2 Copyright (C) 1986, 1987, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
3 1999, 2000, 2001, 2002, 2003, 2004 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, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
28 static void init_library (void);
29 static void mark_named_operators (cpp_reader
*);
30 static void read_original_filename (cpp_reader
*);
31 static void read_original_directory (cpp_reader
*);
32 static void post_options (cpp_reader
*);
34 /* If we have designated initializers (GCC >2.7) these tables can be
35 initialized, constant data. Otherwise, they have to be filled in at
37 #if HAVE_DESIGNATED_INITIALIZERS
39 #define init_trigraph_map() /* Nothing. */
40 #define TRIGRAPH_MAP \
41 __extension__ const uchar _cpp_trigraph_map[UCHAR_MAX + 1] = {
44 #define s(p, v) [p] = v,
48 #define TRIGRAPH_MAP uchar _cpp_trigraph_map[UCHAR_MAX + 1] = { 0 }; \
49 static void init_trigraph_map (void) { \
50 unsigned char *x = _cpp_trigraph_map;
53 #define s(p, v) x[p] = v;
58 s('=', '#') s(')', ']') s('!', '|')
59 s('(', '[') s('\'', '^') s('>', '}')
60 s('/', '\\') s('<', '{') s('-', '~')
67 /* A set of booleans indicating what CPP features each source language
73 char extended_numbers
;
75 char cplusplus_comments
;
79 static const struct lang_flags lang_defaults
[] =
80 { /* c99 c++ xnum std // digr */
81 /* GNUC89 */ { 0, 0, 1, 0, 1, 1 },
82 /* GNUC99 */ { 1, 0, 1, 0, 1, 1 },
83 /* STDC89 */ { 0, 0, 0, 1, 0, 0 },
84 /* STDC94 */ { 0, 0, 0, 1, 0, 1 },
85 /* STDC99 */ { 1, 0, 1, 1, 1, 1 },
86 /* GNUCXX */ { 0, 1, 1, 0, 1, 1 },
87 /* CXX98 */ { 0, 1, 1, 1, 1, 1 },
88 /* ASM */ { 0, 0, 1, 0, 1, 0 }
91 /* Sets internal flags correctly for a given language. */
93 cpp_set_lang (cpp_reader
*pfile
, enum c_lang lang
)
95 const struct lang_flags
*l
= &lang_defaults
[(int) lang
];
97 CPP_OPTION (pfile
, lang
) = lang
;
99 CPP_OPTION (pfile
, c99
) = l
->c99
;
100 CPP_OPTION (pfile
, cplusplus
) = l
->cplusplus
;
101 CPP_OPTION (pfile
, extended_numbers
) = l
->extended_numbers
;
102 CPP_OPTION (pfile
, std
) = l
->std
;
103 CPP_OPTION (pfile
, trigraphs
) = l
->std
;
104 CPP_OPTION (pfile
, cplusplus_comments
) = l
->cplusplus_comments
;
105 CPP_OPTION (pfile
, digraphs
) = l
->digraphs
;
108 /* Initialize library global state. */
112 static int initialized
= 0;
118 /* Set up the trigraph map. This doesn't need to do anything if
119 we were compiled with a compiler that supports C99 designated
121 init_trigraph_map ();
125 /* Initialize a cpp_reader structure. */
127 cpp_create_reader (enum c_lang lang
, hash_table
*table
,
128 struct line_maps
*line_table
)
132 /* Initialize this instance of the library if it hasn't been already. */
135 pfile
= xcalloc (1, sizeof (cpp_reader
));
137 cpp_set_lang (pfile
, lang
);
138 CPP_OPTION (pfile
, warn_multichar
) = 1;
139 CPP_OPTION (pfile
, discard_comments
) = 1;
140 CPP_OPTION (pfile
, discard_comments_in_macro_exp
) = 1;
141 CPP_OPTION (pfile
, show_column
) = 1;
142 CPP_OPTION (pfile
, tabstop
) = 8;
143 CPP_OPTION (pfile
, operator_names
) = 1;
144 CPP_OPTION (pfile
, warn_trigraphs
) = 2;
145 CPP_OPTION (pfile
, warn_endif_labels
) = 1;
146 CPP_OPTION (pfile
, warn_deprecated
) = 1;
147 CPP_OPTION (pfile
, warn_long_long
) = !CPP_OPTION (pfile
, c99
);
148 CPP_OPTION (pfile
, dollars_in_ident
) = 1;
149 CPP_OPTION (pfile
, warn_dollars
) = 1;
150 CPP_OPTION (pfile
, warn_variadic_macros
) = 1;
