1 /* Definitions for CPP library.
2 Copyright (C) 1995-2022 Free Software Foundation, Inc.
3 Written by Per Bothner, 1994-95.
4
5 This program is free software; you can redistribute it and/or modify it
6 under the terms of the GNU General Public License as published by the
7 Free Software Foundation; either version 3, or (at your option) any
8 later version.
9
10 This program is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
14
15 You should have received a copy of the GNU General Public License
16 along with this program; see the file COPYING3. If not see
17 <http://www.gnu.org/licenses/>.
18
19 In other words, you are welcome to use, share and improve this program.
20 You are forbidden to forbid anyone else to use, share and improve
21 what you give them. Help stamp out software-hoarding! */
22 #ifndef LIBCPP_CPPLIB_H
23 #define LIBCPP_CPPLIB_H
24
25 #include <sys/types.h>
26 #include "symtab.h"
27 #include "line-map.h"
28
29 typedef struct cpp_reader cpp_reader;
30 typedef struct cpp_buffer cpp_buffer;
31 typedef struct cpp_options cpp_options;
32 typedef struct cpp_token cpp_token;
33 typedef struct cpp_string cpp_string;
34 typedef struct cpp_hashnode cpp_hashnode;
35 typedef struct cpp_macro cpp_macro;
36 typedef struct cpp_callbacks cpp_callbacks;
37 typedef struct cpp_dir cpp_dir;
38
39 struct _cpp_file;
40
41 /* The first three groups, apart from '=', can appear in preprocessor
42 expressions (+= and -= are used to indicate unary + and - resp.).
43 This allows a lookup table to be implemented in _cpp_parse_expr.
44
45 The first group, to CPP_LAST_EQ, can be immediately followed by an
46 '='. The lexer needs operators ending in '=', like ">>=", to be in
47 the same order as their counterparts without the '=', like ">>".
48
49 See the cpp_operator table optab in expr.cc if you change the order or
50 add or remove anything in the first group. */
51
52 #define TTYPE_TABLE \
53 OP(EQ, "=") \
54 OP(NOT, "!") \
55 OP(GREATER, ">") /* compare */ \
56 OP(LESS, "<") \
57 OP(PLUS, "+") /* math */ \
58 OP(MINUS, "-") \
59 OP(MULT, "*") \
60 OP(DIV, "/") \
61 OP(MOD, "%") \
62 OP(AND, "&") /* bit ops */ \
63 OP(OR, "|") \
64 OP(XOR, "^") \
65 OP(RSHIFT, ">>") \
66 OP(LSHIFT, "<<") \
67 \
68 OP(COMPL, "~") \
69 OP(AND_AND, "&&") /* logical */ \
70 OP(OR_OR, "||") \
71 OP(QUERY, "?") \
72 OP(COLON, ":") \
73 OP(COMMA, ",") /* grouping */ \
74 OP(OPEN_PAREN, "(") \
75 OP(CLOSE_PAREN, ")") \
76 TK(EOF, NONE) \
77 OP(EQ_EQ, "==") /* compare */ \
78 OP(NOT_EQ, "!=") \
79 OP(GREATER_EQ, ">=") \
80 OP(LESS_EQ, "<=") \
81 OP(SPACESHIP, "<=>") \
82 \
83 /* These two are unary + / - in preprocessor expressions. */ \
84 OP(PLUS_EQ, "+=") /* math */ \
85 OP(MINUS_EQ, "-=") \
86 \
87 OP(MULT_EQ, "*=") \
88 OP(DIV_EQ, "/=") \
89 OP(MOD_EQ, "%=") \
90 OP(AND_EQ, "&=") /* bit ops */ \
91 OP(OR_EQ, "|=") \
92 OP(XOR_EQ, "^=") \
93 OP(RSHIFT_EQ, ">>=") \
94 OP(LSHIFT_EQ, "<<=") \
95 /* Digraphs together, beginning with CPP_FIRST_DIGRAPH. */ \
96 OP(HASH, "#") /* digraphs */ \
97 OP(PASTE, "##") \
98 OP(OPEN_SQUARE, "[") \
