |
1066 | } | |
1067 | else | |
1068 | /* We got an exception from parse_float() */ | |
1069 | ERROR ("floating constant out of range"); | |
e8bbfc4e | 1070 | |
0e5921e8 ZW |
1071 | /* Receive output from parse_float() */ |
1072 | conversion_errno = args.conversion_errno; | |
1073 | type = args.type; | |
1074 | ||
1075 | #ifdef ERANGE | |
1076 | /* ERANGE is also reported for underflow, | |
1077 | so test the value to distinguish overflow from that. */ | |
1078 | if (conversion_errno == ERANGE && !flag_traditional && pedantic | |
1079 | && (REAL_VALUES_LESS (dconst1, real) | |
1080 | || REAL_VALUES_LESS (real, dconstm1))) | |
1081 | warning ("floating point number exceeds range of 'double'"); | |
56f48ce9 | 1082 | #endif |
e8bbfc4e | 1083 | |
0e5921e8 ZW |
1084 | /* Create a node with determined type and value. */ |
1085 | if (imag) | |
1086 | value = build_complex (NULL_TREE, convert (type, integer_zero_node), | |
1087 | build_real (type, real)); | |
1088 | else | |
1089 | value = build_real (type, real); | |
1090 | } | |
1091 | else | |
1092 | { | |
1093 | tree trad_type, ansi_type, type; | |
1094 | HOST_WIDE_INT high, low; | |
1095 | int spec_unsigned = 0; | |
1096 | int spec_long = 0; | |
1097 | int spec_long_long = 0; | |
1098 | int spec_imag = 0; | |
1099 | int suffix_lu = 0; | |
1100 | int warn = 0, i; | |
1101 | ||
1102 | trad_type = ansi_type = type = NULL_TREE; | |
1103 | while (p < str + len) | |
1104 | { | |
1105 | c = *p++; | |
1106 | switch (c) | |
1107 | { | |
1108 | case 'u': case 'U': | |
1109 | if (spec_unsigned) | |
1110 | error ("two 'u' suffixes on integer constant"); | |
7065e130 | 1111 | else if (warn_traditional && !in_system_header |
c691145a | 1112 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 ZW |
1113 | warning ("traditional C rejects the 'u' suffix"); |
1114 | ||
1115 | spec_unsigned = 1; | |
1116 | if (spec_long) | |
1117 | suffix_lu = 1; | |
1118 | break; | |
e8bbfc4e | 1119 | |
0e5921e8 ZW |
1120 | case 'l': case 'L': |
1121 | if (spec_long) | |
1122 | { | |
1123 | if (spec_long_long) | |
1124 | error ("three 'l' suffixes on integer constant"); | |
1125 | else if (suffix_lu) | |
1126 | error ("'lul' is not a valid integer suffix"); | |
1127 | else if (c != spec_long) | |
1128 | error ("'Ll' and 'lL' are not valid integer suffixes"); | |
1129 | else if (pedantic && ! flag_isoc99 | |
1130 | && ! in_system_header && warn_long_long) | |
1131 | pedwarn ("ISO C89 forbids long long integer constants"); | |
1132 | spec_long_long = 1; | |
1133 | } | |
1134 | spec_long = c; | |
1135 | break; | |
56f48ce9 | 1136 | |
0e5921e8 ZW |
1137 | case 'i': case 'I': case 'j': case 'J': |
1138 | if (spec_imag) | |
1139 | error ("more than one 'i' or 'j' suffix on integer constant"); | |
1140 | else if (pedantic) | |
1141 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
1142 | spec_imag = 1; | |
1143 | break; | |
56f48ce9 | 1144 | |
0e5921e8 ZW |
1145 | default: |
1146 | ERROR ("invalid suffix on integer constant"); | |
1147 | } | |
1148 | } | |
56f48ce9 | 1149 | |
ec5c56db | 1150 | /* If the literal overflowed, pedwarn about it now. */ |
0e5921e8 ZW |
1151 | if (overflow) |
1152 | { | |
1153 | warn = 1; | |
1154 | pedwarn ("integer constant is too large for this configuration of the compiler - truncated to %d bits", HOST_BITS_PER_WIDE_INT * 2); | |
1155 | } | |
e8bbfc4e | 1156 | |
0e5921e8 ZW |
1157 | /* This is simplified by the fact that our constant |
1158 | is always positive. */ | |
56f48ce9 | 1159 | |
0e5921e8 | 1160 | high = low = 0; |
e8bbfc4e | 1161 | |
0e5921e8 ZW |
1162 | for (i = 0; i < HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR; i++) |
