cpplib.h (struct cpp_options): Add canonical_system_headers.
[gcc.git] / libcpp / include / cpplib.h
1 /* Definitions for CPP library.
2 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
3 2004, 2005, 2007, 2008, 2009, 2010, 2011, 2012
4 Free Software Foundation, Inc.
5 Written by Per Bothner, 1994-95.
6
7 This program is free software; you can redistribute it and/or modify it
8 under the terms of the GNU General Public License as published by the
9 Free Software Foundation; either version 3, or (at your option) any
10 later version.
11
12 This program 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 this program; see the file COPYING3. If not see
19 <http://www.gnu.org/licenses/>.
20
21 In other words, you are welcome to use, share and improve this program.
22 You are forbidden to forbid anyone else to use, share and improve
23 what you give them. Help stamp out software-hoarding! */
24 #ifndef LIBCPP_CPPLIB_H
25 #define LIBCPP_CPPLIB_H
26
27 #include <sys/types.h>
28 #include "symtab.h"
29 #include "line-map.h"
30
31 typedef struct cpp_reader cpp_reader;
32 typedef struct cpp_buffer cpp_buffer;
33 typedef struct cpp_options cpp_options;
34 typedef struct cpp_token cpp_token;
35 typedef struct cpp_string cpp_string;
36 typedef struct cpp_hashnode cpp_hashnode;
37 typedef struct cpp_macro cpp_macro;
38 typedef struct cpp_callbacks cpp_callbacks;
39 typedef struct cpp_dir cpp_dir;
40
41 struct answer;
42 struct _cpp_file;
43
44 /* The first three groups, apart from '=', can appear in preprocessor
45 expressions (+= and -= are used to indicate unary + and - resp.).
46 This allows a lookup table to be implemented in _cpp_parse_expr.
47
48 The first group, to CPP_LAST_EQ, can be immediately followed by an
49 '='. The lexer needs operators ending in '=', like ">>=", to be in
50 the same order as their counterparts without the '=', like ">>".
51
52 See the cpp_operator table optab in expr.c if you change the order or
53 add or remove anything in the first group. */
54
55 #define TTYPE_TABLE \
56 OP(EQ, "=") \
57 OP(NOT, "!") \
58 OP(GREATER, ">") /* compare */ \
59 OP(LESS, "<") \
60 OP(PLUS, "+") /* math */ \
61 OP(MINUS, "-") \
62 OP(MULT, "*") \
63 OP(DIV, "/") \
64 OP(MOD, "%") \
65 OP(AND, "&") /* bit ops */ \
66 OP(OR, "|") \
67 OP(XOR, "^") \
68 OP(RSHIFT, ">>") \
69 OP(LSHIFT, "<<") \
70 \
71 OP(COMPL, "~") \
72 OP(AND_AND, "&&") /* logical */ \
73 OP(OR_OR, "||") \
74 OP(QUERY, "?") \
75 OP(COLON, ":") \
76 OP(COMMA, ",") /* grouping */ \
77 OP(OPEN_PAREN, "(") \
78 OP(CLOSE_PAREN, ")") \
79 TK(EOF, NONE) \
80 OP(EQ_EQ, "==") /* compare */ \
81 OP(NOT_EQ, "!=") \
82 OP(GREATER_EQ, ">=") \
83 OP(LESS_EQ, "<=") \
84 \
85 /* These two are unary + / - in preprocessor expressions. */ \
86 OP(PLUS_EQ, "+=") /* math */ \
87 OP(MINUS_EQ, "-=") \
88 \
89 OP(MULT_EQ, "*=") \
90 OP(DIV_EQ, "/=") \
91 OP(MOD_EQ, "%=") \
92 OP(AND_EQ, "&=") /* bit ops */ \
93 OP(OR_EQ, "|=") \
94 OP(XOR_EQ, "^=") \
95 OP(RSHIFT_EQ, ">>=") \
96 OP(LSHIFT_EQ, "<<=") \
97 /* Digraphs together, beginning with CPP_FIRST_DIGRAPH. */ \
98 OP(HASH, "#") /* digraphs */ \
99 OP(PASTE, "##") \
100 OP(OPEN_SQUARE, "[") \
101 OP(CLOSE_SQUARE, "]") \
102 OP(OPEN_BRACE, "{") \
103 OP(CLOSE_BRACE, "}") \
104 /* The remainder of the punctuation. Order is not significant. */ \
105 OP(SEMICOLON, ";") /* structure */ \
106 OP(ELLIPSIS, "...") \
107 OP(PLUS_PLUS, "++") /* increment */ \
108 OP(MINUS_MINUS, "--") \
109 OP(DEREF, "->") /* accessors */ \
110 OP(DOT, ".") \
111 OP(SCOPE, "::") \
112 OP(DEREF_STAR, "->*") \
113 OP(DOT_STAR, ".*") \
114 OP(ATSIGN, "@") /* used in Objective-C */ \
115 \
116 TK(NAME, IDENT) /* word */ \
117 TK(AT_NAME, IDENT) /* @word - Objective-C */ \
118 TK(NUMBER, LITERAL) /* 34_be+ta */ \
119 \
120 TK(CHAR, LITERAL) /* 'char' */ \
121 TK(WCHAR, LITERAL) /* L'char' */ \
122 TK(CHAR16, LITERAL) /* u'char' */ \
123 TK(CHAR32, LITERAL) /* U'char' */ \
124 TK(OTHER, LITERAL) /* stray punctuation */ \
125 \
126 TK(STRING, LITERAL) /* "string" */ \
127 TK(WSTRING, LITERAL) /* L"string" */ \
128 TK(STRING16, LITERAL) /* u"string" */ \
129 TK(STRING32, LITERAL) /* U"string" */ \
130 TK(UTF8STRING, LITERAL) /* u8"string" */ \
131 TK(OBJC_STRING, LITERAL) /* @"string" - Objective-C */ \
132 TK(HEADER_NAME, LITERAL) /* <stdio.h> in #include */ \
133 \
134 TK(CHAR_USERDEF, LITERAL) /* 'char'_suffix - C++-0x */ \
135 TK(WCHAR_USERDEF, LITERAL) /* L'char'_suffix - C++-0x */ \