99 OP(CLOSE_SQUARE, "]") \
100 OP(OPEN_BRACE, "{") \
101 OP(CLOSE_BRACE, "}") \
102 /* The remainder of the punctuation. Order is not significant. */ \
103 OP(SEMICOLON, ";") /* structure */ \
104 OP(ELLIPSIS, "...") \
105 OP(PLUS_PLUS, "++") /* increment */ \
106 OP(MINUS_MINUS, "--") \
107 OP(DEREF, "->") /* accessors */ \
108 OP(DOT, ".") \
109 OP(SCOPE, "::") \
110 OP(DEREF_STAR, "->*") \
111 OP(DOT_STAR, ".*") \
112 OP(ATSIGN, "@") /* used in Objective-C */ \
113 \
114 TK(NAME, IDENT) /* word */ \
115 TK(AT_NAME, IDENT) /* @word - Objective-C */ \
116 TK(NUMBER, LITERAL) /* 34_be+ta */ \
117 \
118 TK(CHAR, LITERAL) /* 'char' */ \
119 TK(WCHAR, LITERAL) /* L'char' */ \
120 TK(CHAR16, LITERAL) /* u'char' */ \
121 TK(CHAR32, LITERAL) /* U'char' */ \
122 TK(UTF8CHAR, LITERAL) /* u8'char' */ \
123 TK(OTHER, LITERAL) /* stray punctuation */ \
124 \
125 TK(STRING, LITERAL) /* "string" */ \
126 TK(WSTRING, LITERAL) /* L"string" */ \
127 TK(STRING16, LITERAL) /* u"string" */ \
128 TK(STRING32, LITERAL) /* U"string" */ \
129 TK(UTF8STRING, LITERAL) /* u8"string" */ \
130 TK(OBJC_STRING, LITERAL) /* @"string" - Objective-C */ \
131 TK(HEADER_NAME, LITERAL) /* <stdio.h> in #include */ \
132 \
133 TK(CHAR_USERDEF, LITERAL) /* 'char'_suffix - C++-0x */ \
134 TK(WCHAR_USERDEF, LITERAL) /* L'char'_suffix - C++-0x */ \
135 TK(CHAR16_USERDEF, LITERAL) /* u'char'_suffix - C++-0x */ \
136 TK(CHAR32_USERDEF, LITERAL) /* U'char'_suffix - C++-0x */ \
137 TK(UTF8CHAR_USERDEF, LITERAL) /* u8'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 /* A token type for keywords, as opposed to ordinary identifiers. */
159 CPP_KEYWORD,
160
161 /* Positions in the table. */
162 CPP_LAST_EQ = CPP_LSHIFT,
163 CPP_FIRST_DIGRAPH = CPP_HASH,
164 CPP_LAST_PUNCTUATOR= CPP_ATSIGN,
165 CPP_LAST_CPP_OP = CPP_LESS_EQ
166 };
167 #undef OP
168 #undef TK
169
170 /* C language kind, used when calling cpp_create_reader. */
171 enum c_lang {CLK_GNUC89 = 0, CLK_GNUC99, CLK_GNUC11, CLK_GNUC17, CLK_GNUC2X,
172 CLK_STDC89, CLK_STDC94, CLK_STDC99, CLK_STDC11, CLK_STDC17,
173 CLK_STDC2X,
174 CLK_GNUCXX, CLK_CXX98, CLK_GNUCXX11, CLK_CXX11,
175 CLK_GNUCXX14, CLK_CXX14, CLK_GNUCXX17, CLK_CXX17,
176 CLK_GNUCXX20, CLK_CXX20, CLK_GNUCXX23, CLK_CXX23,
177 CLK_ASM};
178
179 /* Payload of a NUMBER, STRING, CHAR or COMMENT token. */
180 struct GTY(()) cpp_string {
181 unsigned int len;
182 const unsigned char *text;
183 };
184
185 /* Flags for the cpp_token structure. */
186 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */
187 #define DIGRAPH (1 << 1) /* If it was a digraph. */
188 #define STRINGIFY_ARG (1 << 2) /* If macro argument to be stringified. */
189 #define PASTE_LEFT (1 << 3) /* If on LHS of a ## operator. */
190 #define NAMED_OP (1 << 4) /* C++ named operators. */
191 #define PREV_FALLTHROUGH (1 << 5) /* On a token preceeded by FALLTHROUGH
192 comment. */
193 #define BOL (1 << 6) /* Token at beginning of line. */
194 #define PURE_ZERO (1 << 7) /* Single 0 digit, used by the C++ frontend,
195 set in c-lex.cc. */
