2 Copyright (C) 1986, 1987, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
3 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2007 Free Software Foundation, Inc.
4 Contributed by Per Bothner, 1994-95.
5 Based on CCCP program by Paul Rubin, June 1986
6 Adapted to ANSI C, Richard Stallman, Jan 1987
8 This program is free software; you can redistribute it and/or modify it
9 under the terms of the GNU General Public License as published by the
10 Free Software Foundation; either version 2, or (at your option) any
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
27 #include "localedir.h"
29 static void init_library (void);
30 static void mark_named_operators (cpp_reader
*);
31 static void read_original_filename (cpp_reader
*);
32 static void read_original_directory (cpp_reader
*);
33 static void post_options (cpp_reader
*);
35 /* If we have designated initializers (GCC >2.7) these tables can be
36 initialized, constant data. Otherwise, they have to be filled in at
38 #if HAVE_DESIGNATED_INITIALIZERS
40 #define init_trigraph_map() /* Nothing. */
41 #define TRIGRAPH_MAP \
42 __extension__ const uchar _cpp_trigraph_map[UCHAR_MAX + 1] = {
45 #define s(p, v) [p] = v,
49 #define TRIGRAPH_MAP uchar _cpp_trigraph_map[UCHAR_MAX + 1] = { 0 }; \
50 static void init_trigraph_map (void) { \
51 unsigned char *x = _cpp_trigraph_map;
54 #define s(p, v) x[p] = v;
59 s('=', '#') s(')', ']') s('!', '|')
60 s('(', '[') s('\'', '^') s('>', '}')
61 s('/', '\\') s('<', '{') s('-', '~')
68 /* A set of booleans indicating what CPP features each source language
74 char extended_numbers
;
75 char extended_identifiers
;
77 char cplusplus_comments
;
81 static const struct lang_flags lang_defaults
[] =
82 { /* c99 c++ xnum xid std // digr */
83 /* GNUC89 */ { 0, 0, 1, 0, 0, 1, 1 },
84 /* GNUC99 */ { 1, 0, 1, 0, 0, 1, 1 },
85 /* STDC89 */ { 0, 0, 0, 0, 1, 0, 0 },
86 /* STDC94 */ { 0, 0, 0, 0, 1, 0, 1 },
87 /* STDC99 */ { 1, 0, 1, 0, 1, 1, 1 },
88 /* GNUCXX */ { 0, 1, 1, 0, 0, 1, 1 },
89 /* CXX98 */ { 0, 1, 1, 0, 1, 1, 1 },
90 /* GNUCXX0X */ { 1, 1, 1, 0, 0, 1, 1 },
91 /* CXX0X */ { 1, 1, 1, 0, 1, 1, 1 },
92 /* ASM */ { 0, 0, 1, 0, 0, 1, 0 }
93 /* xid should be 1 for GNUC99, STDC99, GNUCXX, CXX98, GNUCXX0X, and
94 CXX0X when no longer experimental (when all uses of identifiers
95 in the compiler have been audited for correct handling of
96 extended identifiers). */
99 /* Sets internal flags correctly for a given language. */
101 cpp_set_lang (cpp_reader
*pfile
, enum c_lang lang
)
103 const struct lang_flags
*l
= &lang_defaults
[(int) lang
];
105 CPP_OPTION (pfile
, lang
) = lang
;
107 CPP_OPTION (pfile
, c99
) = l
->c99
;
108 CPP_OPTION (pfile
, cplusplus
) = l
->cplusplus
;
109 CPP_OPTION (pfile
, extended_numbers
) = l
->extended_numbers
;
110 CPP_OPTION (pfile
, extended_identifiers
) = l
->extended_identifiers
;
111 CPP_OPTION (pfile
, std
) = l
->std
;
112 CPP_OPTION (pfile
, trigraphs
) = l
->std
;
113 CPP_OPTION (pfile
, cplusplus_comments
) = l
->cplusplus_comments
;
114 CPP_OPTION (pfile
, digraphs
) = l
->digraphs
;
117 /* Initialize library global state. */
121 static int initialized
= 0;
127 /* Set up the trigraph map. This doesn't need to do anything if
