1 /* Lexical analyzer for C and Objective C.
2 Copyright (C) 1987, 1988, 1989, 1992, 1994, 1995, 1996, 1997
3 1998, 1999, 2000 Free Software Foundation, Inc.
5 This file is part of GNU CC.
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)
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.
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
19 the Free Software Foundation, 59 Temple Place - Suite 330,
20 Boston, MA 02111-1307, USA. */
39 #include "splay-tree.h"
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. */
45 #undef MULTIBYTE_CHARS
48 #ifdef MULTIBYTE_CHARS
51 #endif /* MULTIBYTE_CHARS */
52 #ifndef GET_ENVIRONMENT
53 #define GET_ENVIRONMENT(ENV_VALUE,ENV_NAME) ((ENV_VALUE) = getenv (ENV_NAME))
56 /* The input filename as understood by CPP, where "" represents stdin. */
57 static const char *cpp_filename
;
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
;
63 /* Cause the `yydebug' variable to be defined. */
66 /* File used for outputting assembler code. */
67 extern FILE *asm_out_file
;
69 #undef WCHAR_TYPE_SIZE
70 #define WCHAR_TYPE_SIZE TYPE_PRECISION (wchar_type_node)
72 /* Number of bytes in a wide character. */
73 #define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT)
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 */
79 /* Nonzero tells yylex to ignore \ in string constants. */
80 static int ignore_escape_flag
;
82 static void parse_float
PARAMS ((PTR
));
83 static tree lex_number
PARAMS ((const char *, unsigned int));
84 static tree lex_string
PARAMS ((const char *, unsigned int, int));
85 static tree lex_charconst
PARAMS ((const cpp_token
*));
86 static void update_header_times
PARAMS ((const char *));
87 static int dump_one_header
PARAMS ((splay_tree_node
, void *));
88 static void cb_ident
PARAMS ((cpp_reader
*, const cpp_string
*));
89 static void cb_file_change
PARAMS ((cpp_reader
*, const cpp_file_change
*));
90 static void cb_def_pragma
PARAMS ((cpp_reader
*));
91 static void cb_define
PARAMS ((cpp_reader
*, cpp_hashnode
*));
92 static void cb_undef
PARAMS ((cpp_reader
*, cpp_hashnode
*));
98 struct cpp_callbacks
*cb
;
99 struct c_fileinfo
*toplevel
;
101 /* Set up filename timing. Must happen before cpp_start_read. */
102 file_info_tree
= splay_tree_new ((splay_tree_compare_fn
)strcmp
,
104 (splay_tree_delete_value_fn
)free
);
105 toplevel
= get_fileinfo ("<top level>");
106 if (flag_detailed_statistics
)
109 body_time
= get_run_time ();
110 toplevel
->time
= body_time
;
113 #ifdef MULTIBYTE_CHARS
114 /* Change to the native locale for multibyte conversions. */
115 setlocale (LC_CTYPE
, "");
116 GET_ENVIRONMENT (literal_codeset
, "LANG");
119 cb
= cpp_get_callbacks (parse_in
);
121 cb
->ident
= cb_ident
;
122 cb
->file_change
= cb_file_change
;
123 cb
->def_pragma
= cb_def_pragma
;
125 /* Set the debug callbacks if we can use them. */
126 if (debug_info_level
== DINFO_LEVEL_VERBOSE
127 && (write_symbols
== DWARF_DEBUG
|| write_symbols
== DWARF2_DEBUG
))
129 cb
->define
= cb_define
;
130 cb
->undef
= cb_undef
;
134 if (filename
== 0 || !strcmp (filename
, "-"))
135 filename
= "stdin", cpp_filename
= "";
137 cpp_filename
= filename
;
145 /* A thin wrapper around the real parser that initializes the
146 integrated preprocessor after debug output has been initialized. */
151 if (! cpp_start_read (parse_in
, cpp_filename
))
152 return 1; /* cpplib has emitted an error. */
162 struct c_fileinfo
*fi
;
164 n
= splay_tree_lookup (file_info_tree
, (splay_tree_key
) name
);
166 return (struct c_fileinfo
*) n
->value
;
168 fi
= (struct c_fileinfo
*) xmalloc (sizeof (struct c_fileinfo
));
170 fi
->interface_only
= 0;
171 fi
->interface_unknown
= 1;
172 splay_tree_insert (file_info_tree
, (splay_tree_key
) name
,
173 (splay_tree_value
) fi
);
178 update_header_times (name
)
181 /* Changing files again. This means currently collected time
182 is charged against header time, and body time starts back at 0. */
183 if (flag_detailed_statistics
)
185 int this_time
= get_run_time ();
186 struct c_fileinfo
*file
= get_fileinfo (name
);
187 header_time
+= this_time
- body_time
;
188 file
->time
+= this_time
- body_time
;
189 body_time
= this_time
;
194 dump_one_header (n
, dummy
)
196 void *dummy ATTRIBUTE_UNUSED
;
198 print_time ((const char *) n
->key
,
199 ((struct c_fileinfo
*) n
->value
)->time
);
204 dump_time_statistics ()
206 struct c_fileinfo
*file
= get_fileinfo (input_filename
);
207 int this_time
= get_run_time ();
208 file
->time
+= this_time
- body_time
;
210 fprintf (stderr
, "\n******\n");
211 print_time ("header files (total)", header_time
);
212 print_time ("main file (total)", this_time
- body_time
);
213 fprintf (stderr
, "ratio = %g : 1\n",
214 (double)header_time
/ (double)(this_time
- body_time
));
215 fprintf (stderr
, "\n******\n");
217 splay_tree_foreach (file_info_tree
, dump_one_header
, 0);
220 /* Not yet handled: #pragma, #define, #undef.