136 TK(CHAR16_USERDEF, LITERAL) /* u'char'_suffix - C++-0x */ \
137 TK(CHAR32_USERDEF, LITERAL) /* U'char'_suffix - C++-0x */ \
138 TK(STRING_USERDEF, LITERAL) /* "string"_suffix - C++-0x */ \
139 TK(WSTRING_USERDEF, LITERAL) /* L"string"_suffix - C++-0x */ \
140 TK(STRING16_USERDEF, LITERAL) /* u"string"_suffix - C++-0x */ \
141 TK(STRING32_USERDEF, LITERAL) /* U"string"_suffix - C++-0x */ \
142 TK(UTF8STRING_USERDEF,LITERAL) /* u8"string"_suffix - C++-0x */ \
143 \
144 TK(COMMENT, LITERAL) /* Only if output comments. */ \
145 /* SPELL_LITERAL happens to DTRT. */ \
146 TK(MACRO_ARG, NONE) /* Macro argument. */ \
147 TK(PRAGMA, NONE) /* Only for deferred pragmas. */ \
148 TK(PRAGMA_EOL, NONE) /* End-of-line for deferred pragmas. */ \
149 TK(PADDING, NONE) /* Whitespace for -E. */
150
151 #define OP(e, s) CPP_ ## e,
152 #define TK(e, s) CPP_ ## e,
153 enum cpp_ttype
154 {
155 TTYPE_TABLE
156 N_TTYPES,
157
158 /* Positions in the table. */
159 CPP_LAST_EQ = CPP_LSHIFT,
160 CPP_FIRST_DIGRAPH = CPP_HASH,
161 CPP_LAST_PUNCTUATOR= CPP_ATSIGN,
162 CPP_LAST_CPP_OP = CPP_LESS_EQ
163 };
164 #undef OP
165 #undef TK
166
167 /* C language kind, used when calling cpp_create_reader. */
168 enum c_lang {CLK_GNUC89 = 0, CLK_GNUC99, CLK_GNUC11,
169 CLK_STDC89, CLK_STDC94, CLK_STDC99, CLK_STDC11,
170 CLK_GNUCXX, CLK_CXX98, CLK_GNUCXX11, CLK_CXX11, CLK_ASM};
171
172 /* Payload of a NUMBER, STRING, CHAR or COMMENT token. */
173 struct GTY(()) cpp_string {
174 unsigned int len;
175 const unsigned char *text;
176 };
177
178 /* Flags for the cpp_token structure. */
179 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */
180 #define DIGRAPH (1 << 1) /* If it was a digraph. */
181 #define STRINGIFY_ARG (1 << 2) /* If macro argument to be stringified. */
182 #define PASTE_LEFT (1 << 3) /* If on LHS of a ## operator. */
183 #define NAMED_OP (1 << 4) /* C++ named operators. */
184 #define NO_EXPAND (1 << 5) /* Do not macro-expand this token. */
185 #define BOL (1 << 6) /* Token at beginning of line. */
186 #define PURE_ZERO (1 << 7) /* Single 0 digit, used by the C++ frontend,
187 set in c-lex.c. */
188 #define SP_DIGRAPH (1 << 8) /* # or ## token was a digraph. */
189 #define SP_PREV_WHITE (1 << 9) /* If whitespace before a ##
190 operator, or before this token
191 after a # operator. */
192
193 /* Specify which field, if any, of the cpp_token union is used. */
194
195 enum cpp_token_fld_kind {
196 CPP_TOKEN_FLD_NODE,
197 CPP_TOKEN_FLD_SOURCE,
198 CPP_TOKEN_FLD_STR,
199 CPP_TOKEN_FLD_ARG_NO,
200 CPP_TOKEN_FLD_TOKEN_NO,
201 CPP_TOKEN_FLD_PRAGMA,
202 CPP_TOKEN_FLD_NONE
203 };
204
205 /* A macro argument in the cpp_token union. */
206 struct GTY(()) cpp_macro_arg {
207 /* Argument number. */
208 unsigned int arg_no;
209 };
210
211 /* An identifier in the cpp_token union. */
212 struct GTY(()) cpp_identifier {
213 /* The canonical (UTF-8) spelling of the identifier. */
214 cpp_hashnode *
215 GTY ((nested_ptr (union tree_node,
216 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
217 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL")))
218 node;
219 };
220
221 /* A preprocessing token. This has been carefully packed and should
222 occupy 16 bytes on 32-bit hosts and 24 bytes on 64-bit hosts. */
223 struct GTY(()) cpp_token {
224 source_location src_loc; /* Location of first char of token. */
225 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
226 unsigned short flags; /* flags - see above */
227
228 union cpp_token_u
229 {
230 /* An identifier. */
231 struct cpp_identifier GTY ((tag ("CPP_TOKEN_FLD_NODE"))) node;
232
233 /* Inherit padding from this token. */
234 cpp_token * GTY ((tag ("CPP_TOKEN_FLD_SOURCE"))) source;
235
236 /* A string, or number. */
237 struct cpp_string GTY ((tag ("CPP_TOKEN_FLD_STR"))) str;
238
239 /* Argument no. for a CPP_MACRO_ARG. */
240 struct cpp_macro_arg GTY ((tag ("CPP_TOKEN_FLD_ARG_NO"))) macro_arg;
241
242 /* Original token no. for a CPP_PASTE (from a sequence of
243 consecutive paste tokens in a macro expansion). */
244 unsigned int GTY ((tag ("CPP_TOKEN_FLD_TOKEN_NO"))) token_no;
245
246 /* Caller-supplied identifier for a CPP_PRAGMA. */
247 unsigned int GTY ((tag ("CPP_TOKEN_FLD_PRAGMA"))) pragma;
248 } GTY ((desc ("cpp_token_val_index (&%1)"))) val;
249 };
250
251 /* Say which field is in use. */
252 extern enum cpp_token_fld_kind cpp_token_val_index (cpp_token *tok);
253
254 /* A type wide enough to hold any multibyte source character.