128 we were compiled with a compiler that supports C99 designated
130 init_trigraph_map ();
133 (void) bindtextdomain (PACKAGE
, LOCALEDIR
);
138 /* Initialize a cpp_reader structure. */
140 cpp_create_reader (enum c_lang lang
, hash_table
*table
,
141 struct line_maps
*line_table
)
145 /* Initialize this instance of the library if it hasn't been already. */
148 pfile
= XCNEW (cpp_reader
);
150 cpp_set_lang (pfile
, lang
);
151 CPP_OPTION (pfile
, warn_multichar
) = 1;
152 CPP_OPTION (pfile
, discard_comments
) = 1;
153 CPP_OPTION (pfile
, discard_comments_in_macro_exp
) = 1;
154 CPP_OPTION (pfile
, show_column
) = 1;
155 CPP_OPTION (pfile
, tabstop
) = 8;
156 CPP_OPTION (pfile
, operator_names
) = 1;
157 CPP_OPTION (pfile
, warn_trigraphs
) = 2;
158 CPP_OPTION (pfile
, warn_endif_labels
) = 1;
159 CPP_OPTION (pfile
, warn_deprecated
) = 1;
160 CPP_OPTION (pfile
, warn_long_long
) = !CPP_OPTION (pfile
, c99
);
161 CPP_OPTION (pfile
, dollars_in_ident
) = 1;
162 CPP_OPTION (pfile
, warn_dollars
) = 1;
163 CPP_OPTION (pfile
, warn_variadic_macros
) = 1;
164 CPP_OPTION (pfile
, warn_normalize
) = normalized_C
;
166 /* Default CPP arithmetic to something sensible for the host for the
167 benefit of dumb users like fix-header. */
168 CPP_OPTION (pfile
, precision
) = CHAR_BIT
* sizeof (long);
169 CPP_OPTION (pfile
, char_precision
) = CHAR_BIT
;
170 CPP_OPTION (pfile
, wchar_precision
) = CHAR_BIT
* sizeof (int);
171 CPP_OPTION (pfile
, int_precision
) = CHAR_BIT
* sizeof (int);
172 CPP_OPTION (pfile
, unsigned_char
) = 0;
173 CPP_OPTION (pfile
, unsigned_wchar
) = 1;
174 CPP_OPTION (pfile
, bytes_big_endian
) = 1; /* does not matter */
176 /* Default to no charset conversion. */
177 CPP_OPTION (pfile
, narrow_charset
) = _cpp_default_encoding ();
178 CPP_OPTION (pfile
, wide_charset
) = 0;
180 /* Default the input character set to UTF-8. */
181 CPP_OPTION (pfile
, input_charset
) = _cpp_default_encoding ();
183 /* A fake empty "directory" used as the starting point for files
184 looked up without a search path. Name cannot be '/' because we
185 don't want to prepend anything at all to filenames using it. All
186 other entries are correct zero-initialized. */
187 pfile
->no_search_path
.name
= (char *) "";
189 /* Initialize the line map. */
190 pfile
->line_table
= line_table
;
192 /* Initialize lexer state. */
193 pfile
->state
.save_comments
= ! CPP_OPTION (pfile
, discard_comments
);
195 /* Set up static tokens. */
196 pfile
->avoid_paste
.type
= CPP_PADDING
;
197 pfile
->avoid_paste
.val
.source
= NULL
;
198 pfile
->eof
.type
= CPP_EOF
;
199 pfile
->eof
.flags
= 0;
201 /* Create a token buffer for the lexer. */
202 _cpp_init_tokenrun (&pfile
->base_run
, 250);
203 pfile
->cur_run
= &pfile
->base_run
;
204 pfile
->cur_token
= pfile
->base_run
.base
;
206 /* Initialize the base context. */
207 pfile
->context
= &pfile
->base_context
;
208 pfile
->base_context
.macro
= 0;
209 pfile
->base_context
.prev
= pfile
->base_context
.next
= 0;
211 /* Aligned and unaligned storage. */
212 pfile
->a_buff
= _cpp_get_buff (pfile
, 0);
213 pfile
->u_buff
= _cpp_get_buff (pfile
, 0);