221 No need to deal with linemarkers under normal conditions. */
224 cb_ident (pfile
, str
)
225 cpp_reader
*pfile ATTRIBUTE_UNUSED
;
226 const cpp_string
*str ATTRIBUTE_UNUSED
;
228 #ifdef ASM_OUTPUT_IDENT
231 /* Convert escapes in the string. */
232 tree value
= lex_string ((const char *)str
->text
, str
->len
, 0);
233 ASM_OUTPUT_IDENT (asm_out_file
, TREE_STRING_POINTER (value
));
239 cb_file_change (pfile
, fc
)
240 cpp_reader
*pfile ATTRIBUTE_UNUSED
;
241 const cpp_file_change
*fc
;
243 if (fc
->reason
== FC_ENTER
)
245 /* Don't stack the main buffer on the input stack;
246 we already did in compile_file. */
247 if (fc
->from
.filename
)
249 lineno
= fc
->from
.lineno
;
250 push_srcloc (fc
->to
.filename
, 1);
251 input_file_stack
->indent_level
= indent_level
;
252 debug_start_source_file (fc
->to
.filename
);
253 #ifndef NO_IMPLICIT_EXTERN_C
256 else if (fc
->externc
)
259 ++pending_lang_change
;
264 main_input_filename
= fc
->to
.filename
;
266 else if (fc
->reason
== FC_LEAVE
)
268 /* Popping out of a file. */
269 if (input_file_stack
->next
)
271 #ifndef NO_IMPLICIT_EXTERN_C
272 if (c_header_level
&& --c_header_level
== 0)
275 warning ("badly nested C headers from preprocessor");
276 --pending_lang_change
;
280 if (indent_level
!= input_file_stack
->indent_level
)
282 warning_with_file_and_line
283 (input_filename
, lineno
,
284 "This file contains more '%c's than '%c's.",
285 indent_level
> input_file_stack
->indent_level
? '{' : '}',
286 indent_level
> input_file_stack
->indent_level
? '}' : '{');
290 debug_end_source_file (input_file_stack
->line
);
293 error ("leaving more files than we entered");
296 update_header_times (fc
->to
.filename
);
297 in_system_header
= fc
->sysp
!= 0;
298 input_filename
= fc
->to
.filename
;
299 lineno
= fc
->to
.lineno
; /* Do we need this? */
302 extract_interface_info ();
306 cb_def_pragma (pfile
)
309 /* Issue a warning message if we have been asked to do so. Ignore
310 unknown pragmas in system headers unless an explicit
311 -Wunknown-pragmas has been given. */
312 if (warn_unknown_pragmas
> in_system_header
)
314 const unsigned char *space
, *name
= 0;
317 cpp_get_token (pfile
, &s
);
318 space
= cpp_token_as_text (pfile
, &s
);
319 cpp_get_token (pfile
, &s
);
320 if (s
.type
== CPP_NAME
)
321 name
= cpp_token_as_text (pfile
, &s
);
323 lineno
= cpp_get_line (parse_in
)->line
;
325 warning ("ignoring #pragma %s %s", space
, name
);
327 warning ("ignoring #pragma %s", space
);
331 /* #define callback for DWARF and DWARF2 debug info. */
333 cb_define (pfile
, node
)
337 debug_define (lineno
, (const char *) cpp_macro_definition (pfile
, node
));
340 /* #undef callback for DWARF and DWARF2 debug info. */
342 cb_undef (pfile
, node
)
343 cpp_reader
*pfile ATTRIBUTE_UNUSED
;
346 debug_undef (lineno
, (const char *) NODE_NAME (node
));
350 /* Returns nonzero if C is a universal-character-name. Give an error if it
351 is not one which may appear in an identifier, as per [extendid].