152 /* Default CPP arithmetic to something sensible for the host for the
153 benefit of dumb users like fix-header. */
154 CPP_OPTION (pfile
, precision
) = CHAR_BIT
* sizeof (long);
155 CPP_OPTION (pfile
, char_precision
) = CHAR_BIT
;
156 CPP_OPTION (pfile
, wchar_precision
) = CHAR_BIT
* sizeof (int);
157 CPP_OPTION (pfile
, int_precision
) = CHAR_BIT
* sizeof (int);
158 CPP_OPTION (pfile
, unsigned_char
) = 0;
159 CPP_OPTION (pfile
, unsigned_wchar
) = 1;
160 CPP_OPTION (pfile
, bytes_big_endian
) = 1; /* does not matter */
162 /* Default to no charset conversion. */
163 CPP_OPTION (pfile
, narrow_charset
) = _cpp_default_encoding ();
164 CPP_OPTION (pfile
, wide_charset
) = 0;
166 /* Default the input character set to UTF-8. */
167 CPP_OPTION (pfile
, input_charset
) = _cpp_default_encoding ();
169 /* A fake empty "directory" used as the starting point for files
170 looked up without a search path. Name cannot be '/' because we
171 don't want to prepend anything at all to filenames using it. All
172 other entries are correct zero-initialized. */
173 pfile
->no_search_path
.name
= (char *) "";
175 /* Initialize the line map. */
176 pfile
->line_table
= line_table
;
178 /* Initialize lexer state. */
179 pfile
->state
.save_comments
= ! CPP_OPTION (pfile
, discard_comments
);
181 /* Set up static tokens. */
182 pfile
->avoid_paste
.type
= CPP_PADDING
;
183 pfile
->avoid_paste
.val
.source
= NULL
;
184 pfile
->eof
.type
= CPP_EOF
;
185 pfile
->eof
.flags
= 0;
187 /* Create a token buffer for the lexer. */
188 _cpp_init_tokenrun (&pfile
->base_run
, 250);
189 pfile
->cur_run
= &pfile
->base_run
;
190 pfile
->cur_token
= pfile
->base_run
.base
;
192 /* Initialize the base context. */
193 pfile
->context
= &pfile
->base_context
;
194 pfile
->base_context
.macro
= 0;
195 pfile
->base_context
.prev
= pfile
->base_context
.next
= 0;
197 /* Aligned and unaligned storage. */
198 pfile
->a_buff
= _cpp_get_buff (pfile
, 0);
199 pfile
->u_buff
= _cpp_get_buff (pfile
, 0);
201 /* The expression parser stack. */
202 _cpp_expand_op_stack (pfile
);
204 /* Initialize the buffer obstack. */
205 _obstack_begin (&pfile
->buffer_ob
, 0, 0,
206 (void *(*) (long)) xmalloc
,
207 (void (*) (void *)) free
);
209 _cpp_init_files (pfile
);
211 _cpp_init_hashtable (pfile
, table
);
216 /* Free resources used by PFILE. Accessing PFILE after this function
217 returns leads to undefined behavior. Returns the error count. */
219 cpp_destroy (cpp_reader
*pfile
)
221 cpp_context
*context
, *contextn
;
222 tokenrun
*run
, *runn
;
224 free (pfile
->op_stack
);
226 while (CPP_BUFFER (pfile
) != NULL
)
227 _cpp_pop_buffer (pfile
);
230 free (pfile
->out
.base
);
232 if (pfile
->macro_buffer
)
234 free (pfile
->macro_buffer
);
235 pfile
->macro_buffer
= NULL
;
236 pfile
->macro_buffer_len
= 0;
240 deps_free (pfile
->deps
);
241 obstack_free (&pfile
->buffer_ob
, 0);
243 _cpp_destroy_hashtable (pfile
);
244 _cpp_cleanup_files (pfile
);
245 _cpp_destroy_iconv (pfile
);
247 _cpp_free_buff (pfile
->a_buff
);
248 _cpp_free_buff (pfile
->u_buff
);
249 _cpp_free_buff (pfile
->free_buffs
);
251 for (run
= &pfile
->base_run
; run
; run
= runn
)
255 if (run
!= &pfile
->base_run
)
259 for (context
= pfile
->base_context
.next
; context
; context
= contextn
)
261 contextn
= context
->next
;
268 /* This structure defines one built-in identifier. A node will be
269 entered in the hash table under the name NAME, with value VALUE.
