]>
Commit | Line | Data |
---|---|---|
d45cf215 | 1 | /* Lexical analyzer for C and Objective C. |
517cbe13 JL |
2 | Copyright (C) 1987, 1988, 1989, 1992, 1994, 1995, 1996, 1997 |
3 | 1998, 1999, 2000 Free Software Foundation, Inc. | |
e8bbfc4e RK |
4 | |
5 | This file is part of GNU CC. | |
6 | ||
7 | GNU CC is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
9 | the Free Software Foundation; either version 2, or (at your option) | |
10 | any later version. | |
11 | ||
12 | GNU CC is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with GNU CC; see the file COPYING. If not, write to | |
940d9d63 RK |
19 | the Free Software Foundation, 59 Temple Place - Suite 330, |
20 | Boston, MA 02111-1307, USA. */ | |
e8bbfc4e | 21 | |
e9a25f70 | 22 | #include "config.h" |
670ee920 | 23 | #include "system.h" |
e8bbfc4e | 24 | |
e8bbfc4e | 25 | #include "rtl.h" |
8f17b5c5 | 26 | #include "expr.h" |
e8bbfc4e RK |
27 | #include "tree.h" |
28 | #include "input.h" | |
d6f4ec51 | 29 | #include "output.h" |
e8bbfc4e RK |
30 | #include "c-lex.h" |
31 | #include "c-tree.h" | |
32 | #include "flags.h" | |
0e5921e8 | 33 | #include "timevar.h" |
8b97c5f8 | 34 | #include "cpplib.h" |
3d6f7931 | 35 | #include "c-pragma.h" |
5f6da302 | 36 | #include "toplev.h" |
ab87f8c8 | 37 | #include "intl.h" |
7bdb32b9 | 38 | #include "tm_p.h" |
0e5921e8 | 39 | #include "splay-tree.h" |
ab87f8c8 | 40 | |
ab87f8c8 JL |
41 | /* MULTIBYTE_CHARS support only works for native compilers. |
42 | ??? Ideally what we want is to model widechar support after | |
43 | the current floating point support. */ | |
44 | #ifdef CROSS_COMPILE | |
45 | #undef MULTIBYTE_CHARS | |
46 | #endif | |
e8bbfc4e RK |
47 | |
48 | #ifdef MULTIBYTE_CHARS | |
56f48ce9 | 49 | #include "mbchar.h" |
e8bbfc4e | 50 | #include <locale.h> |
56f48ce9 | 51 | #endif /* MULTIBYTE_CHARS */ |
c5c76735 JL |
52 | #ifndef GET_ENVIRONMENT |
53 | #define GET_ENVIRONMENT(ENV_VALUE,ENV_NAME) ((ENV_VALUE) = getenv (ENV_NAME)) | |
54 | #endif | |
e8bbfc4e | 55 | |
373e2177 NB |
56 | /* The input filename as understood by CPP, where "" represents stdin. */ |
57 | static const char *cpp_filename; | |
3aac38d7 | 58 | |
0e5921e8 ZW |
59 | /* We may keep statistics about how long which files took to compile. */ |
60 | static int header_time, body_time; | |
61 | static splay_tree file_info_tree; | |
3ab6dd7c | 62 | |
e8bbfc4e RK |
63 | /* Cause the `yydebug' variable to be defined. */ |
64 | #define YYDEBUG 1 | |
65 | ||
e8bbfc4e RK |
66 | /* File used for outputting assembler code. */ |
67 | extern FILE *asm_out_file; | |
68 | ||
12a39b12 JM |
69 | #undef WCHAR_TYPE_SIZE |
70 | #define WCHAR_TYPE_SIZE TYPE_PRECISION (wchar_type_node) | |
e8bbfc4e RK |
71 | |
72 | /* Number of bytes in a wide character. */ | |
73 | #define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT) | |
74 | ||
0e5921e8 ZW |
75 | int indent_level; /* Number of { minus number of }. */ |
76 | int pending_lang_change; /* If we need to switch languages - C++ only */ | |
77 | int c_header_level; /* depth in C headers - C++ only */ | |
fbb18613 JM |
78 | |
79 | /* Nonzero tells yylex to ignore \ in string constants. */ | |
80 | static int ignore_escape_flag; | |
e9a25f70 | 81 | |
0e5921e8 ZW |
82 | static const char *readescape PARAMS ((const char *, const char *, |
83 | unsigned int *)); | |
84 | static const char *read_ucs PARAMS ((const char *, const char *, | |
85 | unsigned int *, int)); | |
86 | static void parse_float PARAMS ((PTR)); | |
87 | static tree lex_number PARAMS ((const char *, unsigned int)); | |
88 | static tree lex_string PARAMS ((const char *, unsigned int, int)); | |
c8a96070 | 89 | static tree lex_charconst PARAMS ((const cpp_token *)); |
0e5921e8 ZW |
90 | static void update_header_times PARAMS ((const char *)); |
91 | static int dump_one_header PARAMS ((splay_tree_node, void *)); | |
23356f93 | 92 | static void cb_ident PARAMS ((cpp_reader *, const cpp_string *)); |
eb1f4d9d | 93 | static void cb_file_change PARAMS ((cpp_reader *, const cpp_file_change *)); |
8b97c5f8 | 94 | static void cb_def_pragma PARAMS ((cpp_reader *)); |
65289a3a NB |
95 | static void cb_define PARAMS ((cpp_reader *, cpp_hashnode *)); |
96 | static void cb_undef PARAMS ((cpp_reader *, cpp_hashnode *)); | |
e31c7eec | 97 | \f |
3b304f5b | 98 | const char * |
0e5921e8 | 99 | init_c_lex (filename) |
3b304f5b | 100 | const char *filename; |
e3d1fd32 | 101 | { |
b61c5ed0 | 102 | struct cpp_callbacks *cb; |
0e5921e8 ZW |
103 | struct c_fileinfo *toplevel; |
104 | ||
105 | /* Set up filename timing. Must happen before cpp_start_read. */ | |
106 | file_info_tree = splay_tree_new ((splay_tree_compare_fn)strcmp, | |
107 | 0, | |
108 | (splay_tree_delete_value_fn)free); | |
a8a05998 | 109 | toplevel = get_fileinfo ("<top level>"); |
0e5921e8 ZW |
110 | if (flag_detailed_statistics) |
111 | { | |
112 | header_time = 0; | |
113 | body_time = get_run_time (); | |
114 | toplevel->time = body_time; | |
115 | } | |
116 | ||
117 | #ifdef MULTIBYTE_CHARS | |
118 | /* Change to the native locale for multibyte conversions. */ | |
119 | setlocale (LC_CTYPE, ""); | |
120 | GET_ENVIRONMENT (literal_codeset, "LANG"); | |
121 | #endif | |
122 | ||
b61c5ed0 NB |
123 | cb = cpp_get_callbacks (parse_in); |
124 | ||
125 | cb->ident = cb_ident; | |
126 | cb->file_change = cb_file_change; | |
127 | cb->def_pragma = cb_def_pragma; | |
0e5921e8 | 128 | |
65289a3a NB |
129 | /* Set the debug callbacks if we can use them. */ |
130 | if (debug_info_level == DINFO_LEVEL_VERBOSE | |
131 | && (write_symbols == DWARF_DEBUG || write_symbols == DWARF2_DEBUG)) | |
132 | { | |
b61c5ed0 NB |
133 | cb->define = cb_define; |
134 | cb->undef = cb_undef; | |
65289a3a NB |
135 | } |
136 | ||
373e2177 | 137 | |
b12da25e | 138 | if (filename == 0 || !strcmp (filename, "-")) |
373e2177 NB |
139 | filename = "stdin", cpp_filename = ""; |
140 | else | |
141 | cpp_filename = filename; | |
5c60e5c0 | 142 | |
851ef036 NB |
143 | /* Start it at 0. */ |
144 | lineno = 0; | |
b12da25e | 145 | |
5c60e5c0 | 146 | return filename; |
e3d1fd32 PB |
147 | } |
148 | ||
3aac38d7 RH |
149 | /* A thin wrapper around the real parser that initializes the |
150 | integrated preprocessor after debug output has been initialized. */ | |
151 | ||
152 | int | |
153 | yyparse() | |
154 | { | |
373e2177 | 155 | if (! cpp_start_read (parse_in, cpp_filename)) |
3aac38d7 RH |
156 | return 1; /* cpplib has emitted an error. */ |
157 | ||
158 | return yyparse_1(); | |
159 | } | |
160 | ||
0e5921e8 ZW |
161 | struct c_fileinfo * |
162 | get_fileinfo (name) | |
163 | const char *name; | |
e3d1fd32 | 164 | { |
0e5921e8 ZW |
165 | splay_tree_node n; |
166 | struct c_fileinfo *fi; | |
167 | ||
168 | n = splay_tree_lookup (file_info_tree, (splay_tree_key) name); | |
169 | if (n) | |
170 | return (struct c_fileinfo *) n->value; | |
171 | ||
172 | fi = (struct c_fileinfo *) xmalloc (sizeof (struct c_fileinfo)); | |
173 | fi->time = 0; | |
174 | fi->interface_only = 0; | |
175 | fi->interface_unknown = 1; | |
176 | splay_tree_insert (file_info_tree, (splay_tree_key) name, | |
177 | (splay_tree_value) fi); | |
178 | return fi; | |
e56e519d | 179 | } |
e3d1fd32 | 180 | |
0e5921e8 ZW |
181 | static void |
182 | update_header_times (name) | |
183 | const char *name; | |
e8bbfc4e | 184 | { |
0e5921e8 ZW |
185 | /* Changing files again. This means currently collected time |
186 | is charged against header time, and body time starts back at 0. */ | |