215 /* The expression parser stack. */
216 _cpp_expand_op_stack (pfile
);
218 /* Initialize the buffer obstack. */
219 _obstack_begin (&pfile
->buffer_ob
, 0, 0,
220 (void *(*) (long)) xmalloc
,
221 (void (*) (void *)) free
);
223 _cpp_init_files (pfile
);
225 _cpp_init_hashtable (pfile
, table
);
230 /* Set the line_table entry in PFILE. This is called after reading a
231 PCH file, as the old line_table will be incorrect. */
233 cpp_set_line_map (cpp_reader
*pfile
, struct line_maps
*line_table
)
235 pfile
->line_table
= line_table
;
238 /* Free resources used by PFILE. Accessing PFILE after this function
239 returns leads to undefined behavior. Returns the error count. */
241 cpp_destroy (cpp_reader
*pfile
)
243 cpp_context
*context
, *contextn
;
244 tokenrun
*run
, *runn
;
246 free (pfile
->op_stack
);
248 while (CPP_BUFFER (pfile
) != NULL
)
249 _cpp_pop_buffer (pfile
);
252 free (pfile
->out
.base
);
254 if (pfile
->macro_buffer
)
256 free (pfile
->macro_buffer
);
257 pfile
->macro_buffer
= NULL
;
258 pfile
->macro_buffer_len
= 0;
262 deps_free (pfile
->deps
);
263 obstack_free (&pfile
->buffer_ob
, 0);
265 _cpp_destroy_hashtable (pfile
);
266 _cpp_cleanup_files (pfile
);
267 _cpp_destroy_iconv (pfile
);
269 _cpp_free_buff (pfile
->a_buff
);
270 _cpp_free_buff (pfile
->u_buff
);
271 _cpp_free_buff (pfile
->free_buffs
);
273 for (run
= &pfile
->base_run
; run
; run
= runn
)
277 if (run
!= &pfile
->base_run
)
281 for (context
= pfile
->base_context
.next
; context
; context
= contextn
)
283 contextn
= context
->next
;
290 /* This structure defines one built-in identifier. A node will be
291 entered in the hash table under the name NAME, with value VALUE.
293 There are two tables of these. builtin_array holds all the
294 "builtin" macros: these are handled by builtin_macro() in
295 macro.c. Builtin is somewhat of a misnomer -- the property of
296 interest is that these macros require special code to compute their
297 expansions. The value is a "builtin_type" enumerator.
299 operator_array holds the C++ named operators. These are keywords
300 which act as aliases for punctuators. In C++, they cannot be
301 altered through #define, and #if recognizes them as operators. In
302 C, these are not entered into the hash table at all (but see
303 <iso646.h>). The value is a token-type enumerator. */
308 unsigned short value
;
311 #define B(n, t) { DSC(n), t }
312 static const struct builtin builtin_array
[] =
314 B("__TIMESTAMP__", BT_TIMESTAMP
),
315 B("__TIME__", BT_TIME
),
316 B("__DATE__", BT_DATE
),
317 B("__FILE__", BT_FILE
),
318 B("__BASE_FILE__", BT_BASE_FILE
),
319 B("__LINE__", BT_SPECLINE
),
320 B("__INCLUDE_LEVEL__", BT_INCLUDE_LEVEL
),
321 B("__COUNTER__", BT_COUNTER
),
322 /* Keep builtins not used for -traditional-cpp at the end, and
323 update init_builtins() if any more are added. */
324 B("_Pragma", BT_PRAGMA
),
325 B("__STDC__", BT_STDC
),
328 static const struct builtin operator_array
[] =
330 B("and", CPP_AND_AND
),
331 B("and_eq", CPP_AND_EQ
),
332 B("bitand", CPP_AND
),
334 B("compl", CPP_COMPL
),
336 B("not_eq", CPP_NOT_EQ
),
338 B("or_eq", CPP_OR_EQ
),
340 B("xor_eq", CPP_XOR_EQ
)
344 /* Mark the C++ named operators in the hash table. */