255 cpplib's character constant interpreter requires an unsigned type.
256 Also, a typedef for the signed equivalent.
257 The width of this type is capped at 32 bits; there do exist targets
258 where wchar_t is 64 bits, but only in a non-default mode, and there
259 would be no meaningful interpretation for a wchar_t value greater
260 than 2^32 anyway -- the widest wide-character encoding around is
261 ISO 10646, which stops at 2^31. */
262 #if CHAR_BIT * SIZEOF_INT >= 32
263 # define CPPCHAR_SIGNED_T int
264 #elif CHAR_BIT * SIZEOF_LONG >= 32
265 # define CPPCHAR_SIGNED_T long
266 #else
267 # error "Cannot find a least-32-bit signed integer type"
268 #endif
269 typedef unsigned CPPCHAR_SIGNED_T cppchar_t;
270 typedef CPPCHAR_SIGNED_T cppchar_signed_t;
271
272 /* Style of header dependencies to generate. */
273 enum cpp_deps_style { DEPS_NONE = 0, DEPS_USER, DEPS_SYSTEM };
274
275 /* The possible normalization levels, from most restrictive to least. */
276 enum cpp_normalize_level {
277 /* In NFKC. */
278 normalized_KC = 0,
279 /* In NFC. */
280 normalized_C,
281 /* In NFC, except for subsequences where being in NFC would make
282 the identifier invalid. */
283 normalized_identifier_C,
284 /* Not normalized at all. */
285 normalized_none
286 };
287
288 /* This structure is nested inside struct cpp_reader, and
289 carries all the options visible to the command line. */
290 struct cpp_options
291 {
292 /* Characters between tab stops. */
293 unsigned int tabstop;
294
295 /* The language we're preprocessing. */
296 enum c_lang lang;
297
298 /* Nonzero means use extra default include directories for C++. */
299 unsigned char cplusplus;
300
301 /* Nonzero means handle cplusplus style comments. */
302 unsigned char cplusplus_comments;
303
304 /* Nonzero means define __OBJC__, treat @ as a special token, use
305 the OBJC[PLUS]_INCLUDE_PATH environment variable, and allow
306 "#import". */
307 unsigned char objc;
308
309 /* Nonzero means don't copy comments into the output file. */
310 unsigned char discard_comments;
311
312 /* Nonzero means don't copy comments into the output file during
313 macro expansion. */
314 unsigned char discard_comments_in_macro_exp;
315
316 /* Nonzero means process the ISO trigraph sequences. */
317 unsigned char trigraphs;
318
319 /* Nonzero means process the ISO digraph sequences. */
320 unsigned char digraphs;
321
322 /* Nonzero means to allow hexadecimal floats and LL suffixes. */
323 unsigned char extended_numbers;
324
325 /* Nonzero means process u/U prefix literals (UTF-16/32). */
326 unsigned char uliterals;
327
328 /* Nonzero means process r/R raw strings. If this is set, uliterals
329 must be set as well. */
330 unsigned char rliterals;
331
332 /* Nonzero means print names of header files (-H). */
333 unsigned char print_include_names;
334
335 /* Nonzero means complain about deprecated features. */
336 unsigned char cpp_warn_deprecated;
337
338 /* Nonzero means warn if slash-star appears in a comment. */
339 unsigned char warn_comments;
340
341 /* Nonzero means warn if a user-supplied include directory does not
342 exist. */
343 unsigned char warn_missing_include_dirs;
344
345 /* Nonzero means warn if there are any trigraphs. */
346 unsigned char warn_trigraphs;
347
348 /* Nonzero means warn about multicharacter charconsts. */
349 unsigned char warn_multichar;
350
351 /* Nonzero means warn about various incompatibilities with
352 traditional C. */
353 unsigned char cpp_warn_traditional;
354
355 /* Nonzero means warn about long long numeric constants. */
356 unsigned char cpp_warn_long_long;
357
358 /* Nonzero means warn about text after an #endif (or #else). */
359 unsigned char warn_endif_labels;
360
361 /* Nonzero means warn about implicit sign changes owing to integer
362 promotions. */
363 unsigned char warn_num_sign_change;
364
365 /* Zero means don't warn about __VA_ARGS__ usage in c89 pedantic mode.