187 | if (flag_detailed_statistics) | |
e8bbfc4e | 188 | { |
0e5921e8 ZW |
189 | int this_time = get_run_time (); |
190 | struct c_fileinfo *file = get_fileinfo (name); | |
191 | header_time += this_time - body_time; | |
192 | file->time += this_time - body_time; | |
193 | body_time = this_time; | |
e8bbfc4e RK |
194 | } |
195 | } | |
196 | ||
0e5921e8 ZW |
197 | static int |
198 | dump_one_header (n, dummy) | |
199 | splay_tree_node n; | |
200 | void *dummy ATTRIBUTE_UNUSED; | |
e8bbfc4e | 201 | { |
0e5921e8 ZW |
202 | print_time ((const char *) n->key, |
203 | ((struct c_fileinfo *) n->value)->time); | |
204 | return 0; | |
e8bbfc4e | 205 | } |
e8bbfc4e RK |
206 | |
207 | void | |
0e5921e8 | 208 | dump_time_statistics () |
e8bbfc4e | 209 | { |
0e5921e8 ZW |
210 | struct c_fileinfo *file = get_fileinfo (input_filename); |
211 | int this_time = get_run_time (); | |
212 | file->time += this_time - body_time; | |
213 | ||
214 | fprintf (stderr, "\n******\n"); | |
215 | print_time ("header files (total)", header_time); | |
216 | print_time ("main file (total)", this_time - body_time); | |
217 | fprintf (stderr, "ratio = %g : 1\n", | |
218 | (double)header_time / (double)(this_time - body_time)); | |
219 | fprintf (stderr, "\n******\n"); | |
220 | ||
221 | splay_tree_foreach (file_info_tree, dump_one_header, 0); | |
e8bbfc4e | 222 | } |
a6124a42 | 223 | |
27e2564a NB |
224 | /* Not yet handled: #pragma, #define, #undef. |
225 | No need to deal with linemarkers under normal conditions. */ | |
226 | ||
0e5921e8 | 227 | static void |
27e2564a NB |
228 | cb_ident (pfile, str) |
229 | cpp_reader *pfile ATTRIBUTE_UNUSED; | |
f5720527 | 230 | const cpp_string *str ATTRIBUTE_UNUSED; |
0e5921e8 | 231 | { |
0e5921e8 | 232 | #ifdef ASM_OUTPUT_IDENT |
27e2564a | 233 | if (! flag_no_ident) |
0e5921e8 | 234 | { |
27e2564a NB |
235 | /* Convert escapes in the string. */ |
236 | tree value = lex_string ((const char *)str->text, str->len, 0); | |
237 | ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (value)); | |
0e5921e8 ZW |
238 | } |
239 | #endif | |
27e2564a NB |
240 | } |
241 | ||
242 | static void | |
eb1f4d9d | 243 | cb_file_change (pfile, fc) |
27e2564a NB |
244 | cpp_reader *pfile ATTRIBUTE_UNUSED; |
245 | const cpp_file_change *fc; | |
246 | { | |
27e2564a | 247 | if (fc->reason == FC_ENTER) |
fbb18613 | 248 | { |
5bea1ccf JM |
249 | /* Don't stack the main buffer on the input stack; |
250 | we already did in compile_file. */ | |
27e2564a | 251 | if (fc->from.filename) |
0e5921e8 | 252 | { |
851ef036 | 253 | lineno = fc->from.lineno; |
27e2564a NB |
254 | push_srcloc (fc->to.filename, 1); |
255 | input_file_stack->indent_level = indent_level; | |
256 | debug_start_source_file (fc->to.filename); | |
257 | #ifndef NO_IMPLICIT_EXTERN_C | |
258 | if (c_header_level) | |
259 | ++c_header_level; | |
260 | else if (fc->externc) | |
261 | { | |
262 | c_header_level = 1; | |
263 | ++pending_lang_change; | |
264 | } | |
0e5921e8 | 265 | #endif |
27e2564a | 266 | } |
ad2a084d NB |
267 | else |
268 | main_input_filename = fc->to.filename; | |
fbb18613 | 269 | } |
27e2564a | 270 | else if (fc->reason == FC_LEAVE) |
fbb18613 JM |
271 | { |
272 | /* Popping out of a file. */ | |
273 | if (input_file_stack->next) | |
0468bc75 | 274 | { |
0e5921e8 ZW |
275 | #ifndef NO_IMPLICIT_EXTERN_C |
276 | if (c_header_level && --c_header_level == 0) | |
277 | { | |
27e2564a | 278 | if (fc->externc) |
0e5921e8 ZW |
279 | warning ("badly nested C headers from preprocessor"); |
280 | --pending_lang_change; | |
281 | } | |
282 | #endif | |
283 | #if 0 | |
284 | if (indent_level != input_file_stack->indent_level) | |
fbb18613 JM |
285 | { |
286 | warning_with_file_and_line | |
5bea1ccf | 287 | (input_filename, lineno, |
0e5921e8 ZW |
288 | "This file contains more '%c's than '%c's.", |
289 | indent_level > input_file_stack->indent_level ? '{' : '}', | |
290 | indent_level > input_file_stack->indent_level ? '}' : '{'); | |
fbb18613 | 291 | } |
0e5921e8 ZW |
292 | #endif |
293 | pop_srcloc (); | |
fbb18613 | 294 | debug_end_source_file (input_file_stack->line); |
0468bc75 | 295 | } |
fbb18613 | 296 | else |
27e2564a | 297 | error ("leaving more files than we entered"); |
e8bbfc4e | 298 | } |
fbb18613 | 299 | |
27e2564a | 300 | update_header_times (fc->to.filename); |
665ad678 | 301 | in_system_header = fc->sysp != 0; |
27e2564a | 302 | input_filename = fc->to.filename; |
851ef036 | 303 | lineno = fc->to.lineno; /* Do we need this? */ |
0e5921e8 | 304 | |
0e5921e8 ZW |
305 | /* Hook for C++. */ |
306 | extract_interface_info (); | |
307 | } | |
8b97c5f8 ZW |
308 | |
309 | static void | |
310 | cb_def_pragma (pfile) | |
311 | cpp_reader *pfile; | |
312 | { | |
313 | /* Issue a warning message if we have been asked to do so. Ignore | |
314 | unknown pragmas in system headers unless an explicit | |
315 | -Wunknown-pragmas has been given. */ | |
316 | if (warn_unknown_pragmas > in_system_header) | |
317 | { | |
23356f93 ZW |
318 | const unsigned char *space, *name = 0; |
319 | cpp_token s; | |
320 | ||
321 | cpp_get_token (pfile, &s); | |
322 | space = cpp_token_as_text (pfile, &s); | |
323 | cpp_get_token (pfile, &s); | |
324 | if (s.type == CPP_NAME) | |
325 | name = cpp_token_as_text (pfile, &s); | |
8b97c5f8 | 326 | |
851ef036 | 327 | lineno = cpp_get_line (parse_in)->line; |
8b97c5f8 ZW |
328 | if (name) |
329 | warning ("ignoring #pragma %s %s", space, name); | |
330 | else | |
331 | warning ("ignoring #pragma %s", space); | |
332 | } | |
333 | } | |
0e5921e8 | 334 | |
65289a3a NB |
335 | /* #define callback for DWARF and DWARF2 debug info. */ |
336 | static void | |
337 | cb_define (pfile, node) | |
338 | cpp_reader *pfile; | |
339 | cpp_hashnode *node; | |
340 | { | |
341 | debug_define (lineno, (const char *) cpp_macro_definition (pfile, node)); | |
342 | } | |
343 | ||
344 | /* #undef callback for DWARF and DWARF2 debug info. */ | |
345 | static void | |
346 | cb_undef (pfile, node) | |
347 | cpp_reader *pfile ATTRIBUTE_UNUSED; | |
348 | cpp_hashnode *node; | |
349 | { | |
a28c5035 | 350 | debug_undef (lineno, (const char *) NODE_NAME (node)); |
65289a3a NB |
351 | } |
352 | ||
0e5921e8 ZW |
353 | /* Parse a '\uNNNN' or '\UNNNNNNNN' sequence. |
354 | ||
355 | [lex.charset]: The character designated by the universal-character-name | |
356 | \UNNNNNNNN is that character whose character short name in ISO/IEC 10646 | |
357 | is NNNNNNNN; the character designated by the universal-character-name | |
358 | \uNNNN is that character whose character short name in ISO/IEC 10646 is | |
359 | 0000NNNN. If the hexadecimal value for a universal character name is | |
360 | less than 0x20 or in the range 0x7F-0x9F (inclusive), or if the | |
361 | universal character name designates a character in the basic source | |
362 | character set, then the program is ill-formed. | |
363 | ||
364 | We assume that wchar_t is Unicode, so we don't need to do any | |
365 | mapping. Is this ever wrong? */ | |
366 | ||
367 | static const char * | |
368 | read_ucs (p, limit, cptr, length) | |
369 | const char *p; | |
370 | const char *limit; | |
371 | unsigned int *cptr; | |
372 | int length; | |
373 | { | |
374 | unsigned int code = 0; | |
375 | int c; | |
376 | ||
377 | for (; length; --length) | |
ca5b800a | 378 | { |
0e5921e8 | 379 | if (p >= limit) |
a3100298 | 380 | { |
0e5921e8 | 381 | error ("incomplete universal-character-name"); |
a3100298 | 382 | break; |