353 Note that extended character support in identifiers has not yet been
354 implemented. It is my personal opinion that this is not a desirable
355 feature. Portable code cannot count on support for more than the basic
356 identifier character set. */
369 /* None of the valid chars are outside the Basic Multilingual Plane (the
373 error ("universal-character-name '\\U%08x' not valid in identifier", c
);
378 if ((c
>= 0x00c0 && c
<= 0x00d6)
379 || (c
>= 0x00d8 && c
<= 0x00f6)
380 || (c
>= 0x00f8 && c
<= 0x01f5)
381 || (c
>= 0x01fa && c
<= 0x0217)
382 || (c
>= 0x0250 && c
<= 0x02a8)
383 || (c
>= 0x1e00 && c
<= 0x1e9a)
384 || (c
>= 0x1ea0 && c
<= 0x1ef9))
389 || (c
>= 0x0388 && c
<= 0x038a)
391 || (c
>= 0x038e && c
<= 0x03a1)
392 || (c
>= 0x03a3 && c
<= 0x03ce)
393 || (c
>= 0x03d0 && c
<= 0x03d6)
398 || (c
>= 0x03e2 && c
<= 0x03f3)
399 || (c
>= 0x1f00 && c
<= 0x1f15)
400 || (c
>= 0x1f18 && c
<= 0x1f1d)
401 || (c
>= 0x1f20 && c
<= 0x1f45)
402 || (c
>= 0x1f48 && c
<= 0x1f4d)
403 || (c
>= 0x1f50 && c
<= 0x1f57)
407 || (c
>= 0x1f5f && c
<= 0x1f7d)
408 || (c
>= 0x1f80 && c
<= 0x1fb4)
409 || (c
>= 0x1fb6 && c
<= 0x1fbc)
410 || (c
>= 0x1fc2 && c
<= 0x1fc4)
411 || (c
>= 0x1fc6 && c
<= 0x1fcc)
412 || (c
>= 0x1fd0 && c
<= 0x1fd3)
413 || (c
>= 0x1fd6 && c
<= 0x1fdb)
414 || (c
>= 0x1fe0 && c
<= 0x1fec)
415 || (c
>= 0x1ff2 && c
<= 0x1ff4)
416 || (c
>= 0x1ff6 && c
<= 0x1ffc))
420 if ((c
>= 0x0401 && c
<= 0x040d)
421 || (c
>= 0x040f && c
<= 0x044f)
422 || (c
>= 0x0451 && c
<= 0x045c)
423 || (c
>= 0x045e && c
<= 0x0481)
424 || (c
>= 0x0490 && c
<= 0x04c4)
425 || (c
>= 0x04c7 && c
<= 0x04c8)
426 || (c
>= 0x04cb && c
<= 0x04cc)
427 || (c
>= 0x04d0 && c
<= 0x04eb)
428 || (c
>= 0x04ee && c
<= 0x04f5)
429 || (c
>= 0x04f8 && c
<= 0x04f9))
433 if ((c
>= 0x0531 && c
<= 0x0556)
434 || (c
>= 0x0561 && c
<= 0x0587))
438 if ((c
>= 0x05d0 && c
<= 0x05ea)
439 || (c
>= 0x05f0 && c
<= 0x05f4))
443 if ((c
>= 0x0621 && c
<= 0x063a)
444 || (c
>= 0x0640 && c
<= 0x0652)
445 || (c
>= 0x0670 && c
<= 0x06b7)
446 || (c
>= 0x06ba && c
<= 0x06be)
447 || (c
>= 0x06c0 && c
<= 0x06ce)
448 || (c
>= 0x06e5 && c
<= 0x06e7))
452 if ((c
>= 0x0905 && c
<= 0x0939)
453 || (c
>= 0x0958 && c
<= 0x0962))
457 if ((c
>= 0x0985 && c
<= 0x098c)
458 || (c
>= 0x098f && c
<= 0x0990)
459 || (c
>= 0x0993 && c
<= 0x09a8)
460 || (c
>= 0x09aa && c
<= 0x09b0)
462 || (c
>= 0x09b6 && c
<= 0x09b9)
463 || (c
>= 0x09dc && c
<= 0x09dd)
464 || (c
>= 0x09df && c
<= 0x09e1)
465 || (c
>= 0x09f0 && c
<= 0x09f1))
469 if ((c
>= 0x0a05 && c
<= 0x0a0a)
470 || (c
>= 0x0a0f && c
<= 0x0a10)
471 || (c
>= 0x0a13 && c
<= 0x0a28)
472 || (c
>= 0x0a2a && c
<= 0x0a30)
473 || (c
>= 0x0a32 && c
<= 0x0a33)
474 || (c
>= 0x0a35 && c
<= 0x0a36)
475 || (c
>= 0x0a38 && c
<= 0x0a39)
476 || (c
>= 0x0a59 && c
<= 0x0a5c)
481 if ((c
>= 0x0a85 && c
<= 0x0a8b)
483 || (c
>= 0x0a8f && c
<= 0x0a91)
484 || (c
>= 0x0a93 && c
<= 0x0aa8)
485 || (c
>= 0x0aaa && c
<= 0x0ab0)
486 || (c
>= 0x0ab2 && c
<= 0x0ab3)
487 || (c
>= 0x0ab5 && c
<= 0x0ab9)
492 if ((c
>= 0x0b05 && c
<= 0x0b0c)
493 || (c
>= 0x0b0f && c
<= 0x0b10)
494 || (c
>= 0x0b13 && c
<= 0x0b28)
495 || (c
>= 0x0b2a && c
<= 0x0b30)
496 || (c
>= 0x0b32 && c
<= 0x0b33)
497 || (c
>= 0x0b36 && c
<= 0x0b39)
498 || (c
>= 0x0b5c && c
<= 0x0b5d)
499 || (c
>= 0x0b5f && c
<= 0x0b61))
503 if ((c
>= 0x0b85 && c
<= 0x0b8a)
504 || (c
>= 0x0b8e && c
<= 0x0b90)
505 || (c
>= 0x0b92 && c
<= 0x0b95)
506 || (c
>= 0x0b99 && c
<= 0x0b9a)
508 || (c
>= 0x0b9e && c
<= 0x0b9f)