271 There are two tables of these. builtin_array holds all the
272 "builtin" macros: these are handled by builtin_macro() in
273 cppmacro.c. Builtin is somewhat of a misnomer -- the property of
274 interest is that these macros require special code to compute their
275 expansions. The value is a "builtin_type" enumerator.
277 operator_array holds the C++ named operators. These are keywords
278 which act as aliases for punctuators. In C++, they cannot be
279 altered through #define, and #if recognizes them as operators. In
280 C, these are not entered into the hash table at all (but see
281 <iso646.h>). The value is a token-type enumerator. */
286 unsigned short value
;
289 #define B(n, t) { DSC(n), t }
290 static const struct builtin builtin_array
[] =
292 B("__TIME__", BT_TIME
),
293 B("__DATE__", BT_DATE
),
294 B("__FILE__", BT_FILE
),
295 B("__BASE_FILE__", BT_BASE_FILE
),
296 B("__LINE__", BT_SPECLINE
),
297 B("__INCLUDE_LEVEL__", BT_INCLUDE_LEVEL
),
298 /* Keep builtins not used for -traditional-cpp at the end, and
299 update init_builtins() if any more are added. */
300 B("_Pragma", BT_PRAGMA
),
301 B("__STDC__", BT_STDC
),
304 static const struct builtin operator_array
[] =
306 B("and", CPP_AND_AND
),
307 B("and_eq", CPP_AND_EQ
),
308 B("bitand", CPP_AND
),
310 B("compl", CPP_COMPL
),
312 B("not_eq", CPP_NOT_EQ
),
314 B("or_eq", CPP_OR_EQ
),
316 B("xor_eq", CPP_XOR_EQ
)
320 /* Mark the C++ named operators in the hash table. */
322 mark_named_operators (cpp_reader
*pfile
)
324 const struct builtin
*b
;
326 for (b
= operator_array
;
327 b
< (operator_array
+ ARRAY_SIZE (operator_array
));
330 cpp_hashnode
*hp
= cpp_lookup (pfile
, b
->name
, b
->len
);
331 hp
->flags
|= NODE_OPERATOR
;
332 hp
->is_directive
= 0;
333 hp
->directive_index
= b
->value
;
337 /* Read the builtins table above and enter them, and language-specific
338 macros, into the hash table. HOSTED is true if this is a hosted
341 cpp_init_builtins (cpp_reader
*pfile
, int hosted
)
343 const struct builtin
*b
;
344 size_t n
= ARRAY_SIZE (builtin_array
);
346 if (CPP_OPTION (pfile
, traditional
))
349 for(b
= builtin_array
; b
< builtin_array
+ n
; b
++)
351 cpp_hashnode
*hp
= cpp_lookup (pfile
, b
->name
, b
->len
);
353 hp
->flags
|= NODE_BUILTIN
| NODE_WARN
;
354 hp
->value
.builtin
= b
->value
;
357 if (CPP_OPTION (pfile
, cplusplus
))
358 _cpp_define_builtin (pfile
, "__cplusplus 1");
359 else if (CPP_OPTION (pfile
, lang
) == CLK_ASM
)
360 _cpp_define_builtin (pfile
, "__ASSEMBLER__ 1");
361 else if (CPP_OPTION (pfile
, lang
) == CLK_STDC94
)
362 _cpp_define_builtin (pfile
, "__STDC_VERSION__ 199409L");
363 else if (CPP_OPTION (pfile
, c99
))
364 _cpp_define_builtin (pfile
, "__STDC_VERSION__ 199901L");
367 _cpp_define_builtin (pfile
, "__STDC_HOSTED__ 1");
369 _cpp_define_builtin (pfile
, "__STDC_HOSTED__ 0");
371 if (CPP_OPTION (pfile
, objc
))
372 _cpp_define_builtin (pfile
, "__OBJC__ 1");
375 /* Sanity-checks are dependent on command-line options, so it is
376 called as a subroutine of cpp_read_main_file (). */
378 static void sanity_checks (cpp_reader
*);
379 static void sanity_checks (cpp_reader
*pfile
)
382 size_t max_precision
= 2 * CHAR_BIT
* sizeof (cpp_num_part
);
384 /* Sanity checks for assumptions about CPP arithmetic and target