1163 | { | |
1164 | high |= ((HOST_WIDE_INT) parts[i + (HOST_BITS_PER_WIDE_INT | |
1165 | / HOST_BITS_PER_CHAR)] | |
1166 | << (i * HOST_BITS_PER_CHAR)); | |
1167 | low |= (HOST_WIDE_INT) parts[i] << (i * HOST_BITS_PER_CHAR); | |
1168 | } | |
0468bc75 | 1169 | |
0e5921e8 ZW |
1170 | value = build_int_2 (low, high); |
1171 | TREE_TYPE (value) = long_long_unsigned_type_node; | |
e8bbfc4e | 1172 | |
0e5921e8 ZW |
1173 | /* If warn_traditional, calculate both the ISO type and the |
1174 | traditional type, then see if they disagree. | |
1175 | Otherwise, calculate only the type for the dialect in use. */ | |
1176 | if (warn_traditional || flag_traditional) | |
1177 | { | |
1178 | /* Calculate the traditional type. */ | |
1179 | /* Traditionally, any constant is signed; but if unsigned is | |
1180 | specified explicitly, obey that. Use the smallest size | |
1181 | with the right number of bits, except for one special | |
1182 | case with decimal constants. */ | |
1183 | if (! spec_long && base != 10 | |
1184 | && int_fits_type_p (value, unsigned_type_node)) | |
1185 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
1186 | /* A decimal constant must be long if it does not fit in | |
1187 | type int. I think this is independent of whether the | |
1188 | constant is signed. */ | |
1189 | else if (! spec_long && base == 10 | |
1190 | && int_fits_type_p (value, integer_type_node)) | |
1191 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
1192 | else if (! spec_long_long) | |
1193 | trad_type = (spec_unsigned | |
1194 | ? long_unsigned_type_node | |
1195 | : long_integer_type_node); | |
1196 | else if (int_fits_type_p (value, | |
1197 | spec_unsigned | |
1198 | ? long_long_unsigned_type_node | |
1199 | : long_long_integer_type_node)) | |
1200 | trad_type = (spec_unsigned | |
1201 | ? long_long_unsigned_type_node | |
1202 | : long_long_integer_type_node); | |
1203 | else | |
1204 | trad_type = (spec_unsigned | |
1205 | ? widest_unsigned_literal_type_node | |
1206 | : widest_integer_literal_type_node); | |
1207 | } | |
1208 | if (warn_traditional || ! flag_traditional) | |
1209 | { | |
1210 | /* Calculate the ISO type. */ | |
1211 | if (! spec_long && ! spec_unsigned | |
1212 | && int_fits_type_p (value, integer_type_node)) | |
1213 | ansi_type = integer_type_node; | |
1214 | else if (! spec_long && (base != 10 || spec_unsigned) | |
1215 | && int_fits_type_p (value, unsigned_type_node)) | |
1216 | ansi_type = unsigned_type_node; | |
1217 | else if (! spec_unsigned && !spec_long_long | |
1218 | && int_fits_type_p (value, long_integer_type_node)) | |
1219 | ansi_type = long_integer_type_node; | |
1220 | else if (! spec_long_long | |
1221 | && int_fits_type_p (value, long_unsigned_type_node)) | |
1222 | ansi_type = long_unsigned_type_node; | |
1223 | else if (! spec_unsigned | |
1224 | && int_fits_type_p (value, long_long_integer_type_node)) | |
1225 | ansi_type = long_long_integer_type_node; | |
1226 | else if (int_fits_type_p (value, long_long_unsigned_type_node)) | |
1227 | ansi_type = long_long_unsigned_type_node; | |
1228 | else if (! spec_unsigned | |
1229 | && int_fits_type_p (value, widest_integer_literal_type_node)) | |
1230 | ansi_type = widest_integer_literal_type_node; | |
1231 | else | |
1232 | ansi_type = widest_unsigned_literal_type_node; | |
1233 | } | |
e8bbfc4e | 1234 | |
0e5921e8 | 1235 | type = flag_traditional ? trad_type : ansi_type; |
e8bbfc4e | 1236 | |
0e5921e8 ZW |
1237 | /* We assume that constants specified in a non-decimal |
1238 | base are bit patterns, and that the programmer really | |
1239 | meant what they wrote. */ | |
1240 | if (warn_traditional && !in_system_header | |