0e5921e8 | 383 | } |
fbb18613 | 384 | |
0e5921e8 ZW |
385 | c = *p++; |
386 | if (! ISXDIGIT (c)) | |
387 | { | |
388 | error ("non hex digit '%c' in universal-character-name", c); | |
389 | p--; | |
390 | break; | |
a3100298 | 391 | } |
f09db6e0 | 392 | |
0e5921e8 ZW |
393 | code <<= 4; |
394 | if (c >= 'a' && c <= 'f') | |
395 | code += c - 'a' + 10; | |
396 | if (c >= 'A' && c <= 'F') | |
397 | code += c - 'A' + 10; | |
398 | if (c >= '0' && c <= '9') | |
399 | code += c - '0'; | |
ca5b800a | 400 | } |
ca5b800a | 401 | |
0e5921e8 ZW |
402 | #ifdef TARGET_EBCDIC |
403 | sorry ("universal-character-name on EBCDIC target"); | |
404 | *cptr = 0x3f; /* EBCDIC invalid character */ | |
405 | return p; | |
406 | #endif | |
407 | ||
408 | if (code > 0x9f && !(code & 0x80000000)) | |
409 | /* True extended character, OK. */; | |
410 | else if (code >= 0x20 && code < 0x7f) | |
411 | { | |
412 | /* ASCII printable character. The C character set consists of all of | |
413 | these except $, @ and `. We use hex escapes so that this also | |
414 | works with EBCDIC hosts. */ | |
415 | if (code != 0x24 && code != 0x40 && code != 0x60) | |
416 | error ("universal-character-name used for '%c'", code); | |
417 | } | |
418 | else | |
419 | error ("invalid universal-character-name"); | |
420 | ||
421 | *cptr = code; | |
422 | return p; | |
423 | } | |
e8bbfc4e | 424 | |
0e5921e8 ZW |
425 | /* Read an escape sequence and write its character equivalent into *CPTR. |
426 | P is the input pointer, which is just after the backslash. LIMIT | |
427 | is how much text we have. | |
428 | Returns the updated input pointer. */ | |
e8bbfc4e | 429 | |
0e5921e8 ZW |
430 | static const char * |
431 | readescape (p, limit, cptr) | |
432 | const char *p; | |
433 | const char *limit; | |
434 | unsigned int *cptr; | |
e8bbfc4e | 435 | { |
0e5921e8 | 436 | unsigned int c, code, count; |
1c7b145e | 437 | unsigned firstdig = 0; |
8696da34 | 438 | int nonnull; |
e8bbfc4e | 439 | |
0e5921e8 ZW |
440 | if (p == limit) |
441 | { | |
442 | /* cpp has already issued an error for this. */ | |
443 | *cptr = 0; | |
444 | return p; | |
445 | } | |
446 | ||
447 | c = *p++; | |
448 | ||
e8bbfc4e RK |
449 | switch (c) |
450 | { | |
451 | case 'x': | |
cde6e684 | 452 | if (warn_traditional && !in_system_header) |
e8bbfc4e RK |
453 | warning ("the meaning of `\\x' varies with -traditional"); |
454 | ||
455 | if (flag_traditional) | |
0e5921e8 ZW |
456 | { |
457 | *cptr = 'x'; | |
458 | return p; | |
459 | } | |
e8bbfc4e RK |
460 | |
461 | code = 0; | |
462 | count = 0; | |
8696da34 | 463 | nonnull = 0; |
0e5921e8 | 464 | while (p < limit) |
e8bbfc4e | 465 | { |
0e5921e8 | 466 | c = *p++; |
fbb18613 | 467 | if (! ISXDIGIT (c)) |
e8bbfc4e | 468 | { |
0e5921e8 | 469 | p--; |
e8bbfc4e RK |
470 | break; |
471 | } | |
472 | code *= 16; | |
473 | if (c >= 'a' && c <= 'f') | |
474 | code += c - 'a' + 10; | |
475 | if (c >= 'A' && c <= 'F') | |
476 | code += c - 'A' + 10; | |
477 | if (c >= '0' && c <= '9') | |
478 | code += c - '0'; | |
8696da34 RS |
479 | if (code != 0 || count != 0) |
480 | { | |
481 | if (count == 0) | |
482 | firstdig = code; | |
483 | count++; | |
484 | } | |
485 | nonnull = 1; | |
e8bbfc4e | 486 | } |
8696da34 | 487 | if (! nonnull) |
4082292a AO |
488 | { |
489 | warning ("\\x used with no following hex digits"); | |
0e5921e8 ZW |
490 | *cptr = 'x'; |
491 | return p; | |
4082292a | 492 | } |
be63d912 RS |
493 | else if (count == 0) |
494 | /* Digits are all 0's. Ok. */ | |
495 | ; | |
e8bbfc4e RK |
496 | else if ((count - 1) * 4 >= TYPE_PRECISION (integer_type_node) |
497 | || (count > 1 | |
fbb18613 JM |
498 | && (((unsigned)1 |
499 | << (TYPE_PRECISION (integer_type_node) | |
500 | - (count - 1) * 4)) | |
e8bbfc4e RK |
501 | <= firstdig))) |
502 | pedwarn ("hex escape out of range"); | |
0e5921e8 ZW |
503 | *cptr = code; |
504 | return p; | |
e8bbfc4e RK |
505 | |
506 | case '0': case '1': case '2': case '3': case '4': | |
507 | case '5': case '6': case '7': | |
508 | code = 0; | |
0e5921e8 | 509 | for (count = 0; count < 3; count++) |
e8bbfc4e | 510 | { |
0e5921e8 ZW |
511 | if (c < '0' || c > '7') |
512 | { | |
513 | p--; | |
514 | break; | |
515 | } | |
e8bbfc4e | 516 | code = (code * 8) + (c - '0'); |
0e5921e8 ZW |
517 | if (p == limit) |
518 | break; | |
519 | c = *p++; | |
e8bbfc4e | 520 | } |
e8bbfc4e | 521 | |
0e5921e8 ZW |
522 | if (count == 3) |
523 | p--; | |
e8bbfc4e | 524 | |
0e5921e8 ZW |
525 | *cptr = code; |
526 | return p; | |
e8bbfc4e | 527 | |
0e5921e8 ZW |
528 | case '\\': case '\'': case '"': case '?': |
529 | *cptr = c; | |
530 | return p; | |
e8bbfc4e | 531 | |
0e5921e8 ZW |
532 | case 'n': *cptr = TARGET_NEWLINE; return p; |
533 | case 't': *cptr = TARGET_TAB; return p; | |
534 | case 'r': *cptr = TARGET_CR; return p; | |
535 | case 'f': *cptr = TARGET_FF; return p; | |
536 | case 'b': *cptr = TARGET_BS; return p; | |
537 | case 'v': *cptr = TARGET_VT; return p; | |
538 | case 'a': | |
539 | if (warn_traditional && !in_system_header) | |
540 | warning ("the meaning of '\\a' varies with -traditional"); | |
541 | *cptr = flag_traditional ? c : TARGET_BELL; | |
542 | return p; | |
e8bbfc4e | 543 | |
0e5921e8 ZW |
544 | /* Warnings and support checks handled by read_ucs(). */ |
545 | case 'u': case 'U': | |
546 | if (c_language != clk_cplusplus && !flag_isoc99) | |
547 | break; | |
e8bbfc4e | 548 | |
cde6e684 | 549 | if (warn_traditional && !in_system_header) |
0e5921e8 | 550 | warning ("the meaning of '\\%c' varies with -traditional", c); |
e8bbfc4e | 551 | |
0e5921e8 ZW |
552 | return read_ucs (p, limit, cptr, c == 'u' ? 4 : 8); |
553 | ||
554 | case 'e': case 'E': | |
dad112ca | 555 | if (pedantic) |
0e5921e8 ZW |
556 | pedwarn ("non-ISO-standard escape sequence, '\\%c'", c); |
557 | *cptr = TARGET_ESC; return p; | |
e8bbfc4e | 558 | |
0e5921e8 ZW |
559 | /* '\(', etc, are used at beginning of line to avoid confusing Emacs. |
560 | '\%' is used to prevent SCCS from getting confused. */ | |
561 | case '(': case '{': case '[': case '%': | |
e8bbfc4e | 562 | if (pedantic) |
0e5921e8 ZW |
563 | pedwarn ("unknown escape sequence '\\%c'", c); |
564 | *cptr = c; | |
565 | return p; | |
e8bbfc4e | 566 | } |
0e5921e8 | 567 | |
fbb18613 | 568 | if (ISGRAPH (c)) |
0e5921e8 | 569 | pedwarn ("unknown escape sequence '\\%c'", c); |
e8bbfc4e | 570 | else |
a52fb89b | 571 | pedwarn ("unknown escape sequence: '\\' followed by char 0x%x", c); |
0e5921e8 ZW |
572 | |
573 | *cptr = c; | |
574 | return p; | |
e8bbfc4e | 575 | } |
0e5921e8 ZW |
576 | |
577 | #if 0 /* not yet */ | |
578 | /* Returns nonzero if C is a universal-character-name. Give an error if it | |
579 | is not one which may appear in an identifier, as per [extendid]. | |
580 | ||
581 | Note that extended character support in identifiers has not yet been | |
582 | implemented. It is my personal opinion that this is not a desirable | |
583 | feature. Portable code cannot count on support for more than the basic | |
584 | identifier character set. */ | |
585 | ||
586 | static inline int | |
587 | is_extended_char (c) | |
588 | int c; | |
589 | { | |
590 | #ifdef TARGET_EBCDIC | |
591 | return 0; | |
592 | #else | |
593 | /* ASCII. */ | |
594 | if (c < 0x7f) | |
595 | return 0; | |
596 | ||
597 | /* None of the valid chars are outside the Basic Multilingual Plane (the | |
598 | low 16 bits). */ | |
599 | if (c > 0xffff) | |
600 | { | |
601 | error ("universal-character-name '\\U%08x' not valid in identifier", c); | |