385 type precisions made by cpplib. */
388 cpp_error (pfile
, CPP_DL_ICE
, "cppchar_t must be an unsigned type");
390 if (CPP_OPTION (pfile
, precision
) > max_precision
)
391 cpp_error (pfile
, CPP_DL_ICE
,
392 "preprocessor arithmetic has maximum precision of %lu bits;"
393 " target requires %lu bits",
394 (unsigned long) max_precision
,
395 (unsigned long) CPP_OPTION (pfile
, precision
));
397 if (CPP_OPTION (pfile
, precision
) < CPP_OPTION (pfile
, int_precision
))
398 cpp_error (pfile
, CPP_DL_ICE
,
399 "CPP arithmetic must be at least as precise as a target int");
401 if (CPP_OPTION (pfile
, char_precision
) < 8)
402 cpp_error (pfile
, CPP_DL_ICE
, "target char is less than 8 bits wide");
404 if (CPP_OPTION (pfile
, wchar_precision
) < CPP_OPTION (pfile
, char_precision
))
405 cpp_error (pfile
, CPP_DL_ICE
,
406 "target wchar_t is narrower than target char");
408 if (CPP_OPTION (pfile
, int_precision
) < CPP_OPTION (pfile
, char_precision
))
409 cpp_error (pfile
, CPP_DL_ICE
,
410 "target int is narrower than target char");
412 /* This is assumed in eval_token() and could be fixed if necessary. */
413 if (sizeof (cppchar_t
) > sizeof (cpp_num_part
))
414 cpp_error (pfile
, CPP_DL_ICE
,
415 "CPP half-integer narrower than CPP character");
417 if (CPP_OPTION (pfile
, wchar_precision
) > BITS_PER_CPPCHAR_T
)
418 cpp_error (pfile
, CPP_DL_ICE
,
419 "CPP on this host cannot handle wide character constants over"
420 " %lu bits, but the target requires %lu bits",
421 (unsigned long) BITS_PER_CPPCHAR_T
,
422 (unsigned long) CPP_OPTION (pfile
, wchar_precision
));
425 # define sanity_checks(PFILE)
428 /* Add a dependency target. Can be called any number of times before
429 cpp_read_main_file(). If no targets have been added before
430 cpp_read_main_file(), then the default target is used. */
432 cpp_add_dependency_target (cpp_reader
*pfile
, const char *target
, int quote
)
435 pfile
->deps
= deps_init ();
437 deps_add_target (pfile
->deps
, target
, quote
);
440 /* This is called after options have been parsed, and partially
443 cpp_post_options (cpp_reader
*pfile
)
445 sanity_checks (pfile
);
447 post_options (pfile
);
449 /* Mark named operators before handling command line macros. */
450 if (CPP_OPTION (pfile
, cplusplus
) && CPP_OPTION (pfile
, operator_names
))
451 mark_named_operators (pfile
);
454 /* Setup for processing input from the file named FNAME, or stdin if
455 it is the empty string. Return the original filename
456 on success (e.g. foo.i->foo.c), or NULL on failure. */
458 cpp_read_main_file (cpp_reader
*pfile
, const char *fname
)
460 if (CPP_OPTION (pfile
, deps
.style
) != DEPS_NONE
)
463 pfile
->deps
= deps_init ();
465 /* Set the default target (if there is none already). */
466 deps_add_default_target (pfile
->deps
, fname
);
470 = _cpp_find_file (pfile
, fname
, &pfile
->no_search_path
, false);
471 if (_cpp_find_failed (pfile
->main_file
))
474 _cpp_stack_file (pfile
, pfile
->main_file
, false);
476 /* For foo.i, read the original filename foo.c now, for the benefit
477 of the front ends. */
478 if (CPP_OPTION (pfile
, preprocessed
))
480 read_original_filename (pfile
);
481 fname
= pfile
->line_table
->maps
[pfile
->line_table
->used
-1].to_file
;
486 /* For preprocessed files, if the first tokens are of the form # NUM.