509 || (c
>= 0x0ba3 && c
<= 0x0ba4)
510 || (c
>= 0x0ba8 && c
<= 0x0baa)
511 || (c
>= 0x0bae && c
<= 0x0bb5)
512 || (c
>= 0x0bb7 && c
<= 0x0bb9))
516 if ((c
>= 0x0c05 && c
<= 0x0c0c)
517 || (c
>= 0x0c0e && c
<= 0x0c10)
518 || (c
>= 0x0c12 && c
<= 0x0c28)
519 || (c
>= 0x0c2a && c
<= 0x0c33)
520 || (c
>= 0x0c35 && c
<= 0x0c39)
521 || (c
>= 0x0c60 && c
<= 0x0c61))
525 if ((c
>= 0x0c85 && c
<= 0x0c8c)
526 || (c
>= 0x0c8e && c
<= 0x0c90)
527 || (c
>= 0x0c92 && c
<= 0x0ca8)
528 || (c
>= 0x0caa && c
<= 0x0cb3)
529 || (c
>= 0x0cb5 && c
<= 0x0cb9)
530 || (c
>= 0x0ce0 && c
<= 0x0ce1))
534 if ((c
>= 0x0d05 && c
<= 0x0d0c)
535 || (c
>= 0x0d0e && c
<= 0x0d10)
536 || (c
>= 0x0d12 && c
<= 0x0d28)
537 || (c
>= 0x0d2a && c
<= 0x0d39)
538 || (c
>= 0x0d60 && c
<= 0x0d61))
542 if ((c
>= 0x0e01 && c
<= 0x0e30)
543 || (c
>= 0x0e32 && c
<= 0x0e33)
544 || (c
>= 0x0e40 && c
<= 0x0e46)
545 || (c
>= 0x0e4f && c
<= 0x0e5b))
549 if ((c
>= 0x0e81 && c
<= 0x0e82)
555 || (c
>= 0x0e94 && c
<= 0x0e97)
556 || (c
>= 0x0e99 && c
<= 0x0e9f)
557 || (c
>= 0x0ea1 && c
<= 0x0ea3)
562 || (c
>= 0x0ead && c
<= 0x0eb0)
566 || (c
>= 0x0ec0 && c
<= 0x0ec4)
571 if ((c
>= 0x10a0 && c
<= 0x10c5)
572 || (c
>= 0x10d0 && c
<= 0x10f6))
576 if ((c
>= 0x3041 && c
<= 0x3094)
577 || (c
>= 0x309b && c
<= 0x309e))
581 if ((c
>= 0x30a1 && c
<= 0x30fe))
585 if ((c
>= 0x3105 && c
<= 0x312c))
589 if ((c
>= 0x1100 && c
<= 0x1159)
590 || (c
>= 0x1161 && c
<= 0x11a2)
591 || (c
>= 0x11a8 && c
<= 0x11f9))
594 /* CJK Unified Ideographs */
595 if ((c
>= 0xf900 && c
<= 0xfa2d)
596 || (c
>= 0xfb1f && c
<= 0xfb36)
597 || (c
>= 0xfb38 && c
<= 0xfb3c)
599 || (c
>= 0xfb40 && c
<= 0xfb41)
600 || (c
>= 0xfb42 && c
<= 0xfb44)
601 || (c
>= 0xfb46 && c
<= 0xfbb1)
602 || (c
>= 0xfbd3 && c
<= 0xfd3f)
603 || (c
>= 0xfd50 && c
<= 0xfd8f)
604 || (c
>= 0xfd92 && c
<= 0xfdc7)
605 || (c
>= 0xfdf0 && c
<= 0xfdfb)
606 || (c
>= 0xfe70 && c
<= 0xfe72)
608 || (c
>= 0xfe76 && c
<= 0xfefc)
609 || (c
>= 0xff21 && c
<= 0xff3a)
610 || (c
>= 0xff41 && c
<= 0xff5a)
611 || (c
>= 0xff66 && c
<= 0xffbe)
612 || (c
>= 0xffc2 && c
<= 0xffc7)
613 || (c
>= 0xffca && c
<= 0xffcf)
614 || (c
>= 0xffd2 && c
<= 0xffd7)
615 || (c
>= 0xffda && c
<= 0xffdc)
616 || (c
>= 0x4e00 && c
<= 0x9fa5))
619 error ("universal-character-name '\\u%04x' not valid in identifier", c
);
624 /* Add the UTF-8 representation of C to the token_buffer. */
627 utf8_extend_token (c
)
637 else if (c
<= 0x000007ff)
638 shift
= 6, mask
= 0xc0;
639 else if (c
<= 0x0000ffff)
640 shift
= 12, mask
= 0xe0;
641 else if (c
<= 0x001fffff)
642 shift
= 18, mask
= 0xf0;
643 else if (c
<= 0x03ffffff)
644 shift
= 24, mask
= 0xf8;
646 shift
= 30, mask
= 0xfc;
648 extend_token (mask
| (c
>> shift
));
652 extend_token ((unsigned char) (0x80 | (c
>> shift
)));
667 struct try_type type_sequence
[] =
669 { &integer_type_node
, 0, 0, 0},
670 { &unsigned_type_node
, 1, 0, 0},
671 { &long_integer_type_node
, 0, 1, 0},
672 { &long_unsigned_type_node
, 1, 1, 0},
673 { &long_long_integer_type_node
, 0, 1, 1},
674 { &long_long_unsigned_type_node
, 1, 1, 1}
686 int conversion_errno
;
687 REAL_VALUE_TYPE value
;
695 struct pf_args
* args
= (struct pf_args
*) data
;
696 const char *typename
;
698 args
->conversion_errno
= 0;
699 args
->type
= double_type_node
;
702 /* The second argument, machine_mode, of REAL_VALUE_ATOF
703 tells the desired precision of the binary result
704 of decimal-to-binary conversion. */
709 error ("both 'f' and 'l' suffixes on floating constant");