366 Presumably the usage is protected by the appropriate #ifdef. */
367 unsigned char warn_variadic_macros;
368
369 /* Nonzero means warn about builtin macros that are redefined or
370 explicitly undefined. */
371 unsigned char warn_builtin_macro_redefined;
372
373 /* Nonzero means we should look for header.gcc files that remap file
374 names. */
375 unsigned char remap;
376
377 /* Zero means dollar signs are punctuation. */
378 unsigned char dollars_in_ident;
379
380 /* Nonzero means UCNs are accepted in identifiers. */
381 unsigned char extended_identifiers;
382
383 /* True if we should warn about dollars in identifiers or numbers
384 for this translation unit. */
385 unsigned char warn_dollars;
386
387 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
388 unsigned char warn_undef;
389
390 /* Nonzero means warn of unused macros from the main file. */
391 unsigned char warn_unused_macros;
392
393 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
394 unsigned char c99;
395
396 /* Nonzero if we are conforming to a specific C or C++ standard. */
397 unsigned char std;
398
399 /* Nonzero means give all the error messages the ANSI standard requires. */
400 unsigned char cpp_pedantic;
401
402 /* Nonzero means we're looking at already preprocessed code, so don't
403 bother trying to do macro expansion and whatnot. */
404 unsigned char preprocessed;
405
406 /* Nonzero means we are going to emit debugging logs during
407 preprocessing. */
408 unsigned char debug;
409
410 /* Nonzero means we are tracking locations of tokens involved in
411 macro expansion. 1 Means we track the location in degraded mode
412 where we do not track locations of tokens resulting from the
413 expansion of arguments of function-like macro. 2 Means we do
414 track all macro expansions. This last option is the one that
415 consumes the highest amount of memory. */
416 unsigned char track_macro_expansion;
417
418 /* Nonzero means handle C++ alternate operator names. */
419 unsigned char operator_names;
420
421 /* Nonzero means warn about use of C++ alternate operator names. */
422 unsigned char warn_cxx_operator_names;
423
424 /* True for traditional preprocessing. */
425 unsigned char traditional;
426
427 /* Nonzero for C++ 2011 Standard user-defnied literals. */
428 unsigned char user_literals;
429
430 /* Nonzero means warn when a string or character literal is followed by a
431 ud-suffix which does not beging with an underscore. */
432 unsigned char warn_literal_suffix;
433
434 /* Nonzero means interpret imaginary, fixed-point, or other gnu extension
435 literal number suffixes as user-defined literal number suffixes. */
436 unsigned char ext_numeric_literals;
437
438 /* Holds the name of the target (execution) character set. */
439 const char *narrow_charset;
440
441 /* Holds the name of the target wide character set. */
442 const char *wide_charset;
443
444 /* Holds the name of the input character set. */
445 const char *input_charset;
446
447 /* The minimum permitted level of normalization before a warning
448 is generated. */
449 enum cpp_normalize_level warn_normalize;
450
451 /* True to warn about precompiled header files we couldn't use. */
452 bool warn_invalid_pch;
453
454 /* True if dependencies should be restored from a precompiled header. */
455 bool restore_pch_deps;
456
457 /* Dependency generation. */
458 struct
459 {
460 /* Style of header dependencies to generate. */
461 enum cpp_deps_style style;
462
463 /* Assume missing files are generated files. */
464 bool missing_files;
465
466 /* Generate phony targets for each dependency apart from the first
467 one. */
468 bool phony_targets;
469
470 /* If true, no dependency is generated on the main file. */
471 bool ignore_main_file;
472
473 /* If true, intend to use the preprocessor output (e.g., for compilation)
474 in addition to the dependency info. */
475 bool need_preprocessor_output;
476 } deps;
477
478 /* Target-specific features set by the front end or client. */
479
480 /* Precision for target CPP arithmetic, target characters, target
481 ints and target wide characters, respectively. */
482 size_t precision, char_precision, int_precision, wchar_precision;
483
484 /* True means chars (wide chars) are unsigned. */
485 bool unsigned_char, unsigned_wchar;
486
487 /* True if the most significant byte in a word has the lowest
488 address in memory. */
489 bool bytes_big_endian;
490
491 /* Nonzero means __STDC__ should have the value 0 in system headers. */
492 unsigned char stdc_0_in_system_headers;
493
494 /* True disables tokenization outside of preprocessing directives. */
495 bool directives_only;
496
497 /* True enables canonicalization of system header file paths. */
498 bool canonical_system_headers;
499 };
500
501 /* Callback for header lookup for HEADER, which is the name of a
502 source file. It is used as a method of last resort to find headers
503 that are not otherwise found during the normal include processing.
504 The return value is the malloced name of a header to try and open,
505 if any, or NULL otherwise. This callback is called only if the
506 header is otherwise unfound. */
507 typedef const char *(*missing_header_cb)(cpp_reader *, const char *header, cpp_dir **);
508
509 /* Call backs to cpplib client. */
510 struct cpp_callbacks
511 {
512 /* Called when a new line of preprocessed output is started. */
513 void (*line_change) (cpp_reader *, const cpp_token *, int);
514
515 /* Called when switching to/from a new file.
516 The line_map is for the new file. It is NULL if there is no new file.
517 (In C this happens when done with <built-in>+<command line> and also
518 when done with a main file.) This can be used for resource cleanup. */
519 void (*file_change) (cpp_reader *, const struct line_map *);
520
521 void (*dir_change) (cpp_reader *, const char *);
522 void (*include) (cpp_reader *, source_location, const unsigned char *,
523 const char *, int, const cpp_token **);
524 void (*define) (cpp_reader *, source_location, cpp_hashnode *);
525 void (*undef) (cpp_reader *, source_location, cpp_hashnode *);
526 void (*ident) (cpp_reader *, source_location, const cpp_string *);
527 void (*def_pragma) (cpp_reader *, source_location);
528 int (*valid_pch) (cpp_reader *, const char *, int);
529 void (*read_pch) (cpp_reader *, const char *, int, const char *);
530 missing_header_cb missing_header;
531
532 /* Context-sensitive macro support. Returns macro (if any) that should
533 be expanded. */
534 cpp_hashnode * (*macro_to_expand) (cpp_reader *, const cpp_token *);
535
536 /* Called to emit a diagnostic. This callback receives the
537 translated message. */
538 bool (*error) (cpp_reader *, int, int, source_location, unsigned int,
539 const char *, va_list *)
540 ATTRIBUTE_FPTR_PRINTF(6,0);
541
542 /* Callbacks for when a macro is expanded, or tested (whether
543 defined or not at the time) in #ifdef, #ifndef or "defined". */
544 void (*used_define) (cpp_reader *, source_location, cpp_hashnode *);
545 void (*used_undef) (cpp_reader *, source_location, cpp_hashnode *);
546 /* Called before #define and #undef or other macro definition
547 changes are processed. */
548 void (*before_define) (cpp_reader *);
549 /* Called whenever a macro is expanded or tested.