602 | return 1; | |
603 | } | |
604 | ||
605 | /* Latin */ | |
606 | if ((c >= 0x00c0 && c <= 0x00d6) | |
607 | || (c >= 0x00d8 && c <= 0x00f6) | |
608 | || (c >= 0x00f8 && c <= 0x01f5) | |
609 | || (c >= 0x01fa && c <= 0x0217) | |
610 | || (c >= 0x0250 && c <= 0x02a8) | |
611 | || (c >= 0x1e00 && c <= 0x1e9a) | |
612 | || (c >= 0x1ea0 && c <= 0x1ef9)) | |
613 | return 1; | |
614 | ||
615 | /* Greek */ | |
616 | if ((c == 0x0384) | |
617 | || (c >= 0x0388 && c <= 0x038a) | |
618 | || (c == 0x038c) | |
619 | || (c >= 0x038e && c <= 0x03a1) | |
620 | || (c >= 0x03a3 && c <= 0x03ce) | |
621 | || (c >= 0x03d0 && c <= 0x03d6) | |
622 | || (c == 0x03da) | |
623 | || (c == 0x03dc) | |
624 | || (c == 0x03de) | |
625 | || (c == 0x03e0) | |
626 | || (c >= 0x03e2 && c <= 0x03f3) | |
627 | || (c >= 0x1f00 && c <= 0x1f15) | |
628 | || (c >= 0x1f18 && c <= 0x1f1d) | |
629 | || (c >= 0x1f20 && c <= 0x1f45) | |
630 | || (c >= 0x1f48 && c <= 0x1f4d) | |
631 | || (c >= 0x1f50 && c <= 0x1f57) | |
632 | || (c == 0x1f59) | |
633 | || (c == 0x1f5b) | |
634 | || (c == 0x1f5d) | |
635 | || (c >= 0x1f5f && c <= 0x1f7d) | |
636 | || (c >= 0x1f80 && c <= 0x1fb4) | |
637 | || (c >= 0x1fb6 && c <= 0x1fbc) | |
638 | || (c >= 0x1fc2 && c <= 0x1fc4) | |
639 | || (c >= 0x1fc6 && c <= 0x1fcc) | |
640 | || (c >= 0x1fd0 && c <= 0x1fd3) | |
641 | || (c >= 0x1fd6 && c <= 0x1fdb) | |
642 | || (c >= 0x1fe0 && c <= 0x1fec) | |
643 | || (c >= 0x1ff2 && c <= 0x1ff4) | |
644 | || (c >= 0x1ff6 && c <= 0x1ffc)) | |
645 | return 1; | |
646 | ||
647 | /* Cyrillic */ | |
648 | if ((c >= 0x0401 && c <= 0x040d) | |
649 | || (c >= 0x040f && c <= 0x044f) | |
650 | || (c >= 0x0451 && c <= 0x045c) | |
651 | || (c >= 0x045e && c <= 0x0481) | |
652 | || (c >= 0x0490 && c <= 0x04c4) | |
653 | || (c >= 0x04c7 && c <= 0x04c8) | |
654 | || (c >= 0x04cb && c <= 0x04cc) | |
655 | || (c >= 0x04d0 && c <= 0x04eb) | |
656 | || (c >= 0x04ee && c <= 0x04f5) | |
657 | || (c >= 0x04f8 && c <= 0x04f9)) | |
658 | return 1; | |
659 | ||
660 | /* Armenian */ | |
661 | if ((c >= 0x0531 && c <= 0x0556) | |
662 | || (c >= 0x0561 && c <= 0x0587)) | |
663 | return 1; | |
664 | ||
665 | /* Hebrew */ | |
666 | if ((c >= 0x05d0 && c <= 0x05ea) | |
667 | || (c >= 0x05f0 && c <= 0x05f4)) | |
668 | return 1; | |
669 | ||
670 | /* Arabic */ | |
671 | if ((c >= 0x0621 && c <= 0x063a) | |
672 | || (c >= 0x0640 && c <= 0x0652) | |
673 | || (c >= 0x0670 && c <= 0x06b7) | |
674 | || (c >= 0x06ba && c <= 0x06be) | |
675 | || (c >= 0x06c0 && c <= 0x06ce) | |
676 | || (c >= 0x06e5 && c <= 0x06e7)) | |
677 | return 1; | |
678 | ||
679 | /* Devanagari */ | |
680 | if ((c >= 0x0905 && c <= 0x0939) | |
681 | || (c >= 0x0958 && c <= 0x0962)) | |
682 | return 1; | |
683 | ||
684 | /* Bengali */ | |
685 | if ((c >= 0x0985 && c <= 0x098c) | |
686 | || (c >= 0x098f && c <= 0x0990) | |
687 | || (c >= 0x0993 && c <= 0x09a8) | |
688 | || (c >= 0x09aa && c <= 0x09b0) | |
689 | || (c == 0x09b2) | |
690 | || (c >= 0x09b6 && c <= 0x09b9) | |
691 | || (c >= 0x09dc && c <= 0x09dd) | |
692 | || (c >= 0x09df && c <= 0x09e1) | |
693 | || (c >= 0x09f0 && c <= 0x09f1)) | |
694 | return 1; | |
695 | ||
696 | /* Gurmukhi */ | |
697 | if ((c >= 0x0a05 && c <= 0x0a0a) | |
698 | || (c >= 0x0a0f && c <= 0x0a10) | |
699 | || (c >= 0x0a13 && c <= 0x0a28) | |
700 | || (c >= 0x0a2a && c <= 0x0a30) | |
701 | || (c >= 0x0a32 && c <= 0x0a33) | |
702 | || (c >= 0x0a35 && c <= 0x0a36) | |
703 | || (c >= 0x0a38 && c <= 0x0a39) | |
704 | || (c >= 0x0a59 && c <= 0x0a5c) | |
705 | || (c == 0x0a5e)) | |
706 | return 1; | |
707 | ||
708 | /* Gujarati */ | |
709 | if ((c >= 0x0a85 && c <= 0x0a8b) | |
710 | || (c == 0x0a8d) | |
711 | || (c >= 0x0a8f && c <= 0x0a91) | |
712 | || (c >= 0x0a93 && c <= 0x0aa8) | |
713 | || (c >= 0x0aaa && c <= 0x0ab0) | |
714 | || (c >= 0x0ab2 && c <= 0x0ab3) | |
715 | || (c >= 0x0ab5 && c <= 0x0ab9) | |
716 | || (c == 0x0ae0)) | |
717 | return 1; | |
718 | ||
719 | /* Oriya */ | |
720 | if ((c >= 0x0b05 && c <= 0x0b0c) | |
721 | || (c >= 0x0b0f && c <= 0x0b10) | |
722 | || (c >= 0x0b13 && c <= 0x0b28) | |
723 | || (c >= 0x0b2a && c <= 0x0b30) | |
724 | || (c >= 0x0b32 && c <= 0x0b33) | |
725 | || (c >= 0x0b36 && c <= 0x0b39) | |
726 | || (c >= 0x0b5c && c <= 0x0b5d) | |
727 | || (c >= 0x0b5f && c <= 0x0b61)) | |
728 | return 1; | |
729 | ||
730 | /* Tamil */ | |
731 | if ((c >= 0x0b85 && c <= 0x0b8a) | |
732 | || (c >= 0x0b8e && c <= 0x0b90) | |
733 | || (c >= 0x0b92 && c <= 0x0b95) | |
734 | || (c >= 0x0b99 && c <= 0x0b9a) | |
735 | || (c == 0x0b9c) | |
736 | || (c >= 0x0b9e && c <= 0x0b9f) | |
737 | || (c >= 0x0ba3 && c <= 0x0ba4) | |
738 | || (c >= 0x0ba8 && c <= 0x0baa) | |
739 | || (c >= 0x0bae && c <= 0x0bb5) | |
740 | || (c >= 0x0bb7 && c <= 0x0bb9)) | |
741 | return 1; | |
742 | ||
743 | /* Telugu */ | |
744 | if ((c >= 0x0c05 && c <= 0x0c0c) | |
745 | || (c >= 0x0c0e && c <= 0x0c10) | |
746 | || (c >= 0x0c12 && c <= 0x0c28) | |
747 | || (c >= 0x0c2a && c <= 0x0c33) | |
748 | || (c >= 0x0c35 && c <= 0x0c39) | |
749 | || (c >= 0x0c60 && c <= 0x0c61)) | |
750 | return 1; | |
751 | ||
752 | /* Kannada */ | |
753 | if ((c >= 0x0c85 && c <= 0x0c8c) | |
754 | || (c >= 0x0c8e && c <= 0x0c90) | |
755 | || (c >= 0x0c92 && c <= 0x0ca8) | |
756 | || (c >= 0x0caa && c <= 0x0cb3) | |
757 | || (c >= 0x0cb5 && c <= 0x0cb9) | |
758 | || (c >= 0x0ce0 && c <= 0x0ce1)) | |
759 | return 1; | |
760 | ||
761 | /* Malayalam */ | |
762 | if ((c >= 0x0d05 && c <= 0x0d0c) | |
763 | || (c >= 0x0d0e && c <= 0x0d10) | |
764 | || (c >= 0x0d12 && c <= 0x0d28) | |
765 | || (c >= 0x0d2a && c <= 0x0d39) | |
766 | || (c >= 0x0d60 && c <= 0x0d61)) | |
767 | return 1; | |
768 | ||
769 | /* Thai */ | |
770 | if ((c >= 0x0e01 && c <= 0x0e30) | |
771 | || (c >= 0x0e32 && c <= 0x0e33) | |
772 | || (c >= 0x0e40 && c <= 0x0e46) | |
773 | || (c >= 0x0e4f && c <= 0x0e5b)) | |
774 | return 1; | |
775 | ||
776 | /* Lao */ | |
777 | if ((c >= 0x0e81 && c <= 0x0e82) | |
778 | || (c == 0x0e84) | |
779 | || (c == 0x0e87) | |
780 | || (c == 0x0e88) | |
781 | || (c == 0x0e8a) | |
782 | || (c == 0x0e0d) | |
783 | || (c >= 0x0e94 && c <= 0x0e97) | |
784 | || (c >= 0x0e99 && c <= 0x0e9f) | |
785 | || (c >= 0x0ea1 && c <= 0x0ea3) | |
786 | || (c == 0x0ea5) | |
787 | || (c == 0x0ea7) | |
788 | || (c == 0x0eaa) | |
789 | || (c == 0x0eab) | |
790 | || (c >= 0x0ead && c <= 0x0eb0) | |
791 | || (c == 0x0eb2) | |
792 | || (c == 0x0eb3) | |
793 | || (c == 0x0ebd) | |
794 | || (c >= 0x0ec0 && c <= 0x0ec4) | |
795 | || (c == 0x0ec6)) | |
796 | return 1; | |
797 | ||
798 | /* Georgian */ | |
799 | if ((c >= 0x10a0 && c <= 0x10c5) | |
800 | || (c >= 0x10d0 && c <= 0x10f6)) | |
801 | return 1; | |
802 | ||
803 | /* Hiragana */ | |
804 | if ((c >= 0x3041 && c <= 0x3094) | |
805 | || (c >= 0x309b && c <= 0x309e)) | |
806 | return 1; | |
807 | ||
808 | /* Katakana */ | |
809 | if ((c >= 0x30a1 && c <= 0x30fe)) | |
810 | return 1; | |
811 | ||
812 | /* Bopmofo */ | |
813 | if ((c >= 0x3105 && c <= 0x312c)) | |
814 | return 1; | |
815 | ||
816 | /* Hangul */ | |
817 | if ((c >= 0x1100 && c <= 0x1159) | |
818 | || (c >= 0x1161 && c <= 0x11a2) | |
819 | || (c >= 0x11a8 && c <= 0x11f9)) | |
820 | return 1; | |
821 | ||
822 | /* CJK Unified Ideographs */ | |
823 | if ((c >= 0xf900 && c <= 0xfa2d) | |
824 | || (c >= 0xfb1f && c <= 0xfb36) | |
825 | || (c >= 0xfb38 && c <= 0xfb3c) | |
826 | || (c == 0xfb3e) | |
827 | || (c >= 0xfb40 && c <= 0xfb41) | |