711 args
->type
= float_type_node
;
714 else if (args
->lflag
)
716 args
->type
= long_double_type_node
;
717 typename
= "long double";
719 else if (flag_single_precision_constant
)
721 args
->type
= float_type_node
;
726 if (args
->base
== 16)
727 args
->value
= REAL_VALUE_HTOF (args
->str
, TYPE_MODE (args
->type
));
729 args
->value
= REAL_VALUE_ATOF (args
->str
, TYPE_MODE (args
->type
));
731 args
->conversion_errno
= errno
;
732 /* A diagnostic is required here by some ISO C testsuites.
733 This is not pedwarn, because some people don't want
734 an error for this. */
735 if (REAL_VALUE_ISINF (args
->value
) && pedantic
)
736 warning ("floating point number exceeds range of '%s'", typename
);
747 timevar_push (TV_CPP
);
748 cpp_get_token (parse_in
, &tok
);
749 timevar_pop (TV_CPP
);
751 /* The C++ front end does horrible things with the current line
752 number. To ensure an accurate line number, we must reset it
753 every time we return a token. */
754 lineno
= cpp_get_line (parse_in
)->line
;
760 case CPP_OPEN_BRACE
: indent_level
++; break;
761 case CPP_CLOSE_BRACE
: indent_level
--; break;
763 /* Issue this error here, where we can get at tok.val.c. */
765 if (ISGRAPH (tok
.val
.c
))
766 error ("stray '%c' in program", tok
.val
.c
);
768 error ("stray '\\%o' in program", tok
.val
.c
);
772 *value
= HT_IDENT_TO_GCC_IDENT (HT_NODE (tok
.val
.node
));
778 *value
= lex_number ((const char *)tok
.val
.str
.text
, tok
.val
.str
.len
);
783 *value
= lex_charconst (&tok
);
788 *value
= lex_string ((const char *)tok
.val
.str
.text
,
789 tok
.val
.str
.len
, tok
.type
== CPP_WSTRING
);
792 /* These tokens should not be visible outside cpplib. */
793 case CPP_HEADER_NAME
:
804 #define ERROR(msgid) do { error(msgid); goto syntax_error; } while(0)
807 lex_number (str
, len
)
813 int largest_digit
= 0;
819 enum anon1
{ NOT_FLOAT
= 0, AFTER_POINT
, AFTER_EXPON
} floatflag
= NOT_FLOAT
;
821 /* We actually store only HOST_BITS_PER_CHAR bits in each part.
822 The code below which fills the parts array assumes that a host
823 int is at least twice as wide as a host char, and that
824 HOST_BITS_PER_WIDE_INT is an even multiple of HOST_BITS_PER_CHAR.
825 Two HOST_WIDE_INTs is the largest int literal we can store.
826 In order to detect overflow below, the number of parts (TOTAL_PARTS)
827 must be exactly the number of parts needed to hold the bits
828 of two HOST_WIDE_INTs. */
829 #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2)
830 unsigned int parts
[TOTAL_PARTS
];
832 /* Optimize for most frequent case. */
836 return integer_zero_node
;
837 else if (*str
== '1')
838 return integer_one_node
;
840 return build_int_2 (*str
- '0', 0);
843 for (count
= 0; count
< TOTAL_PARTS
; count
++)
846 /* len is known to be >1 at this point. */
849 if (len
> 2 && str
[0] == '0' && (str
[1] == 'x' || str
[1] == 'X'))
854 /* The ISDIGIT check is so we are not confused by a suffix on 0. */
855 else if (str
[0] == '0' && ISDIGIT (str
[1]))
867 if (floatflag
== AFTER_POINT
)
868 ERROR ("too many decimal points in floating constant");
869 else if (floatflag
== AFTER_EXPON
)
870 ERROR ("decimal point in exponent - impossible!");
872 floatflag
= AFTER_POINT
;
878 /* Possible future extension: silently ignore _ in numbers,
879 permitting cosmetic grouping - e.g. 0x8000_0000 == 0x80000000
880 but somewhat easier to read. Ada has this? */
881 ERROR ("underscore in number");
885 /* It is not a decimal point.