550 Second argument is the location of the start of the current expansion. */
551 void (*used) (cpp_reader *, source_location, cpp_hashnode *);
552
553 /* Callback that can change a user builtin into normal macro. */
554 bool (*user_builtin_macro) (cpp_reader *, cpp_hashnode *);
555 };
556
557 #ifdef VMS
558 #define INO_T_CPP ino_t ino[3]
559 #else
560 #define INO_T_CPP ino_t ino
561 #endif
562
563 /* Chain of directories to look for include files in. */
564 struct cpp_dir
565 {
566 /* NULL-terminated singly-linked list. */
567 struct cpp_dir *next;
568
569 /* NAME of the directory, NUL-terminated. */
570 char *name;
571 unsigned int len;
572
573 /* One if a system header, two if a system header that has extern
574 "C" guards for C++. */
575 unsigned char sysp;
576
577 /* Is this a user-supplied directory? */
578 bool user_supplied_p;
579
580 /* The canonicalized NAME as determined by lrealpath. This field
581 is only used by hosts that lack reliable inode numbers. */
582 char *canonical_name;
583
584 /* Mapping of file names for this directory for MS-DOS and related
585 platforms. A NULL-terminated array of (from, to) pairs. */
586 const char **name_map;
587
588 /* Routine to construct pathname, given the search path name and the
589 HEADER we are trying to find, return a constructed pathname to
590 try and open. If this is NULL, the constructed pathname is as
591 constructed by append_file_to_dir. */
592 char *(*construct) (const char *header, cpp_dir *dir);
593
594 /* The C front end uses these to recognize duplicated
595 directories in the search path. */
596 INO_T_CPP;
597 dev_t dev;
598 };
599
600 /* The structure of a node in the hash table. The hash table has
601 entries for all identifiers: either macros defined by #define
602 commands (type NT_MACRO), assertions created with #assert
603 (NT_ASSERTION), or neither of the above (NT_VOID). Builtin macros
604 like __LINE__ are flagged NODE_BUILTIN. Poisoned identifiers are
605 flagged NODE_POISONED. NODE_OPERATOR (C++ only) indicates an
606 identifier that behaves like an operator such as "xor".
607 NODE_DIAGNOSTIC is for speed in lex_token: it indicates a
608 diagnostic may be required for this node. Currently this only
609 applies to __VA_ARGS__, poisoned identifiers, and -Wc++-compat
610 warnings about NODE_OPERATOR. */
611
612 /* Hash node flags. */
613 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */
614 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */
615 #define NODE_BUILTIN (1 << 2) /* Builtin macro. */
616 #define NODE_DIAGNOSTIC (1 << 3) /* Possible diagnostic when lexed. */
617 #define NODE_WARN (1 << 4) /* Warn if redefined or undefined. */
618 #define NODE_DISABLED (1 << 5) /* A disabled macro. */
619 #define NODE_MACRO_ARG (1 << 6) /* Used during #define processing. */
620 #define NODE_USED (1 << 7) /* Dumped with -dU. */
621 #define NODE_CONDITIONAL (1 << 8) /* Conditional macro */
622 #define NODE_WARN_OPERATOR (1 << 9) /* Warn about C++ named operator. */
623
624 /* Different flavors of hash node. */
625 enum node_type
626 {
627 NT_VOID = 0, /* No definition yet. */
628 NT_MACRO, /* A macro of some form. */
629 NT_ASSERTION /* Predicate for #assert. */
630 };
631
632 /* Different flavors of builtin macro. _Pragma is an operator, but we
633 handle it with the builtin code for efficiency reasons. */
634 enum cpp_builtin_type
635 {
636 BT_SPECLINE = 0, /* `__LINE__' */
637 BT_DATE, /* `__DATE__' */
638 BT_FILE, /* `__FILE__' */
639 BT_BASE_FILE, /* `__BASE_FILE__' */
640 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
641 BT_TIME, /* `__TIME__' */
642 BT_STDC, /* `__STDC__' */
643 BT_PRAGMA, /* `_Pragma' operator */
644 BT_TIMESTAMP, /* `__TIMESTAMP__' */
645 BT_COUNTER, /* `__COUNTER__' */
646 BT_FIRST_USER, /* User defined builtin macros. */
647 BT_LAST_USER = BT_FIRST_USER + 31
648 };
649
650 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE))
651 #define HT_NODE(NODE) ((ht_identifier *) (NODE))
652 #define NODE_LEN(NODE) HT_LEN (&(NODE)->ident)
653 #define NODE_NAME(NODE) HT_STR (&(NODE)->ident)
654
655 /* Specify which field, if any, of the union is used. */
656
657 enum {
658 NTV_MACRO,
659 NTV_ANSWER,
660 NTV_BUILTIN,
661 NTV_ARGUMENT,
662 NTV_NONE
663 };
664
665 #define CPP_HASHNODE_VALUE_IDX(HNODE) \
666 ((HNODE.flags & NODE_MACRO_ARG) ? NTV_ARGUMENT \
667 : HNODE.type == NT_MACRO ? ((HNODE.flags & NODE_BUILTIN) \
668 ? NTV_BUILTIN : NTV_MACRO) \
669 : HNODE.type == NT_ASSERTION ? NTV_ANSWER \
670 : NTV_NONE)
671
672 /* The common part of an identifier node shared amongst all 3 C front
673 ends. Also used to store CPP identifiers, which are a superset of
674 identifiers in the grammatical sense. */
675
676 union GTY(()) _cpp_hashnode_value {
677 /* If a macro. */
678 cpp_macro * GTY((tag ("NTV_MACRO"))) macro;
679 /* Answers to an assertion. */
680 struct answer * GTY ((tag ("NTV_ANSWER"))) answers;
681 /* Code for a builtin macro. */
682 enum cpp_builtin_type GTY ((tag ("NTV_BUILTIN"))) builtin;
683 /* Macro argument index. */
684 unsigned short GTY ((tag ("NTV_ARGUMENT"))) arg_index;
685 };
686
687 struct GTY(()) cpp_hashnode {
688 struct ht_identifier ident;
689 unsigned int is_directive : 1;
690 unsigned int directive_index : 7; /* If is_directive,
691 then index into directive table.