346 mark_named_operators (cpp_reader
*pfile
)
348 const struct builtin
*b
;
350 for (b
= operator_array
;
351 b
< (operator_array
+ ARRAY_SIZE (operator_array
));
354 cpp_hashnode
*hp
= cpp_lookup (pfile
, b
->name
, b
->len
);
355 hp
->flags
|= NODE_OPERATOR
;
356 hp
->is_directive
= 0;
357 hp
->directive_index
= b
->value
;
362 cpp_init_special_builtins (cpp_reader
*pfile
)
364 const struct builtin
*b
;
365 size_t n
= ARRAY_SIZE (builtin_array
);
367 if (CPP_OPTION (pfile
, traditional
))
369 else if (! CPP_OPTION (pfile
, stdc_0_in_system_headers
)
370 || CPP_OPTION (pfile
, std
))
373 for (b
= builtin_array
; b
< builtin_array
+ n
; b
++)
375 cpp_hashnode
*hp
= cpp_lookup (pfile
, b
->name
, b
->len
);
377 hp
->flags
|= NODE_BUILTIN
| NODE_WARN
;
378 hp
->value
.builtin
= (enum builtin_type
) b
->value
;
382 /* Read the builtins table above and enter them, and language-specific
383 macros, into the hash table. HOSTED is true if this is a hosted
386 cpp_init_builtins (cpp_reader
*pfile
, int hosted
)
388 cpp_init_special_builtins (pfile
);
390 if (!CPP_OPTION (pfile
, traditional
)
391 && (! CPP_OPTION (pfile
, stdc_0_in_system_headers
)
392 || CPP_OPTION (pfile
, std
)))
393 _cpp_define_builtin (pfile
, "__STDC__ 1");
395 if (CPP_OPTION (pfile
, cplusplus
))
396 _cpp_define_builtin (pfile
, "__cplusplus 1");
397 else if (CPP_OPTION (pfile
, lang
) == CLK_ASM
)
398 _cpp_define_builtin (pfile
, "__ASSEMBLER__ 1");
399 else if (CPP_OPTION (pfile
, lang
) == CLK_STDC94
)
400 _cpp_define_builtin (pfile
, "__STDC_VERSION__ 199409L");
401 else if (CPP_OPTION (pfile
, c99
))
402 _cpp_define_builtin (pfile
, "__STDC_VERSION__ 199901L");
405 _cpp_define_builtin (pfile
, "__STDC_HOSTED__ 1");
407 _cpp_define_builtin (pfile
, "__STDC_HOSTED__ 0");
409 if (CPP_OPTION (pfile
, objc
))
410 _cpp_define_builtin (pfile
, "__OBJC__ 1");
413 /* Sanity-checks are dependent on command-line options, so it is
414 called as a subroutine of cpp_read_main_file (). */
416 static void sanity_checks (cpp_reader
*);
417 static void sanity_checks (cpp_reader
*pfile
)
420 size_t max_precision
= 2 * CHAR_BIT
* sizeof (cpp_num_part
);
422 /* Sanity checks for assumptions about CPP arithmetic and target
423 type precisions made by cpplib. */
426 cpp_error (pfile
, CPP_DL_ICE
, "cppchar_t must be an unsigned type");
428 if (CPP_OPTION (pfile
, precision
) > max_precision
)
429 cpp_error (pfile
, CPP_DL_ICE
,
430 "preprocessor arithmetic has maximum precision of %lu bits;"
431 " target requires %lu bits",
432 (unsigned long) max_precision
,
433 (unsigned long) CPP_OPTION (pfile
, precision
));
435 if (CPP_OPTION (pfile
, precision
) < CPP_OPTION (pfile
, int_precision
))
436 cpp_error (pfile
, CPP_DL_ICE
,
437 "CPP arithmetic must be at least as precise as a target int");
439 if (CPP_OPTION (pfile
, char_precision
) < 8)
440 cpp_error (pfile
, CPP_DL_ICE
, "target char is less than 8 bits wide");
442 if (CPP_OPTION (pfile
, wchar_precision
) < CPP_OPTION (pfile
, char_precision
))
443 cpp_error (pfile
, CPP_DL_ICE
,
444 "target wchar_t is narrower than target char");
446 if (CPP_OPTION (pfile
, int_precision
) < CPP_OPTION (pfile