1241 | && base == 10 && trad_type != ansi_type) | |
1242 | { | |
1243 | if (TYPE_PRECISION (trad_type) != TYPE_PRECISION (ansi_type)) | |
1244 | warning ("width of integer constant changes with -traditional"); | |
1245 | else if (TREE_UNSIGNED (trad_type) != TREE_UNSIGNED (ansi_type)) | |
1246 | warning ("integer constant is unsigned in ISO C, signed with -traditional"); | |
1247 | else | |
1248 | warning ("width of integer constant may change on other systems with -traditional"); | |
1249 | } | |
e8bbfc4e | 1250 | |
b15ad712 JM |
1251 | if (pedantic && !flag_traditional && (flag_isoc99 || !spec_long_long) |
1252 | && !warn | |
1253 | && ((flag_isoc99 | |
1254 | ? TYPE_PRECISION (long_long_integer_type_node) | |
1255 | : TYPE_PRECISION (long_integer_type_node)) < TYPE_PRECISION (type))) | |
0e5921e8 ZW |
1256 | { |
1257 | warn = 1; | |
b15ad712 JM |
1258 | pedwarn ("integer constant larger than the maximum value of %s", |
1259 | (flag_isoc99 | |
1260 | ? (TREE_UNSIGNED (type) | |
1261 | ? "an unsigned long long int" | |
1262 | : "a long long int") | |
1263 | : "an unsigned long int")); | |
0e5921e8 | 1264 | } |
e8bbfc4e | 1265 | |
0e5921e8 ZW |
1266 | if (base == 10 && ! spec_unsigned && TREE_UNSIGNED (type)) |
1267 | warning ("decimal constant is so large that it is unsigned"); | |
e8bbfc4e | 1268 | |
0e5921e8 ZW |
1269 | if (spec_imag) |
1270 | { | |
1271 | if (TYPE_PRECISION (type) | |
1272 | <= TYPE_PRECISION (integer_type_node)) | |
1273 | value = build_complex (NULL_TREE, integer_zero_node, | |
1274 | convert (integer_type_node, value)); | |
1275 | else | |
1276 | ERROR ("complex integer constant is too wide for 'complex int'"); | |
1277 | } | |
1278 | else if (flag_traditional && !int_fits_type_p (value, type)) | |
1279 | /* The traditional constant 0x80000000 is signed | |
1280 | but doesn't fit in the range of int. | |
1281 | This will change it to -0x80000000, which does fit. */ | |
1282 | { | |
1283 | TREE_TYPE (value) = unsigned_type (type); | |
1284 | value = convert (type, value); | |
1285 | TREE_OVERFLOW (value) = TREE_CONSTANT_OVERFLOW (value) = 0; | |
1286 | } | |
1287 | else | |
1288 | TREE_TYPE (value) = type; | |
e8bbfc4e | 1289 | |
0e5921e8 | 1290 | /* If it's still an integer (not a complex), and it doesn't |
ec5c56db | 1291 | fit in the type we choose for it, then pedwarn. */ |
fbb18613 | 1292 | |
0e5921e8 ZW |
1293 | if (! warn |
1294 | && TREE_CODE (TREE_TYPE (value)) == INTEGER_TYPE | |
1295 | && ! int_fits_type_p (value, TREE_TYPE (value))) | |
1296 | pedwarn ("integer constant is larger than the maximum value for its type"); | |
1297 | } | |
fbb18613 | 1298 | |
0e5921e8 ZW |
1299 | if (p < str + len) |
1300 | error ("missing white space after number '%.*s'", (int) (p - str), str); | |
e8bbfc4e | 1301 | |
0e5921e8 | 1302 | return value; |
e8bbfc4e | 1303 | |
0e5921e8 ZW |
1304 | syntax_error: |
1305 | return integer_zero_node; | |
1306 | } | |
e8bbfc4e | 1307 | |
0e5921e8 ZW |
1308 | static tree |
1309 | lex_string (str, len, wide) | |
1310 | const char *str; | |
1311 | unsigned int len; | |
1312 | int wide; | |
1313 | { | |
1314 | tree value; | |
1315 | char *buf = alloca ((len + 1) * (wide ? WCHAR_BYTES : 1)); | |
1316 | char *q = buf; | |
1317 | const char *p = str, *limit = str + len; | |
1318 | unsigned int c; | |
1319 | unsigned width = wide ? WCHAR_TYPE_SIZE | |
1320 | : TYPE_PRECISION (char_type_node); | |
e9a25f70 | 1321 | |
0e5921e8 ZW |
1322 | #ifdef MULTIBYTE_CHARS |
1323 | /* Reset multibyte conversion state. */ | |
9714cf43 | 1324 | (void) local_mbtowc (NULL, NULL, 0); |
0e5921e8 | 1325 | #endif |
e9a25f70 | 1326 | |
0e5921e8 ZW |