692 Otherwise, a NODE_OPERATOR. */
693 unsigned char rid_code; /* Rid code - for front ends. */
694 ENUM_BITFIELD(node_type) type : 6; /* CPP node type. */
695 unsigned int flags : 10; /* CPP flags. */
696
697 union _cpp_hashnode_value GTY ((desc ("CPP_HASHNODE_VALUE_IDX (%1)"))) value;
698 };
699
700 /* Call this first to get a handle to pass to other functions.
701
702 If you want cpplib to manage its own hashtable, pass in a NULL
703 pointer. Otherwise you should pass in an initialized hash table
704 that cpplib will share; this technique is used by the C front
705 ends. */
706 extern cpp_reader *cpp_create_reader (enum c_lang, struct ht *,
707 struct line_maps *);
708
709 /* Reset the cpp_reader's line_map. This is only used after reading a
710 PCH file. */
711 extern void cpp_set_line_map (cpp_reader *, struct line_maps *);
712
713 /* Call this to change the selected language standard (e.g. because of
714 command line options). */
715 extern void cpp_set_lang (cpp_reader *, enum c_lang);
716
717 /* Set the include paths. */
718 extern void cpp_set_include_chains (cpp_reader *, cpp_dir *, cpp_dir *, int);
719
720 /* Call these to get pointers to the options, callback, and deps
721 structures for a given reader. These pointers are good until you
722 call cpp_finish on that reader. You can either edit the callbacks
723 through the pointer returned from cpp_get_callbacks, or set them
724 with cpp_set_callbacks. */
725 extern cpp_options *cpp_get_options (cpp_reader *);
726 extern cpp_callbacks *cpp_get_callbacks (cpp_reader *);
727 extern void cpp_set_callbacks (cpp_reader *, cpp_callbacks *);
728 extern struct deps *cpp_get_deps (cpp_reader *);
729
730 /* This function reads the file, but does not start preprocessing. It
731 returns the name of the original file; this is the same as the
732 input file, except for preprocessed input. This will generate at
733 least one file change callback, and possibly a line change callback
734 too. If there was an error opening the file, it returns NULL. */
735 extern const char *cpp_read_main_file (cpp_reader *, const char *);
736
737 /* Set up built-ins with special behavior. Use cpp_init_builtins()
738 instead unless your know what you are doing. */
739 extern void cpp_init_special_builtins (cpp_reader *);
740
741 /* Set up built-ins like __FILE__. */
742 extern void cpp_init_builtins (cpp_reader *, int);
743
744 /* This is called after options have been parsed, and partially
745 processed. */
746 extern void cpp_post_options (cpp_reader *);
747
748 /* Set up translation to the target character set. */
749 extern void cpp_init_iconv (cpp_reader *);
750
751 /* Call this to finish preprocessing. If you requested dependency
752 generation, pass an open stream to write the information to,
753 otherwise NULL. It is your responsibility to close the stream. */
754 extern void cpp_finish (cpp_reader *, FILE *deps_stream);
755
756 /* Call this to release the handle at the end of preprocessing. Any
757 use of the handle after this function returns is invalid. */
758 extern void cpp_destroy (cpp_reader *);
759
760 extern unsigned int cpp_token_len (const cpp_token *);
761 extern unsigned char *cpp_token_as_text (cpp_reader *, const cpp_token *);
762 extern unsigned char *cpp_spell_token (cpp_reader *, const cpp_token *,
763 unsigned char *, bool);
764 extern void cpp_register_pragma (cpp_reader *, const char *, const char *,
765 void (*) (cpp_reader *), bool);
766 extern void cpp_register_deferred_pragma (cpp_reader *, const char *,
767 const char *, unsigned, bool, bool);
768 extern int cpp_avoid_paste (cpp_reader *, const cpp_token *,
769 const cpp_token *);
770 extern const cpp_token *cpp_get_token (cpp_reader *);
771 extern const cpp_token *cpp_get_token_with_location (cpp_reader *,
772 source_location *);
773 extern const unsigned char *cpp_macro_definition (cpp_reader *,
774 cpp_hashnode *);
775 extern void _cpp_backup_tokens (cpp_reader *, unsigned int);
776 extern const cpp_token *cpp_peek_token (cpp_reader *, int);
777
778 /* Evaluate a CPP_*CHAR* token. */
779 extern cppchar_t cpp_interpret_charconst (cpp_reader *, const cpp_token *,
780 unsigned int *, int *);
781 /* Evaluate a vector of CPP_*STRING* tokens. */
782 extern bool cpp_interpret_string (cpp_reader *,
783 const cpp_string *, size_t,
784 cpp_string *, enum cpp_ttype);
785 extern bool cpp_interpret_string_notranslate (cpp_reader *,
786 const cpp_string *, size_t,
787 cpp_string *, enum cpp_ttype);
788
789 /* Convert a host character constant to the execution character set. */
790 extern cppchar_t cpp_host_to_exec_charset (cpp_reader *, cppchar_t);
791
792 /* Used to register macros and assertions, perhaps from the command line.
793 The text is the same as the command line argument. */
794 extern void cpp_define (cpp_reader *, const char *);
795 extern void cpp_define_formatted (cpp_reader *pfile,
796 const char *fmt, ...) ATTRIBUTE_PRINTF_2;
797 extern void cpp_assert (cpp_reader *, const char *);
798 extern void cpp_undef (cpp_reader *, const char *);
799 extern void cpp_unassert (cpp_reader *, const char *);
800
801 /* Undefine all macros and assertions. */
802 extern void cpp_undef_all (cpp_reader *);
803
804 extern cpp_buffer *cpp_push_buffer (cpp_reader *, const unsigned char *,
805 size_t, int);
806 extern int cpp_defined (cpp_reader *, const unsigned char *, int);
807
808 /* A preprocessing number. Code assumes that any unused high bits of
809 the double integer are set to zero. */
810 typedef unsigned HOST_WIDE_INT cpp_num_part;
811 typedef struct cpp_num cpp_num;
812 struct cpp_num
813 {
814 cpp_num_part high;
815 cpp_num_part low;
816 bool unsignedp; /* True if value should be treated as unsigned. */
817 bool overflow; /* True if the most recent calculation overflowed. */
818 };
819
820 /* cpplib provides two interfaces for interpretation of preprocessing
821 numbers.