, char_precision
))
447 cpp_error (pfile
, CPP_DL_ICE
,
448 "target int is narrower than target char");
450 /* This is assumed in eval_token() and could be fixed if necessary. */
451 if (sizeof (cppchar_t
) > sizeof (cpp_num_part
))
452 cpp_error (pfile
, CPP_DL_ICE
,
453 "CPP half-integer narrower than CPP character");
455 if (CPP_OPTION (pfile
, wchar_precision
) > BITS_PER_CPPCHAR_T
)
456 cpp_error (pfile
, CPP_DL_ICE
,
457 "CPP on this host cannot handle wide character constants over"
458 " %lu bits, but the target requires %lu bits",
459 (unsigned long) BITS_PER_CPPCHAR_T
,
460 (unsigned long) CPP_OPTION (pfile
, wchar_precision
));
463 # define sanity_checks(PFILE)
466 /* This is called after options have been parsed, and partially
469 cpp_post_options (cpp_reader
*pfile
)
471 sanity_checks (pfile
);
473 post_options (pfile
);
475 /* Mark named operators before handling command line macros. */
476 if (CPP_OPTION (pfile
, cplusplus
) && CPP_OPTION (pfile
, operator_names
))
477 mark_named_operators (pfile
);
480 /* Setup for processing input from the file named FNAME, or stdin if
481 it is the empty string. Return the original filename
482 on success (e.g. foo.i->foo.c), or NULL on failure. */
484 cpp_read_main_file (cpp_reader
*pfile
, const char *fname
)
486 if (CPP_OPTION (pfile
, deps
.style
) != DEPS_NONE
)
489 pfile
->deps
= deps_init ();
491 /* Set the default target (if there is none already). */
492 deps_add_default_target (pfile
->deps
, fname
);
496 = _cpp_find_file (pfile
, fname
, &pfile
->no_search_path
, false, 0);
497 if (_cpp_find_failed (pfile
->main_file
))
500 _cpp_stack_file (pfile
, pfile
->main_file
, false);
502 /* For foo.i, read the original filename foo.c now, for the benefit
503 of the front ends. */
504 if (CPP_OPTION (pfile
, preprocessed
))
506 read_original_filename (pfile
);
507 fname
= pfile
->line_table
->maps
[pfile
->line_table
->used
-1].to_file
;
512 /* For preprocessed files, if the first tokens are of the form # NUM.
513 handle the directive so we know the original file name. This will
514 generate file_change callbacks, which the front ends must handle
515 appropriately given their state of initialization. */
517 read_original_filename (cpp_reader
*pfile
)
519 const cpp_token
*token
, *token1
;
521 /* Lex ahead; if the first tokens are of the form # NUM, then
522 process the directive, otherwise back up. */
523 token
= _cpp_lex_direct (pfile
);
524 if (token
->type
== CPP_HASH
)
526 pfile
->state
.in_directive
= 1;
527 token1
= _cpp_lex_direct (pfile
);
528 _cpp_backup_tokens (pfile
, 1);
529 pfile
->state
.in_directive
= 0;
531 /* If it's a #line directive, handle it. */
532 if (token1
->type
== CPP_NUMBER
)
534 _cpp_handle_directive (pfile
, token
->flags
& PREV_WHITE
);
535 read_original_directory (pfile
);
540 /* Backup as if nothing happened. */
541 _cpp_backup_tokens (pfile
, 1);
544 /* For preprocessed files, if the tokens following the first filename
545 line is of the form # <line> "/path/name//", handle the
546 directive so we know the original current directory. */
548 read_original_directory (cpp_reader
*pfile
)
550 const cpp_token
*hash
, *token