886 It should be a digit (perhaps a hex digit). */
892 else if (base
<= 10 && (c
== 'e' || c
== 'E'))
895 floatflag
= AFTER_EXPON
;
898 else if (base
== 16 && (c
== 'p' || c
== 'P'))
900 floatflag
= AFTER_EXPON
;
901 break; /* start of exponent */
903 else if (base
== 16 && c
>= 'a' && c
<= 'f')
907 else if (base
== 16 && c
>= 'A' && c
<= 'F')
914 break; /* start of suffix */
917 if (n
>= largest_digit
)
921 for (count
= 0; count
< TOTAL_PARTS
; count
++)
923 parts
[count
] *= base
;
927 += (parts
[count
-1] >> HOST_BITS_PER_CHAR
);
929 &= (1 << HOST_BITS_PER_CHAR
) - 1;
935 /* If the highest-order part overflows (gets larger than
936 a host char will hold) then the whole number has
937 overflowed. Record this and truncate the highest-order
939 if (parts
[TOTAL_PARTS
- 1] >> HOST_BITS_PER_CHAR
)
942 parts
[TOTAL_PARTS
- 1] &= (1 << HOST_BITS_PER_CHAR
) - 1;
946 while (p
< str
+ len
);
948 /* This can happen on input like `int i = 0x;' */
950 ERROR ("numeric constant with no digits");
952 if (largest_digit
>= base
)
953 ERROR ("numeric constant contains digits beyond the radix");
955 if (floatflag
!= NOT_FLOAT
)
958 int imag
, fflag
, lflag
, conversion_errno
;
959 REAL_VALUE_TYPE real
;
963 if (base
== 16 && pedantic
&& !flag_isoc99
)
964 pedwarn ("floating constant may not be in radix 16");
966 if (base
== 16 && floatflag
!= AFTER_EXPON
)
967 ERROR ("hexadecimal floating constant has no exponent");
969 /* Read explicit exponent if any, and put it in tokenbuf. */
970 if ((base
== 10 && ((c
== 'e') || (c
== 'E')))
971 || (base
== 16 && (c
== 'p' || c
== 'P')))
975 if (p
< str
+ len
&& (c
== '+' || c
== '-'))
977 /* Exponent is decimal, even if string is a hex float. */
979 ERROR ("floating constant exponent has no digits");
980 while (p
< str
+ len
&& ISDIGIT (c
))
986 /* Copy the float constant now; we don't want any suffixes in the
987 string passed to parse_float. */
988 copy
= alloca (p
- str
+ 1);
989 memcpy (copy
, str
, p
- str
);
990 copy
[p
- str
] = '\0';
992 /* Now parse suffixes. */
993 fflag
= lflag
= imag
= 0;
994 while (p
< str
+ len
)
999 ERROR ("more than one 'f' suffix on floating constant");
1000 else if (warn_traditional
&& !in_system_header
1001 && ! cpp_sys_macro_p (parse_in
))
1002 warning ("traditional C rejects the 'f' suffix");
1009 ERROR ("more than one 'l' suffix on floating constant");
1010 else if (warn_traditional
&& !in_system_header
1011 && ! cpp_sys_macro_p (parse_in
))
1012 warning ("traditional C rejects the 'l' suffix");
1020 ERROR ("more than one 'i' or 'j' suffix on floating constant");
1022 pedwarn ("ISO C forbids imaginary numeric constants");
1027 ERROR ("invalid suffix on floating constant");
1030 /* Setup input for parse_float() */
1036 /* Convert string to a double, checking for overflow. */
1037 if (do_float_handler (parse_float
, (PTR
) &args
))
1039 /* Receive output from parse_float() */
1043 /* We got an exception from parse_float() */
1044 ERROR ("floating constant out of range");
1046 /* Receive output from parse_float() */
1047 conversion_errno
= args
.conversion_errno
;
1051 /* ERANGE is also reported for underflow,
1052 so test the value to distinguish overflow from that. */
1053 if (conversion_errno
== ERANGE
&& !flag_traditional
&& pedantic
1054 && (REAL_VALUES_LESS (dconst1
, real
)
1055 || REAL_VALUES_LESS (real
, dconstm1
)))
1056 warning ("floating point number exceeds range of 'double'");
1059 /* Create a node with determined type and value. */
1061 value
= build_complex (NULL_TREE
, convert (type
, integer_zero_node
),
1062 build_real (type
, real
));
1064 value
= build_real (type
, real
);
1068 tree trad_type
, ansi_type
, type
;
1069 HOST_WIDE_INT high
, low
;
1070 int spec_unsigned
= 0;
1072 int spec_long_long
= 0;
1077 trad_type
= ansi_type
= type
= NULL_TREE
;
1078 while (p
< str
+ len
)
1085 error ("two 'u' suffixes on integer constant");
1086 else if (warn_traditional
&& !in_system_header
1087 && ! cpp_sys_macro_p (parse_in
))
1088 warning ("traditional C rejects the 'u' suffix");
1099 error ("three 'l' suffixes on integer constant");
1101 error ("'lul' is not a valid integer suffix");
1102 else if (c
!= spec_long
)
1103 error ("'Ll' and 'lL' are not valid integer suffixes");
1104 else if (pedantic
&& ! flag_isoc99
1105 && ! in_system_header
&& warn_long_long
)
1106 pedwarn ("ISO C89 forbids long long integer constants");
1112 case 'i': case 'I': case 'j': case 'J':
1114 error ("more than one 'i' or 'j' suffix on integer constant");
1116 pedwarn ("ISO C forbids imaginary numeric constants");
1121 ERROR ("invalid suffix on integer constant");
1125 /* If the literal overflowed, pedwarn about it now. */