822
823 cpp_classify_number categorizes numeric constants according to
824 their field (integer, floating point, or invalid), radix (decimal,
825 octal, hexadecimal), and type suffixes. */
826
827 #define CPP_N_CATEGORY 0x000F
828 #define CPP_N_INVALID 0x0000
829 #define CPP_N_INTEGER 0x0001
830 #define CPP_N_FLOATING 0x0002
831
832 #define CPP_N_WIDTH 0x00F0
833 #define CPP_N_SMALL 0x0010 /* int, float, shrot _Fract/Accum */
834 #define CPP_N_MEDIUM 0x0020 /* long, double, long _Fract/_Accum. */
835 #define CPP_N_LARGE 0x0040 /* long long, long double,
836 long long _Fract/Accum. */
837
838 #define CPP_N_WIDTH_MD 0xF0000 /* machine defined. */
839 #define CPP_N_MD_W 0x10000
840 #define CPP_N_MD_Q 0x20000
841
842 #define CPP_N_RADIX 0x0F00
843 #define CPP_N_DECIMAL 0x0100
844 #define CPP_N_HEX 0x0200
845 #define CPP_N_OCTAL 0x0400
846 #define CPP_N_BINARY 0x0800
847
848 #define CPP_N_UNSIGNED 0x1000 /* Properties. */
849 #define CPP_N_IMAGINARY 0x2000
850 #define CPP_N_DFLOAT 0x4000
851 #define CPP_N_DEFAULT 0x8000
852
853 #define CPP_N_FRACT 0x100000 /* Fract types. */
854 #define CPP_N_ACCUM 0x200000 /* Accum types. */
855
856 #define CPP_N_USERDEF 0x1000000 /* C++0x user-defined literal. */
857
858 /* Classify a CPP_NUMBER token. The return value is a combination of
859 the flags from the above sets. */
860 extern unsigned cpp_classify_number (cpp_reader *, const cpp_token *,
861 const char **, source_location);
862
863 /* Return the classification flags for a float suffix. */
864 extern unsigned int cpp_interpret_float_suffix (cpp_reader *, const char *,
865 size_t);
866
867 /* Return the classification flags for an int suffix. */
868 extern unsigned int cpp_interpret_int_suffix (cpp_reader *, const char *,
869 size_t);
870
871 /* Evaluate a token classified as category CPP_N_INTEGER. */
872 extern cpp_num cpp_interpret_integer (cpp_reader *, const cpp_token *,
873 unsigned int);
874
875 /* Sign extend a number, with PRECISION significant bits and all
876 others assumed clear, to fill out a cpp_num structure. */
877 cpp_num cpp_num_sign_extend (cpp_num, size_t);
878
879 /* Diagnostic levels. To get a diagnostic without associating a
880 position in the translation unit with it, use cpp_error_with_line
881 with a line number of zero. */
882
883 enum {
884 /* Warning, an error with -Werror. */
885 CPP_DL_WARNING = 0,
886 /* Same as CPP_DL_WARNING, except it is not suppressed in system headers. */
887 CPP_DL_WARNING_SYSHDR,
888 /* Warning, an error with -pedantic-errors or -Werror. */
889 CPP_DL_PEDWARN,
890 /* An error. */
891 CPP_DL_ERROR,
892 /* An internal consistency check failed. Prints "internal error: ",
893 otherwise the same as CPP_DL_ERROR. */
894 CPP_DL_ICE,
895 /* An informative note following a warning. */
896 CPP_DL_NOTE,
897 /* A fatal error. */
898 CPP_DL_FATAL
899 };
900
901 /* Warning reason codes. Use a reason code of zero for unclassified warnings
902 and errors that are not warnings. */
903 enum {
904 CPP_W_NONE = 0,
905 CPP_W_DEPRECATED,
906 CPP_W_COMMENTS,
907 CPP_W_MISSING_INCLUDE_DIRS,
908 CPP_W_TRIGRAPHS,
909 CPP_W_MULTICHAR,
910 CPP_W_TRADITIONAL,
911 CPP_W_LONG_LONG,
912 CPP_W_ENDIF_LABELS,
913 CPP_W_NUM_SIGN_CHANGE,
914 CPP_W_VARIADIC_MACROS,
915 CPP_W_BUILTIN_MACRO_REDEFINED,
916 CPP_W_DOLLARS,
917 CPP_W_UNDEF,
918 CPP_W_UNUSED_MACROS,
919 CPP_W_CXX_OPERATOR_NAMES,
920 CPP_W_NORMALIZE,
921 CPP_W_INVALID_PCH,
922 CPP_W_WARNING_DIRECTIVE,
923 CPP_W_LITERAL_SUFFIX
924 };
925
926 /* Output a diagnostic of some kind. */
927 extern bool cpp_error (cpp_reader *, int, const char *msgid, ...)
928 ATTRIBUTE_PRINTF_3;
929 extern bool cpp_warning (cpp_reader *, int, const char *msgid, ...)
930 ATTRIBUTE_PRINTF_3;
931 extern bool cpp_pedwarning (cpp_reader *, int, const char *msgid, ...)
932 ATTRIBUTE_PRINTF_3;
933 extern bool cpp_warning_syshdr (cpp_reader *, int, const char *msgid, ...)