196 #define COLON_SCOPE PURE_ZERO /* Adjacent colons in C < 23. */
197 #define SP_DIGRAPH (1 << 8) /* # or ## token was a digraph. */
198 #define SP_PREV_WHITE (1 << 9) /* If whitespace before a ##
199 operator, or before this token
200 after a # operator. */
201 #define NO_EXPAND (1 << 10) /* Do not macro-expand this token. */
202 #define PRAGMA_OP (1 << 11) /* _Pragma token. */
203
204 /* Specify which field, if any, of the cpp_token union is used. */
205
206 enum cpp_token_fld_kind {
207 CPP_TOKEN_FLD_NODE,
208 CPP_TOKEN_FLD_SOURCE,
209 CPP_TOKEN_FLD_STR,
210 CPP_TOKEN_FLD_ARG_NO,
211 CPP_TOKEN_FLD_TOKEN_NO,
212 CPP_TOKEN_FLD_PRAGMA,
213 CPP_TOKEN_FLD_NONE
214 };
215
216 /* A macro argument in the cpp_token union. */
217 struct GTY(()) cpp_macro_arg {
218 /* Argument number. */
219 unsigned int arg_no;
220 /* The original spelling of the macro argument token. */
221 cpp_hashnode *
222 GTY ((nested_ptr (union tree_node,
223 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
224 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL")))
225 spelling;
226 };
227
228 /* An identifier in the cpp_token union. */
229 struct GTY(()) cpp_identifier {
230 /* The canonical (UTF-8) spelling of the identifier. */
231 cpp_hashnode *
232 GTY ((nested_ptr (union tree_node,
233 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
234 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL")))
235 node;
236 /* The original spelling of the identifier. */
237 cpp_hashnode *
238 GTY ((nested_ptr (union tree_node,
239 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
240 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL")))
241 spelling;
242 };
243
244 /* A preprocessing token. This has been carefully packed and should
245 occupy 16 bytes on 32-bit hosts and 24 bytes on 64-bit hosts. */
246 struct GTY(()) cpp_token {
247
248 /* Location of first char of token, together with range of full token. */
249 location_t src_loc;
250
251 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
252 unsigned short flags; /* flags - see above */
253
254 union cpp_token_u
255 {
256 /* An identifier. */
257 struct cpp_identifier GTY ((tag ("CPP_TOKEN_FLD_NODE"))) node;
258
259 /* Inherit padding from this token. */
260 cpp_token * GTY ((tag ("CPP_TOKEN_FLD_SOURCE"))) source;
261
262 /* A string, or number. */
263 struct cpp_string GTY ((tag ("CPP_TOKEN_FLD_STR"))) str;
264
265 /* Argument no. (and original spelling) for a CPP_MACRO_ARG. */
266 struct cpp_macro_arg GTY ((tag ("CPP_TOKEN_FLD_ARG_NO"))) macro_arg;
267
268 /* Original token no. for a CPP_PASTE (from a sequence of
269 consecutive paste tokens in a macro expansion). */
270 unsigned int GTY ((tag ("CPP_TOKEN_FLD_TOKEN_NO"))) token_no;
271
272 /* Caller-supplied identifier for a CPP_PRAGMA. */
273 unsigned int GTY ((tag ("CPP_TOKEN_FLD_PRAGMA"))) pragma;
274 } GTY ((desc ("cpp_token_val_index (&%1)"))) val;
275 };
276
277 /* Say which field is in use. */
278 extern enum cpp_token_fld_kind cpp_token_val_index (const cpp_token *tok);
279
280 /* A type wide enough to hold any multibyte source character.
