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
)
131 /* Initialize this instance of the library if it hasn't been already. */
134 pfile
= xcalloc (1, sizeof (cpp_reader
));
136 cpp_set_lang (pfile
, lang
);
137 CPP_OPTION (pfile
, warn_multichar
) = 1;
138 CPP_OPTION (pfile
, discard_comments
) = 1;
139 CPP_OPTION (pfile
, discard_comments_in_macro_exp
) = 1;
140 CPP_OPTION (pfile
, show_column
) = 1;
141 CPP_OPTION (pfile
, tabstop
) = 8;
142 CPP_OPTION (pfile
, operator_names
) = 1;
143 CPP_OPTION (pfile
, warn_trigraphs
) = 2;
144 CPP_OPTION (pfile
, warn_endif_labels
) = 1;
145 CPP_OPTION (pfile
, warn_deprecated
) = 1;
146 CPP_OPTION (pfile
, warn_long_long
) = !CPP_OPTION (pfile
, c99
);
147 CPP_OPTION (pfile
, dollars_in_ident
) = 1;
148 CPP_OPTION (pfile
, warn_dollars
) = 1;
150 /* Default CPP arithmetic to something sensible for the host for the
151 benefit of dumb users like fix-header. */
152 CPP_OPTION (pfile
, precision
) = CHAR_BIT
* sizeof (long);
153 CPP_OPTION (pfile
, char_precision
) = CHAR_BIT
;
154 CPP_OPTION (pfile
, wchar_precision
) = CHAR_BIT
* sizeof (int);
155 CPP_OPTION (pfile
, int_precision
) = CHAR_BIT
* sizeof (int);
156 CPP_OPTION (pfile
, unsigned_char
) = 0;
157 CPP_OPTION (pfile
, unsigned_wchar
) = 1;
158 CPP_OPTION (pfile
, bytes_big_endian
) = 1; /* does not matter */
160 /* Default to locale/UTF-8. */
161 CPP_OPTION (pfile
, narrow_charset
) = _cpp_default_encoding ();
162 CPP_OPTION (pfile
, wide_charset
) = 0;
163 CPP_OPTION (pfile
, input_charset
) = _cpp_default_encoding ();
165 /* A fake empty "directory" used as the starting point for files
166 looked up without a search path. Name cannot be '/' because we
167 don't want to prepend anything at all to filenames using it. All
168 other entries are correct zero-initialized. */
169 pfile
->no_search_path
.name
= (char *) "";
171 /* Initialize the line map. Start at logical line 1, so we can use
172 a line number of zero for special states. */
173 linemap_init (&pfile
->line_maps
);
176 /* Initialize lexer state. */
177 pfile
->state
.save_comments
= ! CPP_OPTION (pfile
, discard_comments
);
179 /* Set up static tokens. */
180 pfile
->avoid_paste
.type
= CPP_PADDING
;
181 pfile
->avoid_paste
.val
.source
= NULL
;
182 pfile
->eof
.type
= CPP_EOF
;
183 pfile
->eof
.flags
= 0;
185 /* Create a token buffer for the lexer. */
186 _cpp_init_tokenrun (&pfile
->base_run
, 250);
187 pfile
->cur_run
= &pfile
->base_run
;
188 pfile
->cur_token
= pfile
->base_run
.base
;
190 /* Initialize the base context. */
191 pfile
->context
= &pfile
->base_context
;
192 pfile
->base_context
.macro
= 0;
193 pfile
->base_context
.prev
= pfile
->base_context
.next
= 0;
195 /* Aligned and unaligned storage. */
196 pfile
->a_buff
= _cpp_get_buff (pfile
, 0);
197 pfile
->u_buff
= _cpp_get_buff (pfile
, 0);
199 /* The expression parser stack. */
200 _cpp_expand_op_stack (pfile
);
202 /* Initialize the buffer obstack. */
203 _obstack_begin (&pfile
->buffer_ob
, 0, 0,
204 (void *(*) (long)) xmalloc
,
205 (void (*) (void *)) free
);
207 _cpp_init_files (pfile
);
209 _cpp_init_hashtable (pfile
, table
);
214 /* Free resources used by PFILE. Accessing PFILE after this function
215 returns leads to undefined behavior. Returns the error count. */
217 cpp_destroy (cpp_reader
*pfile
)
219 cpp_context
*context
, *contextn
;
220 tokenrun
*run
, *runn
;
222 free (pfile
->op_stack
);
224 while (CPP_BUFFER (pfile
) != NULL
)
225 _cpp_pop_buffer (pfile
);
228 free (pfile
->out
.base
);
230 if (pfile
->macro_buffer
)
232 free (pfile
->macro_buffer
);
233 pfile
->macro_buffer
= NULL
;
234 pfile
->macro_buffer_len
= 0;
238 deps_free (pfile
->deps
);
239 obstack_free (&pfile
->buffer_ob
, 0);
241 _cpp_destroy_hashtable (pfile
);
242 _cpp_cleanup_files (pfile
);
243 _cpp_destroy_iconv (pfile
);
245 _cpp_free_buff (pfile
->a_buff
);
246 _cpp_free_buff (pfile
->u_buff
);
247 _cpp_free_buff (pfile
->free_buffs
);
249 for (run
= &pfile
->base_run
; run
; run
= runn
)
253 if (run
!= &pfile
->base_run
)
257 for (context
= pfile
->base_context
.next
; context
; context
= contextn
)
259 contextn
= context
->next
;
263 linemap_free (&pfile
->line_maps
);
267 /* This structure defines one built-in identifier. A node will be
268 entered in the hash table under the name NAME, with value VALUE.