828 | || (c >= 0xfb42 && c <= 0xfb44) | |
829 | || (c >= 0xfb46 && c <= 0xfbb1) | |
830 | || (c >= 0xfbd3 && c <= 0xfd3f) | |
831 | || (c >= 0xfd50 && c <= 0xfd8f) | |
832 | || (c >= 0xfd92 && c <= 0xfdc7) | |
833 | || (c >= 0xfdf0 && c <= 0xfdfb) | |
834 | || (c >= 0xfe70 && c <= 0xfe72) | |
835 | || (c == 0xfe74) | |
836 | || (c >= 0xfe76 && c <= 0xfefc) | |
837 | || (c >= 0xff21 && c <= 0xff3a) | |
838 | || (c >= 0xff41 && c <= 0xff5a) | |
839 | || (c >= 0xff66 && c <= 0xffbe) | |
840 | || (c >= 0xffc2 && c <= 0xffc7) | |
841 | || (c >= 0xffca && c <= 0xffcf) | |
842 | || (c >= 0xffd2 && c <= 0xffd7) | |
843 | || (c >= 0xffda && c <= 0xffdc) | |
844 | || (c >= 0x4e00 && c <= 0x9fa5)) | |
845 | return 1; | |
846 | ||
847 | error ("universal-character-name '\\u%04x' not valid in identifier", c); | |
848 | return 1; | |
849 | #endif | |
850 | } | |
851 | ||
852 | /* Add the UTF-8 representation of C to the token_buffer. */ | |
853 | ||
854 | static void | |
855 | utf8_extend_token (c) | |
856 | int c; | |
e8bbfc4e | 857 | { |
0e5921e8 ZW |
858 | int shift, mask; |
859 | ||
860 | if (c <= 0x0000007f) | |
861 | { | |
862 | extend_token (c); | |
863 | return; | |
864 | } | |
865 | else if (c <= 0x000007ff) | |
866 | shift = 6, mask = 0xc0; | |
867 | else if (c <= 0x0000ffff) | |
868 | shift = 12, mask = 0xe0; | |
869 | else if (c <= 0x001fffff) | |
870 | shift = 18, mask = 0xf0; | |
871 | else if (c <= 0x03ffffff) | |
872 | shift = 24, mask = 0xf8; | |
e8bbfc4e | 873 | else |
0e5921e8 ZW |
874 | shift = 30, mask = 0xfc; |
875 | ||
876 | extend_token (mask | (c >> shift)); | |
877 | do | |
878 | { | |
879 | shift -= 6; | |
880 | extend_token ((unsigned char) (0x80 | (c >> shift))); | |
881 | } | |
882 | while (shift); | |
e8bbfc4e | 883 | } |
0e5921e8 | 884 | #endif |
e8bbfc4e RK |
885 | |
886 | #if 0 | |
e8bbfc4e RK |
887 | struct try_type |
888 | { | |
889 | tree *node_var; | |
890 | char unsigned_flag; | |
891 | char long_flag; | |
892 | char long_long_flag; | |
893 | }; | |
894 | ||
75cb8865 | 895 | struct try_type type_sequence[] = |
e8bbfc4e RK |
896 | { |
897 | { &integer_type_node, 0, 0, 0}, | |
898 | { &unsigned_type_node, 1, 0, 0}, | |
899 | { &long_integer_type_node, 0, 1, 0}, | |
900 | { &long_unsigned_type_node, 1, 1, 0}, | |
901 | { &long_long_integer_type_node, 0, 1, 1}, | |
902 | { &long_long_unsigned_type_node, 1, 1, 1} | |
903 | }; | |
904 | #endif /* 0 */ | |
905 | \f | |
ed513abf KG |
906 | struct pf_args |
907 | { | |
908 | /* Input */ | |
0e5921e8 ZW |
909 | const char *str; |
910 | int fflag; | |
911 | int lflag; | |
ed513abf | 912 | int base; |
fbb18613 | 913 | /* Output */ |
ed513abf | 914 | int conversion_errno; |
ed513abf | 915 | REAL_VALUE_TYPE value; |
0e5921e8 | 916 | tree type; |
ed513abf KG |
917 | }; |
918 | ||
919 | static void | |
920 | parse_float (data) | |
921 | PTR data; | |
922 | { | |
923 | struct pf_args * args = (struct pf_args *) data; | |
0e5921e8 ZW |
924 | const char *typename; |
925 | ||
fbb18613 JM |
926 | args->conversion_errno = 0; |
927 | args->type = double_type_node; | |
0e5921e8 | 928 | typename = "double"; |
ed513abf KG |
929 | |
930 | /* The second argument, machine_mode, of REAL_VALUE_ATOF | |
931 | tells the desired precision of the binary result | |
932 | of decimal-to-binary conversion. */ | |
933 | ||
0e5921e8 | 934 | if (args->fflag) |
ed513abf | 935 | { |
0e5921e8 ZW |
936 | if (args->lflag) |
937 | error ("both 'f' and 'l' suffixes on floating constant"); | |
ed513abf KG |
938 | |
939 | args->type = float_type_node; | |
0e5921e8 | 940 | typename = "float"; |
ed513abf | 941 | } |
0e5921e8 | 942 | else if (args->lflag) |
ed513abf KG |
943 | { |
944 | args->type = long_double_type_node; | |
0e5921e8 | 945 | typename = "long double"; |
ed513abf | 946 | } |
0e5921e8 | 947 | else if (flag_single_precision_constant) |
ed513abf | 948 | { |
0e5921e8 ZW |
949 | args->type = float_type_node; |
950 | typename = "float"; | |
ed513abf | 951 | } |
0e5921e8 ZW |
952 | |
953 | errno = 0; | |
954 | if (args->base == 16) | |
955 | args->value = REAL_VALUE_HTOF (args->str, TYPE_MODE (args->type)); | |
956 | else | |
957 | args->value = REAL_VALUE_ATOF (args->str, TYPE_MODE (args->type)); | |
958 | ||
959 | args->conversion_errno = errno; | |
960 | /* A diagnostic is required here by some ISO C testsuites. | |
961 | This is not pedwarn, because some people don't want | |
962 | an error for this. */ | |
963 | if (REAL_VALUE_ISINF (args->value) && pedantic) | |
964 | warning ("floating point number exceeds range of '%s'", typename); | |
ed513abf KG |
965 | } |
966 | ||
0e5921e8 ZW |
967 | int |
968 | c_lex (value) | |
969 | tree *value; | |
fbb18613 | 970 | { |
23356f93 | 971 | cpp_token tok; |
0e5921e8 ZW |
972 | enum cpp_ttype type; |
973 | ||
974 | retry: | |
975 | timevar_push (TV_CPP); | |
cf44ea52 | 976 | cpp_get_token (parse_in, &tok); |
0e5921e8 ZW |
977 | timevar_pop (TV_CPP); |
978 | ||
979 | /* The C++ front end does horrible things with the current line | |
980 | number. To ensure an accurate line number, we must reset it | |
23356f93 | 981 | every time we return a token. */ |
851ef036 | 982 | lineno = cpp_get_line (parse_in)->line; |
0e5921e8 ZW |
983 | |
984 | *value = NULL_TREE; | |
23356f93 | 985 | type = tok.type; |
0e5921e8 ZW |
986 | switch (type) |
987 | { | |
988 | case CPP_OPEN_BRACE: indent_level++; break; | |
989 | case CPP_CLOSE_BRACE: indent_level--; break; | |
990 | ||
6c53ebff | 991 | /* Issue this error here, where we can get at tok.val.c. */ |
0e5921e8 | 992 | case CPP_OTHER: |
6c53ebff NB |
993 | if (ISGRAPH (tok.val.c)) |
994 | error ("stray '%c' in program", tok.val.c); | |
0e5921e8 | 995 | else |
5677efb0 | 996 | error ("stray '\\%o' in program", tok.val.c); |
0e5921e8 ZW |
997 | goto retry; |
998 | ||
0e5921e8 | 999 | case CPP_NAME: |
083e9f92 | 1000 | *value = HT_IDENT_TO_GCC_IDENT (HT_NODE (tok.val.node)); |
0e5921e8 | 1001 | break; |
fbb18613 | 1002 | |
0e5921e8 ZW |
1003 | case CPP_INT: |
1004 | case CPP_FLOAT: | |
1005 | case CPP_NUMBER: | |
23356f93 | 1006 | *value = lex_number ((const char *)tok.val.str.text, tok.val.str.len); |
0e5921e8 | 1007 | break; |
93868d11 | 1008 | |
0e5921e8 ZW |
1009 | case CPP_CHAR: |
1010 | case CPP_WCHAR: | |
c8a96070 | 1011 | *value = lex_charconst (&tok); |
0e5921e8 | 1012 | break; |
fbb18613 | 1013 | |
0e5921e8 ZW |
1014 | case CPP_STRING: |
1015 | case CPP_WSTRING: | |
23356f93 ZW |
1016 | *value = lex_string ((const char *)tok.val.str.text, |
1017 | tok.val.str.len, tok.type == CPP_WSTRING); | |
0e5921e8 | 1018 | break; |
fbb18613 | 1019 | |
0e5921e8 ZW |
1020 | /* These tokens should not be visible outside cpplib. */ |
1021 | case CPP_HEADER_NAME: | |
1022 | case CPP_COMMENT: | |
1023 | case CPP_MACRO_ARG: | |
0e5921e8 ZW |
1024 | abort (); |
1025 | ||
1026 | default: break; | |
1027 | } | |
1028 | ||
1029 | return type; | |
0e5921e8 | 1030 | } |
8d9bfdc5 | 1031 | |
0e5921e8 | 1032 | #define ERROR(msgid) do { error(msgid); goto syntax_error; } while(0) |
75cb8865 | 1033 | |
0e5921e8 ZW |
1034 | static tree |
1035 | lex_number (str, len) | |
1036 | const char *str; | |
1037 | unsigned int len; | |
1038 | { | |
1039 | int base = 10; | |
1040 | int count = 0; | |
1041 | int largest_digit = 0; | |
1042 | int numdigits = 0; | |
1043 | int overflow = 0; | |
1044 | int c; | |
1045 | tree value; | |
1046 | const char *p; | |
1047 | enum anon1 { NOT_FLOAT = 0, AFTER_POINT, AFTER_EXPON } floatflag = NOT_FLOAT; | |