281 cpplib's character constant interpreter requires an unsigned type.
282 Also, a typedef for the signed equivalent.
283 The width of this type is capped at 32 bits; there do exist targets
284 where wchar_t is 64 bits, but only in a non-default mode, and there
285 would be no meaningful interpretation for a wchar_t value greater
286 than 2^32 anyway -- the widest wide-character encoding around is
287 ISO 10646, which stops at 2^31. */
288 #if CHAR_BIT * SIZEOF_INT >= 32
289 # define CPPCHAR_SIGNED_T int
290 #elif CHAR_BIT * SIZEOF_LONG >= 32
291 # define CPPCHAR_SIGNED_T long
292 #else
293 # error "Cannot find a least-32-bit signed integer type"
294 #endif
295 typedef unsigned CPPCHAR_SIGNED_T cppchar_t;
296 typedef CPPCHAR_SIGNED_T cppchar_signed_t;
297
298 /* Style of header dependencies to generate. */
299 enum cpp_deps_style { DEPS_NONE = 0, DEPS_USER, DEPS_SYSTEM };
300
301 /* The possible normalization levels, from most restrictive to least. */
302 enum cpp_normalize_level {
303 /* In NFKC. */
304 normalized_KC = 0,
305 /* In NFC. */
306 normalized_C,
307 /* In NFC, except for subsequences where being in NFC would make
308 the identifier invalid. */
309 normalized_identifier_C,
310 /* Not normalized at all. */
311 normalized_none
312 };
313
314 enum cpp_main_search
315 {
316 CMS_none, /* A regular source file. */
317 CMS_header, /* Is a directly-specified header file (eg PCH or
318 header-unit). */
319 CMS_user, /* Search the user INCLUDE path. */
320 CMS_system, /* Search the system INCLUDE path. */
321 };
322
323 /* The possible bidirectional control characters checking levels. */
324 enum cpp_bidirectional_level {
325 /* No checking. */
326 bidirectional_none = 0,
327 /* Only detect unpaired uses of bidirectional control characters. */
328 bidirectional_unpaired = 1,
329 /* Detect any use of bidirectional control characters. */
330 bidirectional_any = 2,
331 /* Also warn about UCNs. */
332 bidirectional_ucn = 4
333 };
334
335 /* This structure is nested inside struct cpp_reader, and
336 carries all the options visible to the command line. */
337 struct cpp_options
338 {
339 /* The language we're preprocessing. */
340 enum c_lang lang;
341
342 /* Nonzero means use extra default include directories for C++. */
343 unsigned char cplusplus;
344
345 /* Nonzero means handle cplusplus style comments. */
346 unsigned char cplusplus_comments;
347
348 /* Nonzero means define __OBJC__, treat @ as a special token, use
349 the OBJC[PLUS]_INCLUDE_PATH environment variable, and allow
350 "#import". */
351 unsigned char objc;
352
353 /* Nonzero means don't copy comments into the output file. */
354 unsigned char discard_comments;
355
356 /* Nonzero means don't copy comments into the output file during
357 macro expansion. */
358 unsigned char discard_comments_in_macro_exp;
359
360 /* Nonzero means process the ISO trigraph sequences. */
361 unsigned char trigraphs;
362
363 /* Nonzero means process the ISO digraph sequences. */
364 unsigned char digraphs;
365
366 /* Nonzero means to allow hexadecimal floats and LL suffixes. */
367 unsigned char extended_numbers;
368
369 /* Nonzero means process u/U prefix literals (UTF-16/32). */
370 unsigned char uliterals;
371
372 /* Nonzero means process u8 prefixed character literals (UTF-8). */
373 unsigned char utf8_char_literals;
374
375 /* Nonzero means process r/R raw strings. If this is set, uliterals
376 must be set as well. */
377 unsigned char rliterals;
378
379 /* Nonzero means print names of header files (-H). */
380 unsigned char print_include_names;
381
382 /* Nonzero means complain about deprecated features. */