270 There are two tables of these. builtin_array holds all the
271 "builtin" macros: these are handled by builtin_macro() in
272 cppmacro.c. Builtin is somewhat of a misnomer -- the property of
273 interest is that these macros require special code to compute their
274 expansions. The value is a "builtin_type" enumerator.
276 operator_array holds the C++ named operators. These are keywords
277 which act as aliases for punctuators. In C++, they cannot be
278 altered through #define, and #if recognizes them as operators. In
279 C, these are not entered into the hash table at all (but see
280 <iso646.h>). The value is a token-type enumerator. */
285 unsigned short value
;
288 #define B(n, t) { DSC(n), t }
289 static const struct builtin builtin_array
[] =
291 B("__TIME__", BT_TIME
),
292 B("__DATE__", BT_DATE
),
293 B("__FILE__", BT_FILE
),
294 B("__BASE_FILE__", BT_BASE_FILE
),
295 B("__LINE__", BT_SPECLINE
),
296 B("__INCLUDE_LEVEL__", BT_INCLUDE_LEVEL
),
297 /* Keep builtins not used for -traditional-cpp at the end, and
298 update init_builtins() if any more are added. */
299 B("_Pragma", BT_PRAGMA
),
300 B("__STDC__", BT_STDC
),
303 static const struct builtin operator_array
[] =
305 B("and", CPP_AND_AND
),
306 B("and_eq", CPP_AND_EQ
),
307 B("bitand", CPP_AND
),
309 B("compl", CPP_COMPL
),
311 B("not_eq", CPP_NOT_EQ
),
313 B("or_eq", CPP_OR_EQ
),
315 B("xor_eq", CPP_XOR_EQ
)
319 /* Mark the C++ named operators in the hash table. */
321 mark_named_operators (cpp_reader
*pfile
)
323 const struct builtin
*b
;
325 for (b
= operator_array
;
326 b
< (operator_array
+ ARRAY_SIZE (operator_array
));
329 cpp_hashnode
*hp
= cpp_lookup (pfile
, b
->name
, b
->len
);
330 hp
->flags
|= NODE_OPERATOR
;
331 hp
->is_directive
= 0;
332 hp
->directive_index
= b
->value
;
336 /* Read the builtins table above and enter them, and language-specific
337 macros, into the hash table. HOSTED is true if this is a hosted
340 cpp_init_builtins (cpp_reader
*pfile
, int hosted
)
342 const struct builtin
*b
;
343 size_t n
= ARRAY_SIZE (builtin_array
);
345 if (CPP_OPTION (pfile
, traditional
))
348 for(b
= builtin_array
; b
< builtin_array
+ n
; b
++)
350 cpp_hashnode
*hp
= cpp_lookup (pfile
, b
->name
, b
->len
);
352 hp
->flags
|= NODE_BUILTIN
| NODE_WARN
;
353 hp
->value
.builtin
= b
->value
;
356 if (CPP_OPTION (pfile
, cplusplus
))
357 _cpp_define_builtin (pfile
, "__cplusplus 1");
358 else if (CPP_OPTION (pfile
, lang
) == CLK_ASM
)
359 _cpp_define_builtin (pfile
, "__ASSEMBLER__ 1");
360 else if (CPP_OPTION (pfile
, lang
) == CLK_STDC94
)
361 _cpp_define_builtin (pfile
, "__STDC_VERSION__ 199409L");
362 else if (CPP_OPTION (pfile
, c99
))
363 _cpp_define_builtin (pfile
, "__STDC_VERSION__ 199901L");
366 _cpp_define_builtin (pfile
, "__STDC_HOSTED__ 1");
368 _cpp_define_builtin (pfile
, "__STDC_HOSTED__ 0");
370 if (CPP_OPTION (pfile
, objc
))
371 _cpp_define_builtin (pfile
, "__OBJC__ 1");
374 /* Sanity-checks are dependent on command-line options, so it is
375 called as a subroutine of cpp_read_main_file (). */
377 static void sanity_checks (cpp_reader
*);
378 static void sanity_checks (cpp_reader
*pfile
)
381 size_t max_precision
= 2 * CHAR_BIT
* sizeof (cpp_num_part
);
383 /* Sanity checks for assumptions about CPP arithmetic and target