1048 | ||
1049 | /* We actually store only HOST_BITS_PER_CHAR bits in each part. | |
1050 | The code below which fills the parts array assumes that a host | |
1051 | int is at least twice as wide as a host char, and that | |
1052 | HOST_BITS_PER_WIDE_INT is an even multiple of HOST_BITS_PER_CHAR. | |
1053 | Two HOST_WIDE_INTs is the largest int literal we can store. | |
1054 | In order to detect overflow below, the number of parts (TOTAL_PARTS) | |
1055 | must be exactly the number of parts needed to hold the bits | |
1056 | of two HOST_WIDE_INTs. */ | |
1057 | #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2) | |
1058 | unsigned int parts[TOTAL_PARTS]; | |
1059 | ||
1060 | /* Optimize for most frequent case. */ | |
1061 | if (len == 1) | |
1062 | { | |
1063 | if (*str == '0') | |
1064 | return integer_zero_node; | |
1065 | else if (*str == '1') | |
1066 | return integer_one_node; | |
1067 | else | |
1068 | return build_int_2 (*str - '0', 0); | |
1069 | } | |
e8bbfc4e | 1070 | |
0e5921e8 ZW |
1071 | for (count = 0; count < TOTAL_PARTS; count++) |
1072 | parts[count] = 0; | |
e8bbfc4e | 1073 | |
0e5921e8 ZW |
1074 | /* len is known to be >1 at this point. */ |
1075 | p = str; | |
e8bbfc4e | 1076 | |
0e5921e8 ZW |
1077 | if (len > 2 && str[0] == '0' && (str[1] == 'x' || str[1] == 'X')) |
1078 | { | |
1079 | base = 16; | |
1080 | p = str + 2; | |
1081 | } | |
1082 | /* The ISDIGIT check is so we are not confused by a suffix on 0. */ | |
1083 | else if (str[0] == '0' && ISDIGIT (str[1])) | |
1084 | { | |
1085 | base = 8; | |
1086 | p = str + 1; | |
1087 | } | |
e8bbfc4e | 1088 | |
0e5921e8 ZW |
1089 | do |
1090 | { | |
1091 | c = *p++; | |
88d92ca5 | 1092 | |
0e5921e8 ZW |
1093 | if (c == '.') |
1094 | { | |
1095 | if (base == 16 && pedantic && !flag_isoc99) | |
1096 | pedwarn ("floating constant may not be in radix 16"); | |
1097 | else if (floatflag == AFTER_POINT) | |
1098 | ERROR ("too many decimal points in floating constant"); | |
1099 | else if (floatflag == AFTER_EXPON) | |
1100 | ERROR ("decimal point in exponent - impossible!"); | |
1101 | else | |
1102 | floatflag = AFTER_POINT; | |
c832a30e | 1103 | |
0e5921e8 ZW |
1104 | if (base == 8) |
1105 | base = 10; | |
1106 | } | |
1107 | else if (c == '_') | |
1108 | /* Possible future extension: silently ignore _ in numbers, | |
1109 | permitting cosmetic grouping - e.g. 0x8000_0000 == 0x80000000 | |
1110 | but somewhat easier to read. Ada has this? */ | |
1111 | ERROR ("underscore in number"); | |
1112 | else | |
1113 | { | |
1114 | int n; | |
1115 | /* It is not a decimal point. | |
1116 | It should be a digit (perhaps a hex digit). */ | |
0e0fda0d | 1117 | |
0e5921e8 ZW |
1118 | if (ISDIGIT (c)) |
1119 | { | |
1120 | n = c - '0'; | |
1121 | } | |
1122 | else if (base <= 10 && (c == 'e' || c == 'E')) | |
1123 | { | |
1124 | base = 10; | |
1125 | floatflag = AFTER_EXPON; | |
1126 | break; | |
1127 | } | |
1128 | else if (base == 16 && (c == 'p' || c == 'P')) | |
1129 | { | |
1130 | floatflag = AFTER_EXPON; | |
1131 | break; /* start of exponent */ | |
1132 | } | |
1133 | else if (base == 16 && c >= 'a' && c <= 'f') | |
1134 | { | |
1135 | n = c - 'a' + 10; | |
1136 | } | |
1137 | else if (base == 16 && c >= 'A' && c <= 'F') | |
1138 | { | |
1139 | n = c - 'A' + 10; | |
1140 | } | |
1141 | else | |
1142 | { | |
1143 | p--; | |
1144 | break; /* start of suffix */ | |
1145 | } | |
0e0fda0d | 1146 | |
0e5921e8 ZW |
1147 | if (n >= largest_digit) |
1148 | largest_digit = n; | |
1149 | numdigits++; | |
0e0fda0d | 1150 | |
0e5921e8 ZW |
1151 | for (count = 0; count < TOTAL_PARTS; count++) |
1152 | { | |
1153 | parts[count] *= base; | |
1154 | if (count) | |
1155 | { | |
1156 | parts[count] | |
1157 | += (parts[count-1] >> HOST_BITS_PER_CHAR); | |
1158 | parts[count-1] | |
1159 | &= (1 << HOST_BITS_PER_CHAR) - 1; | |
1160 | } | |
1161 | else | |
1162 | parts[0] += n; | |
1163 | } | |
e8bbfc4e | 1164 | |
0e5921e8 ZW |
1165 | /* If the highest-order part overflows (gets larger than |
1166 | a host char will hold) then the whole number has | |
1167 | overflowed. Record this and truncate the highest-order | |
1168 | part. */ | |
1169 | if (parts[TOTAL_PARTS - 1] >> HOST_BITS_PER_CHAR) | |
1170 | { | |
1171 | overflow = 1; | |
1172 | parts[TOTAL_PARTS - 1] &= (1 << HOST_BITS_PER_CHAR) - 1; | |
1173 | } | |
1174 | } | |
1175 | } | |
1176 | while (p < str + len); | |
91b0989e | 1177 | |
0e5921e8 ZW |
1178 | /* This can happen on input like `int i = 0x;' */ |
1179 | if (numdigits == 0) | |
1180 | ERROR ("numeric constant with no digits"); | |
91b0989e | 1181 | |
0e5921e8 ZW |
1182 | if (largest_digit >= base) |
1183 | ERROR ("numeric constant contains digits beyond the radix"); | |
e8bbfc4e | 1184 | |
0e5921e8 ZW |
1185 | if (floatflag != NOT_FLOAT) |
1186 | { | |
1187 | tree type; | |
1188 | int imag, fflag, lflag, conversion_errno; | |
1189 | REAL_VALUE_TYPE real; | |
1190 | struct pf_args args; | |
1191 | char *copy; | |
1192 | ||
1193 | if (base == 16 && floatflag != AFTER_EXPON) | |
1194 | ERROR ("hexadecimal floating constant has no exponent"); | |
1195 | ||
1196 | /* Read explicit exponent if any, and put it in tokenbuf. */ | |
1197 | if ((base == 10 && ((c == 'e') || (c == 'E'))) | |
1198 | || (base == 16 && (c == 'p' || c == 'P'))) | |
1199 | { | |
1200 | if (p < str + len) | |
1201 | c = *p++; | |
1202 | if (p < str + len && (c == '+' || c == '-')) | |
1203 | c = *p++; | |
1204 | /* Exponent is decimal, even if string is a hex float. */ | |
1205 | if (! ISDIGIT (c)) | |
1206 | ERROR ("floating constant exponent has no digits"); | |
1207 | while (p < str + len && ISDIGIT (c)) | |
1208 | c = *p++; | |
1209 | if (! ISDIGIT (c)) | |
1210 | p--; | |
1211 | } | |
56f48ce9 | 1212 | |
0e5921e8 ZW |
1213 | /* Copy the float constant now; we don't want any suffixes in the |
1214 | string passed to parse_float. */ | |
1215 | copy = alloca (p - str + 1); | |
1216 | memcpy (copy, str, p - str); | |
1217 | copy[p - str] = '\0'; | |
e8bbfc4e | 1218 | |
0e5921e8 ZW |
1219 | /* Now parse suffixes. */ |
1220 | fflag = lflag = imag = 0; | |
1221 | while (p < str + len) | |
1222 | switch (*p++) | |
e8bbfc4e | 1223 | { |
0e5921e8 ZW |
1224 | case 'f': case 'F': |
1225 | if (fflag) | |
1226 | ERROR ("more than one 'f' suffix on floating constant"); | |
7065e130 | 1227 | else if (warn_traditional && !in_system_header |
c691145a | 1228 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 | 1229 | warning ("traditional C rejects the 'f' suffix"); |
e8bbfc4e | 1230 | |
0e5921e8 ZW |
1231 | fflag = 1; |
1232 | break; | |
e8bbfc4e | 1233 | |
0e5921e8 ZW |
1234 | case 'l': case 'L': |
1235 | if (lflag) | |
1236 | ERROR ("more than one 'l' suffix on floating constant"); | |
7065e130 | 1237 | else if (warn_traditional && !in_system_header |
c691145a | 1238 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 | 1239 | warning ("traditional C rejects the 'l' suffix"); |
56f48ce9 | 1240 | |
0e5921e8 ZW |
1241 | lflag = 1; |
1242 | break; | |
e8bbfc4e | 1243 | |
0e5921e8 ZW |
1244 | case 'i': case 'I': |
1245 | case 'j': case 'J': | |
1246 | if (imag) | |
1247 | ERROR ("more than one 'i' or 'j' suffix on floating constant"); | |
1248 | else if (pedantic) | |
1249 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
1250 | imag = 1; | |
1251 | break; | |
e8bbfc4e | 1252 | |
0e5921e8 ZW |
1253 | default: |
1254 | ERROR ("invalid suffix on floating constant"); | |
e8bbfc4e RK |
1255 | } |