383 unsigned char cpp_warn_deprecated;
384
385 /* Nonzero means warn if slash-star appears in a comment. */
386 unsigned char warn_comments;
387
388 /* Nonzero means to warn about __DATA__, __TIME__ and __TIMESTAMP__ usage. */
389 unsigned char warn_date_time;
390
391 /* Nonzero means warn if a user-supplied include directory does not
392 exist. */
393 unsigned char warn_missing_include_dirs;
394
395 /* Nonzero means warn if there are any trigraphs. */
396 unsigned char warn_trigraphs;
397
398 /* Nonzero means warn about multicharacter charconsts. */
399 unsigned char warn_multichar;
400
401 /* Nonzero means warn about various incompatibilities with
402 traditional C. */
403 unsigned char cpp_warn_traditional;
404
405 /* Nonzero means warn about long long numeric constants. */
406 unsigned char cpp_warn_long_long;
407
408 /* Nonzero means warn about text after an #endif (or #else). */
409 unsigned char warn_endif_labels;
410
411 /* Nonzero means warn about implicit sign changes owing to integer
412 promotions. */
413 unsigned char warn_num_sign_change;
414
415 /* Zero means don't warn about __VA_ARGS__ usage in c89 pedantic mode.
416 Presumably the usage is protected by the appropriate #ifdef. */
417 unsigned char warn_variadic_macros;
418
419 /* Nonzero means warn about builtin macros that are redefined or
420 explicitly undefined. */
421 unsigned char warn_builtin_macro_redefined;
422
423 /* Different -Wimplicit-fallthrough= levels. */
424 unsigned char cpp_warn_implicit_fallthrough;
425
426 /* Nonzero means we should look for header.gcc files that remap file
427 names. */
428 unsigned char remap;
429
430 /* Zero means dollar signs are punctuation. */
431 unsigned char dollars_in_ident;
432
433 /* Nonzero means UCNs are accepted in identifiers. */
434 unsigned char extended_identifiers;
435
436 /* True if we should warn about dollars in identifiers or numbers
437 for this translation unit. */
438 unsigned char warn_dollars;
439
440 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
441 unsigned char warn_undef;
442
443 /* Nonzero means warn if "defined" is encountered in a place other than
444 an #if. */
445 unsigned char warn_expansion_to_defined;
446
447 /* Nonzero means warn of unused macros from the main file. */
448 unsigned char warn_unused_macros;
449
450 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
451 unsigned char c99;
452
453 /* Nonzero if we are conforming to a specific C or C++ standard. */
454 unsigned char std;
455
456 /* Nonzero means give all the error messages the ANSI standard requires. */
457 unsigned char cpp_pedantic;
458
459 /* Nonzero means we're looking at already preprocessed code, so don't
460 bother trying to do macro expansion and whatnot. */
461 unsigned char preprocessed;
462
463 /* Nonzero means we are going to emit debugging logs during
464 preprocessing. */
465 unsigned char debug;
466
467 /* Nonzero means we are tracking locations of tokens involved in
468 macro expansion. 1 Means we track the location in degraded mode
469 where we do not track locations of tokens resulting from the
470 expansion of arguments of function-like macro. 2 Means we do
471 track all macro expansions. This last option is the one that
472 consumes the highest amount of memory. */
473 unsigned char track_macro_expansion;
474
475 /* Nonzero means handle C++ alternate operator names. */
476 unsigned char operator_names;
477
478 /* Nonzero means warn about use of C++ alternate operator names. */
479 unsigned char warn_cxx_operator_names;
480
481 /* True for traditional preprocessing. */