;
552 /* Lex ahead; if the first tokens are of the form # NUM, then
553 process the directive, otherwise back up. */
554 hash
= _cpp_lex_direct (pfile
);
555 if (hash
->type
!= CPP_HASH
)
557 _cpp_backup_tokens (pfile
, 1);
561 token
= _cpp_lex_direct (pfile
);
563 if (token
->type
!= CPP_NUMBER
)
565 _cpp_backup_tokens (pfile
, 2);
569 token
= _cpp_lex_direct (pfile
);
571 if (token
->type
!= CPP_STRING
572 || ! (token
->val
.str
.len
>= 5
573 && token
->val
.str
.text
[token
->val
.str
.len
-2] == '/'
574 && token
->val
.str
.text
[token
->val
.str
.len
-3] == '/'))
576 _cpp_backup_tokens (pfile
, 3);
580 if (pfile
->cb
.dir_change
)
582 char *debugdir
= (char *) alloca (token
->val
.str
.len
- 3);
584 memcpy (debugdir
, (const char *) token
->val
.str
.text
+ 1,
585 token
->val
.str
.len
- 4);
586 debugdir
[token
->val
.str
.len
- 4] = '\0';
588 pfile
->cb
.dir_change (pfile
, debugdir
);
592 /* This is called at the end of preprocessing. It pops the last
593 buffer and writes dependency output, and returns the number of
596 Maybe it should also reset state, such that you could call
597 cpp_start_read with a new filename to restart processing. */
599 cpp_finish (cpp_reader
*pfile
, FILE *deps_stream
)
601 /* Warn about unused macros before popping the final buffer. */
602 if (CPP_OPTION (pfile
, warn_unused_macros
))
603 cpp_forall_identifiers (pfile
, _cpp_warn_if_unused_macro
, NULL
);
605 /* lex.c leaves the final buffer on the stack. This it so that
606 it returns an unending stream of CPP_EOFs to the client. If we
607 popped the buffer, we'd dereference a NULL buffer pointer and
608 segfault. It's nice to allow the client to do worry-free excess
609 cpp_get_token calls. */
610 while (pfile
->buffer
)
611 _cpp_pop_buffer (pfile
);
613 /* Don't write the deps file if there are errors. */
614 if (CPP_OPTION (pfile
, deps
.style
) != DEPS_NONE
615 && deps_stream
&& pfile
->errors
== 0)
617 deps_write (pfile
->deps
, deps_stream
, 72);
619 if (CPP_OPTION (pfile
, deps
.phony_targets
))
620 deps_phony_targets (pfile
->deps
, deps_stream
);
623 /* Report on headers that could use multiple include guards. */
624 if (CPP_OPTION (pfile
, print_include_names
))
625 _cpp_report_missing_guards (pfile
);
627 return pfile
->errors
;
631 post_options (cpp_reader
*pfile
)
633 /* -Wtraditional is not useful in C++ mode. */
634 if (CPP_OPTION (pfile
, cplusplus
))
635 CPP_OPTION (pfile
, warn_traditional
) = 0;
637 /* Permanently disable macro expansion if we are rescanning
638 preprocessed text. Read preprocesed source in ISO mode. */
639 if (CPP_OPTION (pfile
, preprocessed
))
641 if (!CPP_OPTION (pfile
, directives_only
))
642 pfile
->state
.prevent_expansion
= 1;
643 CPP_OPTION (pfile
, traditional
) = 0;
646 if (CPP_OPTION (pfile
, warn_trigraphs
) == 2)
647 CPP_OPTION (pfile
, warn_trigraphs
) = !CPP_OPTION (pfile
, trigraphs
);
649 if (CPP_OPTION (pfile
, traditional
))
651 CPP_OPTION (pfile
, cplusplus_comments
) = 0;
653 /* Traditional CPP does not accurately track column information. */
654 CPP_OPTION (pfile
, show_column
) = 0;
655 CPP_OPTION (pfile
, trigraphs
) = 0;
656 CPP_OPTION (pfile
, warn_trigraphs
) = 0;