934 ATTRIBUTE_PRINTF_3;
935
936 /* Output a diagnostic with "MSGID: " preceding the
937 error string of errno. No location is printed. */
938 extern bool cpp_errno (cpp_reader *, int, const char *msgid);
939
940 /* Same as cpp_error, except additionally specifies a position as a
941 (translation unit) physical line and physical column. If the line is
942 zero, then no location is printed. */
943 extern bool cpp_error_with_line (cpp_reader *, int, source_location,
944 unsigned, const char *msgid, ...)
945 ATTRIBUTE_PRINTF_5;
946 extern bool cpp_warning_with_line (cpp_reader *, int, source_location,
947 unsigned, const char *msgid, ...)
948 ATTRIBUTE_PRINTF_5;
949 extern bool cpp_pedwarning_with_line (cpp_reader *, int, source_location,
950 unsigned, const char *msgid, ...)
951 ATTRIBUTE_PRINTF_5;
952 extern bool cpp_warning_with_line_syshdr (cpp_reader *, int, source_location,
953 unsigned, const char *msgid, ...)
954 ATTRIBUTE_PRINTF_5;
955
956 /* In lex.c */
957 extern int cpp_ideq (const cpp_token *, const char *);
958 extern void cpp_output_line (cpp_reader *, FILE *);
959 extern unsigned char *cpp_output_line_to_string (cpp_reader *,
960 const unsigned char *);
961 extern void cpp_output_token (const cpp_token *, FILE *);
962 extern const char *cpp_type2name (enum cpp_ttype, unsigned char flags);
963 /* Returns the value of an escape sequence, truncated to the correct
964 target precision. PSTR points to the input pointer, which is just
965 after the backslash. LIMIT is how much text we have. WIDE is true
966 if the escape sequence is part of a wide character constant or
967 string literal. Handles all relevant diagnostics. */
968 extern cppchar_t cpp_parse_escape (cpp_reader *, const unsigned char ** pstr,
969 const unsigned char *limit, int wide);
970
971 /* Structure used to hold a comment block at a given location in the
972 source code. */
973
974 typedef struct
975 {
976 /* Text of the comment including the terminators. */
977 char *comment;
978
979 /* source location for the given comment. */
980 source_location sloc;
981 } cpp_comment;
982
983 /* Structure holding all comments for a given cpp_reader. */
984
985 typedef struct
986 {
987 /* table of comment entries. */
988 cpp_comment *entries;
989
990 /* number of actual entries entered in the table. */
991 int count;
992
993 /* number of entries allocated currently. */
994 int allocated;
995 } cpp_comment_table;
996
997 /* Returns the table of comments encountered by the preprocessor. This
998 table is only populated when pfile->state.save_comments is true. */
999 extern cpp_comment_table *cpp_get_comments (cpp_reader *);
1000
1001 /* In hash.c */
1002
1003 /* Lookup an identifier in the hashtable. Puts the identifier in the
1004 table if it is not already there. */
1005 extern cpp_hashnode *cpp_lookup (cpp_reader *, const unsigned char *,
1006 unsigned int);
1007
1008 typedef int (*cpp_cb) (cpp_reader *, cpp_hashnode *, void *);
1009 extern void cpp_forall_identifiers (cpp_reader *, cpp_cb, void *);
1010
1011 /* In macro.c */
1012 extern void cpp_scan_nooutput (cpp_reader *);
1013 extern int cpp_sys_macro_p (cpp_reader *);
1014 extern unsigned char *cpp_quote_string (unsigned char *, const unsigned char *,
1015 unsigned int);
1016
1017 /* In files.c */
1018 extern bool cpp_included (cpp_reader *, const char *);
1019 extern bool cpp_included_before (cpp_reader *, const char *, source_location);
1020 extern void cpp_make_system_header (cpp_reader *, int, int);
1021 extern bool cpp_push_include (cpp_reader *, const char *);
1022 extern bool cpp_push_default_include (cpp_reader *, const char *);
1023 extern void cpp_change_file (cpp_reader *, enum lc_reason, const char *);
1024 extern const char *cpp_get_path (struct _cpp_file *);
1025 extern cpp_dir *cpp_get_dir (struct _cpp_file *);
1026 extern cpp_buffer *cpp_get_buffer (cpp_reader *);
1027 extern struct _cpp_file *cpp_get_file (cpp_buffer *);
1028 extern cpp_buffer *cpp_get_prev (cpp_buffer *);
1029 extern void cpp_clear_file_cache (cpp_reader *);
1030
1031 /* In pch.c */
1032 struct save_macro_data;
1033 extern int cpp_save_state (cpp_reader *, FILE *);
1034 extern int cpp_write_pch_deps (cpp_reader *, FILE *);
1035 extern int cpp_write_pch_state (cpp_reader *, FILE *);
1036 extern int cpp_valid_state (cpp_reader *, const char *, int);
1037 extern void cpp_prepare_state (cpp_reader *, struct save_macro_data **);
1038 extern int cpp_read_state (cpp_reader *, const char *, FILE *,
1039 struct save_macro_data *);
1040
1041 /* In lex.c */
1042 extern void cpp_force_token_locations (cpp_reader *, source_location *);
1043 extern void cpp_stop_forcing_token_locations (cpp_reader *);
1044
1045 /* In expr.c */
1046 extern enum cpp_ttype cpp_userdef_string_remove_type
1047 (enum cpp_ttype type);
1048 extern enum cpp_ttype cpp_userdef_string_add_type
1049 (enum cpp_ttype type);
1050 extern enum cpp_ttype cpp_userdef_char_remove_type
1051 (enum cpp_ttype type);
1052 extern enum cpp_ttype cpp_userdef_char_add_type
1053 (enum cpp_ttype type);
1054 extern bool cpp_userdef_string_p
1055 (enum cpp_ttype type);
1056 extern bool cpp_userdef_char_p
1057 (enum cpp_ttype type);
1058 extern const char * cpp_get_userdef_suffix
1059 (const cpp_token *);
1060
1061 #endif /* ! LIBCPP_CPPLIB_H */