1129 pedwarn ("integer constant is too large for this configuration of the compiler - truncated to %d bits", HOST_BITS_PER_WIDE_INT
* 2);
1132 /* This is simplified by the fact that our constant
1133 is always positive. */
1137 for (i
= 0; i
< HOST_BITS_PER_WIDE_INT
/ HOST_BITS_PER_CHAR
; i
++)
1139 high
|= ((HOST_WIDE_INT
) parts
[i
+ (HOST_BITS_PER_WIDE_INT
1140 / HOST_BITS_PER_CHAR
)]
1141 << (i
* HOST_BITS_PER_CHAR
));
1142 low
|= (HOST_WIDE_INT
) parts
[i
] << (i
* HOST_BITS_PER_CHAR
);
1145 value
= build_int_2 (low
, high
);
1146 TREE_TYPE (value
) = long_long_unsigned_type_node
;
1148 /* If warn_traditional, calculate both the ISO type and the
1149 traditional type, then see if they disagree.
1150 Otherwise, calculate only the type for the dialect in use. */
1151 if (warn_traditional
|| flag_traditional
)
1153 /* Calculate the traditional type. */
1154 /* Traditionally, any constant is signed; but if unsigned is
1155 specified explicitly, obey that. Use the smallest size
1156 with the right number of bits, except for one special
1157 case with decimal constants. */
1158 if (! spec_long
&& base
!= 10
1159 && int_fits_type_p (value
, unsigned_type_node
))
1160 trad_type
= spec_unsigned
? unsigned_type_node
: integer_type_node
;
1161 /* A decimal constant must be long if it does not fit in
1162 type int. I think this is independent of whether the
1163 constant is signed. */
1164 else if (! spec_long
&& base
== 10
1165 && int_fits_type_p (value
, integer_type_node
))
1166 trad_type
= spec_unsigned
? unsigned_type_node
: integer_type_node
;
1167 else if (! spec_long_long
)
1168 trad_type
= (spec_unsigned
1169 ? long_unsigned_type_node
1170 : long_integer_type_node
);
1171 else if (int_fits_type_p (value
,
1173 ? long_long_unsigned_type_node
1174 : long_long_integer_type_node
))
1175 trad_type
= (spec_unsigned
1176 ? long_long_unsigned_type_node
1177 : long_long_integer_type_node
);
1179 trad_type
= (spec_unsigned
1180 ? widest_unsigned_literal_type_node
1181 : widest_integer_literal_type_node
);
1183 if (warn_traditional
|| ! flag_traditional
)
1185 /* Calculate the ISO type. */
1186 if (! spec_long
&& ! spec_unsigned
1187 && int_fits_type_p (value
, integer_type_node
))
1188 ansi_type
= integer_type_node
;
1189 else if (! spec_long
&& (base
!= 10 || spec_unsigned
)
1190 && int_fits_type_p (value
, unsigned_type_node
))
1191 ansi_type
= unsigned_type_node
;
1192 else if (! spec_unsigned
&& !spec_long_long
1193 && int_fits_type_p (value
, long_integer_type_node
))
1194 ansi_type
= long_integer_type_node
;
1195 else if (! spec_long_long
1196 && int_fits_type_p (value
, long_unsigned_type_node
))
1197 ansi_type
= long_unsigned_type_node
;
1198 else if (! spec_unsigned
1199 && int_fits_type_p (value
, long_long_integer_type_node
))
1200 ansi_type
= long_long_integer_type_node
;
1201 else if (int_fits_type_p (value
, long_long_unsigned_type_node
))
1202 ansi_type
= long_long_unsigned_type_node
;
1203 else if (! spec_unsigned
1204 && int_fits_type_p (value
, widest_integer_literal_type_node
))
1205 ansi_type
= widest_integer_literal_type_node
;
1207 ansi_type
= widest_unsigned_literal_type_node
;
1210 type
= flag_traditional
? trad_type
: ansi_type
;
1212 /* We assume that constants specified in a non-decimal
1213 base are bit patterns, and that the programmer really
1214 meant what they wrote. */
1215 if (warn_traditional
&& !in_system_header
1216 && base
== 10 && trad_type
!= ansi_type
)
1218 if (TYPE_PRECISION (trad_type
) != TYPE_PRECISION (ansi_type
))
1219 warning ("width of integer constant changes with -traditional");
1220 else if (TREE_UNSIGNED (trad_type
) != TREE_UNSIGNED (ansi_type
))
1221 warning ("integer constant is unsigned in ISO C, signed with -traditional");
1223 warning ("width of integer constant may change on other systems with -traditional");
1226 if (pedantic
&& !flag_traditional
&& (flag_isoc99
|| !spec_long_long
)
1229 ? TYPE_PRECISION (long_long_integer_type_node
)
1230 : TYPE_PRECISION (long_integer_type_node
)) < TYPE_PRECISION (type
)))
1233 pedwarn ("integer constant larger than the maximum value of %s",
1235 ? (TREE_UNSIGNED (type
)
1236 ? "an unsigned long long int"
1237 : "a long long int")
1238 : "an unsigned long int"));
1241 if (base
== 10 && ! spec_unsigned
&& TREE_UNSIGNED (type
))
1242 warning ("decimal constant is so large that it is unsigned");
1246 if (TYPE_PRECISION (type
)
1247 <= TYPE_PRECISION (integer_type_node
))
1248 value
= build_complex (NULL_TREE
, integer_zero_node
,
1249 convert (integer_type_node
, value
));
1251 ERROR ("complex integer constant is too wide for 'complex int'");
1253 else if (flag_traditional
&& !int_fits_type_p (value
, type
))
1254 /* The traditional constant 0x80000000 is signed
1255 but doesn't fit in the range of int.