1327 | while (p < limit) |
1328 | { | |
1329 | #ifdef MULTIBYTE_CHARS | |
1330 | wchar_t wc; | |
1331 | int char_len; | |
1332 | ||
1333 | char_len = local_mbtowc (&wc, p, limit - p); | |
1334 | if (char_len == -1) | |
1335 | { | |
1336 | warning ("Ignoring invalid multibyte character"); | |
1337 | char_len = 1; | |
1338 | c = *p++; | |
1339 | } | |
1340 | else | |
1341 | { | |
1342 | p += char_len; | |
1343 | c = wc; | |
1344 | } | |
1345 | #else | |
1346 | c = *p++; | |
1347 | #endif | |
1348 | ||
1349 | if (c == '\\' && !ignore_escape_flag) | |
1350 | { | |
62729350 NB |
1351 | unsigned int mask; |
1352 | ||
1353 | if (width < HOST_BITS_PER_INT) | |
1354 | mask = ((unsigned int) 1 << width) - 1; | |
1355 | else | |
1356 | mask = ~0; | |
1357 | c = cpp_parse_escape (parse_in, (const unsigned char **) &p, | |
1358 | (const unsigned char *) limit, | |
1359 | mask, flag_traditional); | |
0e5921e8 ZW |
1360 | } |
1361 | ||
1362 | /* Add this single character into the buffer either as a wchar_t | |
1363 | or as a single byte. */ | |
1364 | if (wide) | |
1365 | { | |
1366 | unsigned charwidth = TYPE_PRECISION (char_type_node); | |
5c80f6e6 | 1367 | unsigned bytemask = (1 << charwidth) - 1; |
0e5921e8 ZW |
1368 | int byte; |
1369 | ||
1370 | for (byte = 0; byte < WCHAR_BYTES; ++byte) | |
1371 | { | |
1372 | int n; | |
1373 | if (byte >= (int) sizeof (c)) | |
1374 | n = 0; | |
1375 | else | |
1376 | n = (c >> (byte * charwidth)) & bytemask; | |
1377 | if (BYTES_BIG_ENDIAN) | |
1378 | q[WCHAR_BYTES - byte - 1] = n; | |
1379 | else | |
1380 | q[byte] = n; | |
1381 | } | |
1382 | q += WCHAR_BYTES; | |
1383 | } | |
1384 | else | |
1385 | { | |
1386 | *q++ = c; | |
1387 | } | |
e8bbfc4e RK |
1388 | } |
1389 | ||
0e5921e8 ZW |
1390 | /* Terminate the string value, either with a single byte zero |
1391 | or with a wide zero. */ | |
e8bbfc4e | 1392 | |
0e5921e8 ZW |
1393 | if (wide) |
1394 | { | |
1395 | memset (q, 0, WCHAR_BYTES); | |
1396 | q += WCHAR_BYTES; | |
1397 | } | |
1398 | else | |
1399 | { | |
1400 | *q++ = '\0'; | |
1401 | } | |
1402 | ||
1403 | value = build_string (q - buf, buf); | |
1404 | ||
1405 | if (wide) | |
1406 | TREE_TYPE (value) = wchar_array_type_node; | |
1407 | else | |
1408 | TREE_TYPE (value) = char_array_type_node; | |
e8bbfc4e RK |
1409 | return value; |
1410 | } | |
1411 | ||
c8a96070 | 1412 | /* Converts a (possibly wide) character constant token into a tree. */ |
0e5921e8 | 1413 | static tree |
c8a96070 NB |
1414 | lex_charconst (token) |
1415 | const cpp_token *token; | |
e8bbfc4e | 1416 | { |
c8a96070 | 1417 | HOST_WIDE_INT result; |
0e5921e8 | 1418 | tree value; |
c8a96070 NB |
1419 | unsigned int chars_seen; |
1420 | ||
1421 | result = cpp_interpret_charconst (parse_in, token, warn_multichar, | |
1422 | flag_traditional, &chars_seen); | |
1423 | if (token->type == CPP_WCHAR) | |
0e5921e8 | 1424 | { |
c8a96070 NB |
1425 | value = build_int_2 (result, 0); |
1426 | TREE_TYPE (value) = wchar_type_node; | |
0e5921e8 | 1427 | } |
c8a96070 | 1428 | else |
0e5921e8 | 1429 | { |
c8a96070 NB |
1430 | if (result < 0) |
1431 | value = build_int_2 (result, -1); | |
0e5921e8 | 1432 | else |
c8a96070 NB |
1433 | value = build_int_2 (result, 0); |
1434 | ||
1435 | /* In C, a character constant has type 'int'. | |
1436 | In C++ 'char', but multi-char charconsts have type 'int'. */ | |
1437 | if (c_language == clk_cplusplus && chars_seen <= 1) | |
1438 | TREE_TYPE (value) = char_type_node; | |
0e5921e8 | 1439 | else |
c8a96070 | 1440 | TREE_TYPE (value) = integer_type_node; |
0e5921e8 | 1441 | } |
c8a96070 | 1442 | |
0e5921e8 | 1443 | return value; |
e8bbfc4e | 1444 | } |