1256 | ||
0e5921e8 ZW |
1257 | /* Setup input for parse_float() */ |
1258 | args.str = copy; | |
1259 | args.fflag = fflag; | |
1260 | args.lflag = lflag; | |
1261 | args.base = base; | |
e8bbfc4e | 1262 | |
0e5921e8 ZW |
1263 | /* Convert string to a double, checking for overflow. */ |
1264 | if (do_float_handler (parse_float, (PTR) &args)) | |
1265 | { | |
1266 | /* Receive output from parse_float() */ | |
1267 | real = args.value; | |
1268 | } | |
1269 | else | |
1270 | /* We got an exception from parse_float() */ | |
1271 | ERROR ("floating constant out of range"); | |
e8bbfc4e | 1272 | |
0e5921e8 ZW |
1273 | /* Receive output from parse_float() */ |
1274 | conversion_errno = args.conversion_errno; | |
1275 | type = args.type; | |
1276 | ||
1277 | #ifdef ERANGE | |
1278 | /* ERANGE is also reported for underflow, | |
1279 | so test the value to distinguish overflow from that. */ | |
1280 | if (conversion_errno == ERANGE && !flag_traditional && pedantic | |
1281 | && (REAL_VALUES_LESS (dconst1, real) | |
1282 | || REAL_VALUES_LESS (real, dconstm1))) | |
1283 | warning ("floating point number exceeds range of 'double'"); | |
56f48ce9 | 1284 | #endif |
e8bbfc4e | 1285 | |
0e5921e8 ZW |
1286 | /* Create a node with determined type and value. */ |
1287 | if (imag) | |
1288 | value = build_complex (NULL_TREE, convert (type, integer_zero_node), | |
1289 | build_real (type, real)); | |
1290 | else | |
1291 | value = build_real (type, real); | |
1292 | } | |
1293 | else | |
1294 | { | |
1295 | tree trad_type, ansi_type, type; | |
1296 | HOST_WIDE_INT high, low; | |
1297 | int spec_unsigned = 0; | |
1298 | int spec_long = 0; | |
1299 | int spec_long_long = 0; | |
1300 | int spec_imag = 0; | |
1301 | int suffix_lu = 0; | |
1302 | int warn = 0, i; | |
1303 | ||
1304 | trad_type = ansi_type = type = NULL_TREE; | |
1305 | while (p < str + len) | |
1306 | { | |
1307 | c = *p++; | |
1308 | switch (c) | |
1309 | { | |
1310 | case 'u': case 'U': | |
1311 | if (spec_unsigned) | |
1312 | error ("two 'u' suffixes on integer constant"); | |
7065e130 | 1313 | else if (warn_traditional && !in_system_header |
c691145a | 1314 | && ! cpp_sys_macro_p (parse_in)) |
0e5921e8 ZW |
1315 | warning ("traditional C rejects the 'u' suffix"); |
1316 | ||
1317 | spec_unsigned = 1; | |
1318 | if (spec_long) | |
1319 | suffix_lu = 1; | |
1320 | break; | |
e8bbfc4e | 1321 | |
0e5921e8 ZW |
1322 | case 'l': case 'L': |
1323 | if (spec_long) | |
1324 | { | |
1325 | if (spec_long_long) | |
1326 | error ("three 'l' suffixes on integer constant"); | |
1327 | else if (suffix_lu) | |
1328 | error ("'lul' is not a valid integer suffix"); | |
1329 | else if (c != spec_long) | |
1330 | error ("'Ll' and 'lL' are not valid integer suffixes"); | |
1331 | else if (pedantic && ! flag_isoc99 | |
1332 | && ! in_system_header && warn_long_long) | |
1333 | pedwarn ("ISO C89 forbids long long integer constants"); | |
1334 | spec_long_long = 1; | |
1335 | } | |
1336 | spec_long = c; | |
1337 | break; | |
56f48ce9 | 1338 | |
0e5921e8 ZW |
1339 | case 'i': case 'I': case 'j': case 'J': |
1340 | if (spec_imag) | |
1341 | error ("more than one 'i' or 'j' suffix on integer constant"); | |
1342 | else if (pedantic) | |
1343 | pedwarn ("ISO C forbids imaginary numeric constants"); | |
1344 | spec_imag = 1; | |
1345 | break; | |
56f48ce9 | 1346 | |
0e5921e8 ZW |
1347 | default: |
1348 | ERROR ("invalid suffix on integer constant"); | |
1349 | } | |
1350 | } | |
56f48ce9 | 1351 | |
0e5921e8 ZW |
1352 | /* If the literal overflowed, pedwarn about it now. */ |
1353 | if (overflow) | |
1354 | { | |
1355 | warn = 1; | |
1356 | pedwarn ("integer constant is too large for this configuration of the compiler - truncated to %d bits", HOST_BITS_PER_WIDE_INT * 2); | |
1357 | } | |
e8bbfc4e | 1358 | |
0e5921e8 ZW |
1359 | /* This is simplified by the fact that our constant |
1360 | is always positive. */ | |
56f48ce9 | 1361 | |
0e5921e8 | 1362 | high = low = 0; |
e8bbfc4e | 1363 | |
0e5921e8 ZW |
1364 | for (i = 0; i < HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR; i++) |
1365 | { | |
1366 | high |= ((HOST_WIDE_INT) parts[i + (HOST_BITS_PER_WIDE_INT | |
1367 | / HOST_BITS_PER_CHAR)] | |
1368 | << (i * HOST_BITS_PER_CHAR)); | |
1369 | low |= (HOST_WIDE_INT) parts[i] << (i * HOST_BITS_PER_CHAR); | |
1370 | } | |
0468bc75 | 1371 | |
0e5921e8 ZW |
1372 | value = build_int_2 (low, high); |
1373 | TREE_TYPE (value) = long_long_unsigned_type_node; | |
e8bbfc4e | 1374 | |
0e5921e8 ZW |
1375 | /* If warn_traditional, calculate both the ISO type and the |
1376 | traditional type, then see if they disagree. | |
1377 | Otherwise, calculate only the type for the dialect in use. */ | |
1378 | if (warn_traditional || flag_traditional) | |
1379 | { | |
1380 | /* Calculate the traditional type. */ | |
1381 | /* Traditionally, any constant is signed; but if unsigned is | |
1382 | specified explicitly, obey that. Use the smallest size | |
1383 | with the right number of bits, except for one special | |
1384 | case with decimal constants. */ | |
1385 | if (! spec_long && base != 10 | |
1386 | && int_fits_type_p (value, unsigned_type_node)) | |
1387 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
1388 | /* A decimal constant must be long if it does not fit in | |
1389 | type int. I think this is independent of whether the | |
1390 | constant is signed. */ | |
1391 | else if (! spec_long && base == 10 | |
1392 | && int_fits_type_p (value, integer_type_node)) | |
1393 | trad_type = spec_unsigned ? unsigned_type_node : integer_type_node; | |
1394 | else if (! spec_long_long) | |
1395 | trad_type = (spec_unsigned | |
1396 | ? long_unsigned_type_node | |
1397 | : long_integer_type_node); | |
1398 | else if (int_fits_type_p (value, | |
1399 | spec_unsigned | |
1400 | ? long_long_unsigned_type_node | |
1401 | : long_long_integer_type_node)) | |
1402 | trad_type = (spec_unsigned | |
1403 | ? long_long_unsigned_type_node | |
1404 | : long_long_integer_type_node); | |
1405 | else | |
1406 | trad_type = (spec_unsigned | |
1407 | ? widest_unsigned_literal_type_node | |
1408 | : widest_integer_literal_type_node); | |
1409 | } | |
1410 | if (warn_traditional || ! flag_traditional) | |
1411 | { | |
1412 | /* Calculate the ISO type. */ | |
1413 | if (! spec_long && ! spec_unsigned | |
1414 | && int_fits_type_p (value, integer_type_node)) | |
1415 | ansi_type = integer_type_node; | |
1416 | else if (! spec_long && (base != 10 || spec_unsigned) | |
1417 | && int_fits_type_p (value, unsigned_type_node)) | |
1418 | ansi_type = unsigned_type_node; | |
1419 | else if (! spec_unsigned && !spec_long_long | |
1420 | && int_fits_type_p (value, long_integer_type_node)) | |
1421 | ansi_type = long_integer_type_node; | |
1422 | else if (! spec_long_long | |
1423 | && int_fits_type_p (value, long_unsigned_type_node)) | |
1424 | ansi_type = long_unsigned_type_node; | |
1425 | else if (! spec_unsigned | |
1426 | && int_fits_type_p (value, long_long_integer_type_node)) | |
1427 | ansi_type = long_long_integer_type_node; | |
1428 | else if (int_fits_type_p (value, long_long_unsigned_type_node)) | |
1429 | ansi_type = long_long_unsigned_type_node; | |
1430 | else if (! spec_unsigned | |
1431 | && int_fits_type_p (value, widest_integer_literal_type_node)) | |
1432 | ansi_type = widest_integer_literal_type_node; | |
1433 | else | |
1434 | ansi_type = widest_unsigned_literal_type_node; | |
1435 | } | |
e8bbfc4e | 1436 | |