384 type precisions made by cpplib. */
387 cpp_error (pfile
, CPP_DL_ICE
, "cppchar_t must be an unsigned type");
389 if (CPP_OPTION (pfile
, precision
) > max_precision
)
390 cpp_error (pfile
, CPP_DL_ICE
,
391 "preprocessor arithmetic has maximum precision of %lu bits;"
392 " target requires %lu bits",
393 (unsigned long) max_precision
,
394 (unsigned long) CPP_OPTION (pfile
, precision
));
396 if (CPP_OPTION (pfile
, precision
) < CPP_OPTION (pfile
, int_precision
))
397 cpp_error (pfile
, CPP_DL_ICE
,
398 "CPP arithmetic must be at least as precise as a target int");
400 if (CPP_OPTION (pfile
, char_precision
) < 8)
401 cpp_error (pfile
, CPP_DL_ICE
, "target char is less than 8 bits wide");
403 if (CPP_OPTION (pfile
, wchar_precision
) < CPP_OPTION (pfile
, char_precision
))
404 cpp_error (pfile
, CPP_DL_ICE
,
405 "target wchar_t is narrower than target char");
407 if (CPP_OPTION (pfile
, int_precision
) < CPP_OPTION (pfile
, char_precision
))
408 cpp_error (pfile
, CPP_DL_ICE
,
409 "target int is narrower than target char");
411 /* This is assumed in eval_token() and could be fixed if necessary. */
412 if (sizeof (cppchar_t
) > sizeof (cpp_num_part
))
413 cpp_error (pfile
, CPP_DL_ICE
,
414 "CPP half-integer narrower than CPP character");
416 if (CPP_OPTION (pfile
, wchar_precision
) > BITS_PER_CPPCHAR_T
)
417 cpp_error (pfile
, CPP_DL_ICE
,
418 "CPP on this host cannot handle wide character constants over"
419 " %lu bits, but the target requires %lu bits",
420 (unsigned long) BITS_PER_CPPCHAR_T
,
421 (unsigned long) CPP_OPTION (pfile
, wchar_precision
));
424 # define sanity_checks(PFILE)
427 /* Add a dependency target. Can be called any number of times before
428 cpp_read_main_file(). If no targets have been added before
429 cpp_read_main_file(), then the default target is used. */
431 cpp_add_dependency_target (cpp_reader
*pfile
, const char *target
, int quote
)
434 pfile
->deps
= deps_init ();
436 deps_add_target (pfile
->deps
, target
, quote
);
439 /* This is called after options have been parsed, and partially
442 cpp_post_options (cpp_reader
*pfile
)
444 sanity_checks (pfile
);
446 post_options (pfile
);
448 /* Mark named operators before handling command line macros. */
449 if (CPP_OPTION (pfile
, cplusplus
) && CPP_OPTION (pfile
, operator_names
))
450 mark_named_operators (pfile
);
453 /* Setup for processing input from the file named FNAME, or stdin if
454 it is the empty string. Return the original filename
455 on success (e.g. foo.i->foo.c), or NULL on failure. */
457 cpp_read_main_file (cpp_reader
*pfile
, const char *fname
)
459 if (CPP_OPTION (pfile
, deps
.style
) != DEPS_NONE
)
462 pfile
->deps
= deps_init ();
464 /* Set the default target (if there is none already). */
465 deps_add_default_target (pfile
->deps
, fname
);
469 = _cpp_find_file (pfile
, fname
, &pfile
->no_search_path
, false, 0);
470 if (_cpp_find_failed (pfile
->main_file
))
473 _cpp_stack_file (pfile
, pfile
->main_file
, false);
475 /* For foo.i, read the original filename foo.c now, for the benefit
476 of the front ends. */
477 if (CPP_OPTION (pfile
, preprocessed
))
479 read_original_filename (pfile
);
482 fname
= pfile
->map
->to_file
;
487 /* For preprocessed files, if the first tokens are of the form # NUM.