487 handle the directive so we know the original file name. This will
488 generate file_change callbacks, which the front ends must handle
489 appropriately given their state of initialization. */
491 read_original_filename (cpp_reader
*pfile
)
493 const cpp_token
*token
, *token1
;
495 /* Lex ahead; if the first tokens are of the form # NUM, then
496 process the directive, otherwise back up. */
497 token
= _cpp_lex_direct (pfile
);
498 if (token
->type
== CPP_HASH
)
500 token1
= _cpp_lex_direct (pfile
);
501 _cpp_backup_tokens (pfile
, 1);
503 /* If it's a #line directive, handle it. */
504 if (token1
->type
== CPP_NUMBER
)
506 _cpp_handle_directive (pfile
, token
->flags
& PREV_WHITE
);
507 read_original_directory (pfile
);
512 /* Backup as if nothing happened. */
513 _cpp_backup_tokens (pfile
, 1);
516 /* For preprocessed files, if the tokens following the first filename
517 line is of the form # <line> "/path/name//", handle the
518 directive so we know the original current directory. */
520 read_original_directory (cpp_reader
*pfile
)
522 const cpp_token
*hash
, *token
;
524 /* Lex ahead; if the first tokens are of the form # NUM, then
525 process the directive, otherwise back up. */
526 hash
= _cpp_lex_direct (pfile
);
527 if (hash
->type
!= CPP_HASH
)
529 _cpp_backup_tokens (pfile
, 1);
533 token
= _cpp_lex_direct (pfile
);
535 if (token
->type
!= CPP_NUMBER
)
537 _cpp_backup_tokens (pfile
, 2);
541 token
= _cpp_lex_direct (pfile
);
543 if (token
->type
!= CPP_STRING
544 || ! (token
->val
.str
.len
>= 5
545 && token
->val
.str
.text
[token
->val
.str
.len
-2] == '/'
546 && token
->val
.str
.text
[token
->val
.str
.len
-3] == '/'))
548 _cpp_backup_tokens (pfile
, 3);
552 if (pfile
->cb
.dir_change
)
554 char *debugdir
= alloca (token
->val
.str
.len
- 3);
556 memcpy (debugdir
, (const char *) token
->val
.str
.text
+ 1,
557 token
->val
.str
.len
- 4);
558 debugdir
[token
->val
.str
.len
- 4] = '\0';
560 pfile
->cb
.dir_change (pfile
, debugdir
);
564 /* This is called at the end of preprocessing. It pops the last
565 buffer and writes dependency output, and returns the number of
568 Maybe it should also reset state, such that you could call
569 cpp_start_read with a new filename to restart processing. */
571 cpp_finish (cpp_reader
*pfile
, FILE *deps_stream
)
573 /* Warn about unused macros before popping the final buffer. */
574 if (CPP_OPTION (pfile
, warn_unused_macros
))
575 cpp_forall_identifiers (pfile
, _cpp_warn_if_unused_macro
, NULL
);
577 /* cpplex.c leaves the final buffer on the stack. This it so that
578 it returns an unending stream of CPP_EOFs to the client. If we
579 popped the buffer, we'd dereference a NULL buffer pointer and
580 segfault. It's nice to allow the client to do worry-free excess
581 cpp_get_token calls. */
582 while (pfile
->buffer
)
583 _cpp_pop_buffer (pfile
);
585 /* Don't write the deps file if there are errors. */
586 if (CPP_OPTION (pfile
, deps
.style
) != DEPS_NONE
587 && deps_stream
&& pfile
->errors
== 0)
589 deps_write (pfile
->deps
, deps_stream
, 72);
591 if (CPP_OPTION (pfile
, deps
.phony_targets
))
592 deps_phony_targets (pfile
->deps
, deps_stream
);
595 /* Report on headers that could use multiple include guards. */
596 if (CPP_OPTION (pfile
, print_include_names
))
597 _cpp_report_missing_guards (pfile
);
599 return pfile
->errors
;
603 post_options (cpp_reader
*pfile
)
605 /* -Wtraditional is not useful in C++ mode. */
606 if (CPP_OPTION (pfile
, cplusplus
))
607 CPP_OPTION (pfile
, warn_traditional
) = 0;
609 /* Permanently disable macro expansion if we are rescanning
610 preprocessed text. Read preprocesed source in ISO mode. */
611 if (CPP_OPTION (pfile
, preprocessed
))
613 pfile
->state
.prevent_expansion
= 1;
614 CPP_OPTION (pfile
, traditional
) = 0;
617 if (CPP_OPTION (pfile
, warn_trigraphs
) == 2)
618 CPP_OPTION (pfile
, warn_trigraphs
) = !CPP_OPTION (pfile
, trigraphs
);
620 if (CPP_OPTION (pfile
, traditional
))
622 /* Traditional CPP does not accurately track column information. */
623 CPP_OPTION (pfile
, show_column
) = 0;
624 CPP_OPTION (pfile
, trigraphs
) = 0;
625 CPP_OPTION (pfile
, warn_trigraphs
) = 0;