0e5921e8 | 1437 | type = flag_traditional ? trad_type : ansi_type; |
e8bbfc4e | 1438 | |
0e5921e8 ZW |
1439 | /* We assume that constants specified in a non-decimal |
1440 | base are bit patterns, and that the programmer really | |
1441 | meant what they wrote. */ | |
1442 | if (warn_traditional && !in_system_header | |
1443 | && base == 10 && trad_type != ansi_type) | |
1444 | { | |
1445 | if (TYPE_PRECISION (trad_type) != TYPE_PRECISION (ansi_type)) | |
1446 | warning ("width of integer constant changes with -traditional"); | |
1447 | else if (TREE_UNSIGNED (trad_type) != TREE_UNSIGNED (ansi_type)) | |
1448 | warning ("integer constant is unsigned in ISO C, signed with -traditional"); | |
1449 | else | |
1450 | warning ("width of integer constant may change on other systems with -traditional"); | |
1451 | } | |
e8bbfc4e | 1452 | |
b15ad712 JM |
1453 | if (pedantic && !flag_traditional && (flag_isoc99 || !spec_long_long) |
1454 | && !warn | |
1455 | && ((flag_isoc99 | |
1456 | ? TYPE_PRECISION (long_long_integer_type_node) | |
1457 | : TYPE_PRECISION (long_integer_type_node)) < TYPE_PRECISION (type))) | |
0e5921e8 ZW |
1458 | { |
1459 | warn = 1; | |
b15ad712 JM |
1460 | pedwarn ("integer constant larger than the maximum value of %s", |
1461 | (flag_isoc99 | |
1462 | ? (TREE_UNSIGNED (type) | |
1463 | ? "an unsigned long long int" | |
1464 | : "a long long int") | |
1465 | : "an unsigned long int")); | |
0e5921e8 | 1466 | } |
e8bbfc4e | 1467 | |
0e5921e8 ZW |
1468 | if (base == 10 && ! spec_unsigned && TREE_UNSIGNED (type)) |
1469 | warning ("decimal constant is so large that it is unsigned"); | |
e8bbfc4e | 1470 | |
0e5921e8 ZW |
1471 | if (spec_imag) |
1472 | { | |
1473 | if (TYPE_PRECISION (type) | |
1474 | <= TYPE_PRECISION (integer_type_node)) | |
1475 | value = build_complex (NULL_TREE, integer_zero_node, | |
1476 | convert (integer_type_node, value)); | |
1477 | else | |
1478 | ERROR ("complex integer constant is too wide for 'complex int'"); | |
1479 | } | |
1480 | else if (flag_traditional && !int_fits_type_p (value, type)) | |
1481 | /* The traditional constant 0x80000000 is signed | |
1482 | but doesn't fit in the range of int. | |
1483 | This will change it to -0x80000000, which does fit. */ | |
1484 | { | |
1485 | TREE_TYPE (value) = unsigned_type (type); | |
1486 | value = convert (type, value); | |
1487 | TREE_OVERFLOW (value) = TREE_CONSTANT_OVERFLOW (value) = 0; | |
1488 | } | |
1489 | else | |
1490 | TREE_TYPE (value) = type; | |
e8bbfc4e | 1491 | |
0e5921e8 ZW |
1492 | /* If it's still an integer (not a complex), and it doesn't |
1493 | fit in the type we choose for it, then pedwarn. */ | |
fbb18613 | 1494 | |
0e5921e8 ZW |
1495 | if (! warn |
1496 | && TREE_CODE (TREE_TYPE (value)) == INTEGER_TYPE | |
1497 | && ! int_fits_type_p (value, TREE_TYPE (value))) | |
1498 | pedwarn ("integer constant is larger than the maximum value for its type"); | |
1499 | } | |
fbb18613 | 1500 | |
0e5921e8 ZW |
1501 | if (p < str + len) |
1502 | error ("missing white space after number '%.*s'", (int) (p - str), str); | |
e8bbfc4e | 1503 | |
0e5921e8 | 1504 | return value; |
e8bbfc4e | 1505 | |
0e5921e8 ZW |
1506 | syntax_error: |
1507 | return integer_zero_node; | |
1508 | } | |
e8bbfc4e | 1509 | |
0e5921e8 ZW |
1510 | static tree |
1511 | lex_string (str, len, wide) | |
1512 | const char *str; | |
1513 | unsigned int len; | |
1514 | int wide; | |
1515 | { | |
1516 | tree value; | |
1517 | char *buf = alloca ((len + 1) * (wide ? WCHAR_BYTES : 1)); | |
1518 | char *q = buf; | |
1519 | const char *p = str, *limit = str + len; | |
1520 | unsigned int c; | |
1521 | unsigned width = wide ? WCHAR_TYPE_SIZE | |
1522 | : TYPE_PRECISION (char_type_node); | |
e9a25f70 | 1523 | |
0e5921e8 ZW |
1524 | #ifdef MULTIBYTE_CHARS |
1525 | /* Reset multibyte conversion state. */ | |
9714cf43 | 1526 | (void) local_mbtowc (NULL, NULL, 0); |
0e5921e8 | 1527 | #endif |
e9a25f70 | 1528 | |
0e5921e8 ZW |
1529 | while (p < limit) |
1530 | { | |
1531 | #ifdef MULTIBYTE_CHARS | |
1532 | wchar_t wc; | |
1533 | int char_len; | |
1534 | ||
1535 | char_len = local_mbtowc (&wc, p, limit - p); | |
1536 | if (char_len == -1) | |
1537 | { | |
1538 | warning ("Ignoring invalid multibyte character"); | |
1539 | char_len = 1; | |
1540 | c = *p++; | |
1541 | } | |
1542 | else | |
1543 | { | |
1544 | p += char_len; | |
1545 | c = wc; | |
1546 | } | |
1547 | #else | |
1548 | c = *p++; | |
1549 | #endif | |
1550 | ||
1551 | if (c == '\\' && !ignore_escape_flag) | |
1552 | { | |
1553 | p = readescape (p, limit, &c); | |
1554 | if (width < HOST_BITS_PER_INT | |
1555 | && (unsigned) c >= ((unsigned)1 << width)) | |
1556 | pedwarn ("escape sequence out of range for character"); | |
1557 | } | |
1558 | ||
1559 | /* Add this single character into the buffer either as a wchar_t | |
1560 | or as a single byte. */ | |
1561 | if (wide) | |
1562 | { | |
1563 | unsigned charwidth = TYPE_PRECISION (char_type_node); | |
5c80f6e6 | 1564 | unsigned bytemask = (1 << charwidth) - 1; |
0e5921e8 ZW |
1565 | int byte; |
1566 | ||
1567 | for (byte = 0; byte < WCHAR_BYTES; ++byte) | |
1568 | { | |
1569 | int n; | |
1570 | if (byte >= (int) sizeof (c)) | |
1571 | n = 0; | |
1572 | else | |
1573 | n = (c >> (byte * charwidth)) & bytemask; | |
1574 | if (BYTES_BIG_ENDIAN) | |
1575 | q[WCHAR_BYTES - byte - 1] = n; | |
1576 | else | |
1577 | q[byte] = n; | |
1578 | } | |
1579 | q += WCHAR_BYTES; | |
1580 | } | |
1581 | else | |
1582 | { | |
1583 | *q++ = c; | |
1584 | } | |
e8bbfc4e RK |
1585 | } |
1586 | ||
0e5921e8 ZW |
1587 | /* Terminate the string value, either with a single byte zero |
1588 | or with a wide zero. */ | |
e8bbfc4e | 1589 | |
0e5921e8 ZW |
1590 | if (wide) |
1591 | { | |
1592 | memset (q, 0, WCHAR_BYTES); | |
1593 | q += WCHAR_BYTES; | |
1594 | } | |
1595 | else | |
1596 | { | |
1597 | *q++ = '\0'; | |
1598 | } | |
1599 | ||
1600 | value = build_string (q - buf, buf); | |
1601 | ||
1602 | if (wide) | |
1603 | TREE_TYPE (value) = wchar_array_type_node; | |
1604 | else | |
1605 | TREE_TYPE (value) = char_array_type_node; | |
e8bbfc4e RK |
1606 | return value; |
1607 | } | |
1608 | ||
c8a96070 | 1609 | /* Converts a (possibly wide) character constant token into a tree. */ |
0e5921e8 | 1610 | static tree |
c8a96070 NB |
1611 | lex_charconst (token) |
1612 | const cpp_token *token; | |
e8bbfc4e | 1613 | { |
c8a96070 | 1614 | HOST_WIDE_INT result; |
0e5921e8 | 1615 | tree value; |
c8a96070 NB |
1616 | unsigned int chars_seen; |
1617 | ||
1618 | result = cpp_interpret_charconst (parse_in, token, warn_multichar, | |
1619 | flag_traditional, &chars_seen); | |
1620 | if (token->type == CPP_WCHAR) | |
0e5921e8 | 1621 | { |
c8a96070 NB |
1622 | value = build_int_2 (result, 0); |
1623 | TREE_TYPE (value) = wchar_type_node; | |
0e5921e8 | 1624 | } |
c8a96070 | 1625 | else |
0e5921e8 | 1626 | { |
c8a96070 NB |
1627 | if (result < 0) |
1628 | value = build_int_2 (result, -1); | |
0e5921e8 | 1629 | else |
c8a96070 NB |
1630 | value = build_int_2 (result, 0); |
1631 | ||
1632 | /* In C, a character constant has type 'int'. | |
1633 | In C++ 'char', but multi-char charconsts have type 'int'. */ | |
1634 | if (c_language == clk_cplusplus && chars_seen <= 1) | |
1635 | TREE_TYPE (value) = char_type_node; | |
0e5921e8 | 1636 | else |
c8a96070 | 1637 | TREE_TYPE (value) = integer_type_node; |
0e5921e8 | 1638 | } |
c8a96070 | 1639 | |
0e5921e8 | 1640 | return value; |
e8bbfc4e | 1641 | } |