1256 This will change it to -0x80000000, which does fit. */
1258 TREE_TYPE (value
) = unsigned_type (type
);
1259 value
= convert (type
, value
);
1260 TREE_OVERFLOW (value
) = TREE_CONSTANT_OVERFLOW (value
) = 0;
1263 TREE_TYPE (value
) = type
;
1265 /* If it's still an integer (not a complex), and it doesn't
1266 fit in the type we choose for it, then pedwarn. */
1269 && TREE_CODE (TREE_TYPE (value
)) == INTEGER_TYPE
1270 && ! int_fits_type_p (value
, TREE_TYPE (value
)))
1271 pedwarn ("integer constant is larger than the maximum value for its type");
1275 error ("missing white space after number '%.*s'", (int) (p
- str
), str
);
1280 return integer_zero_node
;
1284 lex_string (str
, len
, wide
)
1290 char *buf
= alloca ((len
+ 1) * (wide
? WCHAR_BYTES
: 1));
1292 const char *p
= str
, *limit
= str
+ len
;
1294 unsigned width
= wide
? WCHAR_TYPE_SIZE
1295 : TYPE_PRECISION (char_type_node
);
1297 #ifdef MULTIBYTE_CHARS
1298 /* Reset multibyte conversion state. */
1299 (void) local_mbtowc (NULL
, NULL
, 0);
1304 #ifdef MULTIBYTE_CHARS
1308 char_len
= local_mbtowc (&wc
, p
, limit
- p
);
1311 warning ("Ignoring invalid multibyte character");
1324 if (c
== '\\' && !ignore_escape_flag
)
1328 if (width
< HOST_BITS_PER_INT
)
1329 mask
= ((unsigned int) 1 << width
) - 1;
1332 c
= cpp_parse_escape (parse_in
, (const unsigned char **) &p
,
1333 (const unsigned char *) limit
,
1334 mask
, flag_traditional
);
1337 /* Add this single character into the buffer either as a wchar_t
1338 or as a single byte. */
1341 unsigned charwidth
= TYPE_PRECISION (char_type_node
);
1342 unsigned bytemask
= (1 << charwidth
) - 1;
1345 for (byte
= 0; byte
< WCHAR_BYTES
; ++byte
)
1348 if (byte
>= (int) sizeof (c
))
1351 n
= (c
>> (byte
* charwidth
)) & bytemask
;
1352 if (BYTES_BIG_ENDIAN
)
1353 q
[WCHAR_BYTES
- byte
- 1] = n
;
1365 /* Terminate the string value, either with a single byte zero
1366 or with a wide zero. */
1370 memset (q
, 0, WCHAR_BYTES
);
1378 value
= build_string (q
- buf
, buf
);
1381 TREE_TYPE (value
) = wchar_array_type_node
;
1383 TREE_TYPE (value
) = char_array_type_node
;
1387 /* Converts a (possibly wide) character constant token into a tree. */
1389 lex_charconst (token
)
1390 const cpp_token
*token
;
1392 HOST_WIDE_INT result
;
1394 unsigned int chars_seen
;
1396 result
= cpp_interpret_charconst (parse_in
, token
, warn_multichar
,
1397 flag_traditional
, &chars_seen
);
1398 if (token
->type
== CPP_WCHAR
)
1400 value
= build_int_2 (result
, 0);
1401 TREE_TYPE (value
) = wchar_type_node
;
1406 value
= build_int_2 (result
, -1);
1408 value
= build_int_2 (result
, 0);
1410 /* In C, a character constant has type 'int'.
1411 In C++ 'char', but multi-char charconsts have type 'int'. */
1412 if (c_language
== clk_cplusplus
&& chars_seen
<= 1)
1413 TREE_TYPE (value
) = char_type_node
;
1415 TREE_TYPE (value
) = integer_type_node
;