488 handle the directive so we know the original file name. This will
489 generate file_change callbacks, which the front ends must handle
490 appropriately given their state of initialization. */
492 read_original_filename (cpp_reader
*pfile
)
494 const cpp_token
*token
, *token1
;
496 /* Lex ahead; if the first tokens are of the form # NUM, then
497 process the directive, otherwise back up. */
498 token
= _cpp_lex_direct (pfile
);
499 if (token
->type
== CPP_HASH
)
501 pfile
->state
.in_directive
= 1;
502 token1
= _cpp_lex_direct (pfile
);
503 _cpp_backup_tokens (pfile
, 1);
504 pfile
->state
.in_directive
= 0;
506 /* If it's a #line directive, handle it. */
507 if (token1
->type
== CPP_NUMBER
)
509 _cpp_handle_directive (pfile
, token
->flags
& PREV_WHITE
);
510 read_original_directory (pfile
);
515 /* Backup as if nothing happened. */
516 _cpp_backup_tokens (pfile
, 1);
519 /* For preprocessed files, if the tokens following the first filename
520 line is of the form # <line> "/path/name//", handle the
521 directive so we know the original current directory. */
523 read_original_directory (cpp_reader
*pfile
)
525 const cpp_token
*hash
, *token
;
527 /* Lex ahead; if the first tokens are of the form # NUM, then
528 process the directive, otherwise back up. */
529 hash
= _cpp_lex_direct (pfile
);
530 if (hash
->type
!= CPP_HASH
)
532 _cpp_backup_tokens (pfile
, 1);
536 token
= _cpp_lex_direct (pfile
);
538 if (token
->type
!= CPP_NUMBER
)
540 _cpp_backup_tokens (pfile
, 2);
544 token
= _cpp_lex_direct (pfile
);
546 if (token
->type
!= CPP_STRING
547 || ! (token
->val
.str
.len
>= 5
548 && token
->val
.str
.text
[token
->val
.str
.len
-2] == '/'
549 && token
->val
.str
.text
[token
->val
.str
.len
-3] == '/'))
551 _cpp_backup_tokens (pfile
, 3);
555 if (pfile
->cb
.dir_change
)
557 char *debugdir
= alloca (token
->val
.str
.len
- 3);
559 memcpy (debugdir
, (const char *) token
->val
.str
.text
+ 1,
560 token
->val
.str
.len
- 4);
561 debugdir
[token
->val
.str
.len
- 4] = '\0';
563 pfile
->cb
.dir_change (pfile
, debugdir
);
567 /* This is called at the end of preprocessing. It pops the last
568 buffer and writes dependency output, and returns the number of
571 Maybe it should also reset state, such that you could call
572 cpp_start_read with a new filename to restart processing. */
574 cpp_finish (cpp_reader
*pfile
, FILE *deps_stream
)
576 /* Warn about unused macros before popping the final buffer. */
577 if (CPP_OPTION (pfile
, warn_unused_macros
))
578 cpp_forall_identifiers (pfile
, _cpp_warn_if_unused_macro
, NULL
);
580 /* cpplex.c leaves the final buffer on the stack. This it so that
581 it returns an unending stream of CPP_EOFs to the client. If we
582 popped the buffer, we'd dereference a NULL buffer pointer and
583 segfault. It's nice to allow the client to do worry-free excess
584 cpp_get_token calls. */
585 while (pfile
->buffer
)
586 _cpp_pop_buffer (pfile
);
588 /* Don't write the deps file if there are errors. */
589 if (CPP_OPTION (pfile
, deps
.style
) != DEPS_NONE
590 && deps_stream
&& pfile
->errors
== 0)
592 deps_write (pfile
->deps
, deps_stream
, 72);
594 if (CPP_OPTION (pfile
, deps
.phony_targets
))
595 deps_phony_targets (pfile
->deps
, deps_stream
);
598 /* Report on headers that could use multiple include guards. */
599 if (CPP_OPTION (pfile
, print_include_names
))
600 _cpp_report_missing_guards (pfile
);
602 return pfile
->errors
;
606 post_options (cpp_reader
*pfile
)
608 /* -Wtraditional is not useful in C++ mode. */
609 if (CPP_OPTION (pfile
, cplusplus
))
610 CPP_OPTION (pfile
, warn_traditional
) = 0;
612 /* Permanently disable macro expansion if we are rescanning
613 preprocessed text. Read preprocesed source in ISO mode. */
614 if (CPP_OPTION (pfile
, preprocessed
))
616 pfile
->state
.prevent_expansion
= 1;
617 CPP_OPTION (pfile
, traditional
) = 0;
620 if (CPP_OPTION (pfile
, warn_trigraphs
) == 2)
621 CPP_OPTION (pfile
, warn_trigraphs
) = !CPP_OPTION (pfile
, trigraphs
);
623 if (CPP_OPTION (pfile
, traditional
))
625 CPP_OPTION (pfile
, cplusplus_comments
) = 0;
627 /* Traditional CPP does not accurately track column information. */
628 CPP_OPTION (pfile
, show_column
) = 0;
629 CPP_OPTION (pfile
, trigraphs
) = 0;
630 CPP_OPTION (pfile
, warn_trigraphs
) = 0;