482 unsigned char traditional;
483
484 /* Nonzero for C++ 2011 Standard user-defined literals. */
485 unsigned char user_literals;
486
487 /* Nonzero means warn when a string or character literal is followed by a
488 ud-suffix which does not beging with an underscore. */
489 unsigned char warn_literal_suffix;
490
491 /* Nonzero means interpret imaginary, fixed-point, or other gnu extension
492 literal number suffixes as user-defined literal number suffixes. */
493 unsigned char ext_numeric_literals;
494
495 /* Nonzero means extended identifiers allow the characters specified
496 in C11. */
497 unsigned char c11_identifiers;
498
499 /* Nonzero for C++ 2014 Standard binary constants. */
500 unsigned char binary_constants;
501
502 /* Nonzero for C++ 2014 Standard digit separators. */
503 unsigned char digit_separators;
504
505 /* Nonzero for C2X decimal floating-point constants. */
506 unsigned char dfp_constants;
507
508 /* Nonzero for C++20 __VA_OPT__ feature. */
509 unsigned char va_opt;
510
511 /* Nonzero for the '::' token. */
512 unsigned char scope;
513
514 /* Nonzero for the '#elifdef' and '#elifndef' directives. */
515 unsigned char elifdef;
516
517 /* Nonzero means tokenize C++20 module directives. */
518 unsigned char module_directives;
519
520 /* Nonzero for C++23 size_t literals. */
521 unsigned char size_t_literals;
522
523 /* Holds the name of the target (execution) character set. */
524 const char *narrow_charset;
525
526 /* Holds the name of the target wide character set. */
527 const char *wide_charset;
528
529 /* Holds the name of the input character set. */
530 const char *input_charset;
531
532 /* The minimum permitted level of normalization before a warning
533 is generated. See enum cpp_normalize_level. */
534 int warn_normalize;
535
536 /* True to warn about precompiled header files we couldn't use. */
537 bool warn_invalid_pch;
538
539 /* True if dependencies should be restored from a precompiled header. */
540 bool restore_pch_deps;
541
542 /* True if warn about differences between C90 and C99. */
543 signed char cpp_warn_c90_c99_compat;
544
545 /* True if warn about differences between C11 and C2X. */
546 signed char cpp_warn_c11_c2x_compat;
547
548 /* True if warn about differences between C++98 and C++11. */
549 bool cpp_warn_cxx11_compat;
550
551 /* Nonzero if bidirectional control characters checking is on. See enum
552 cpp_bidirectional_level. */
553 unsigned char cpp_warn_bidirectional;
554
555 /* Dependency generation. */
556 struct
557 {
558 /* Style of header dependencies to generate. */
559 enum cpp_deps_style style;
560
561 /* Assume missing files are generated files. */
562 bool missing_files;
563
564 /* Generate phony targets for each dependency apart from the first
565 one. */
566 bool phony_targets;
567
568 /* Generate dependency info for modules. */
569 bool modules;
570
571 /* If true, no dependency is generated on the main file. */
572 bool ignore_main_file;
573
574 /* If true, intend to use the preprocessor output (e.g., for compilation)
575 in addition to the dependency info. */
576 bool need_preprocessor_output;
577 } deps;
578
579 /* Target-specific features set by the front end or client. */
580
581 /* Precision for target CPP arithmetic, target characters, target
582 ints and target wide characters, respectively. */
583 size_t precision, char_precision, int_precision, wchar_precision;
584
585 /* True means chars (wide chars) are unsigned. */
586 bool unsigned_char, unsigned_wchar;
587
588 /* True if the most significant byte in a word has the lowest
589 address in memory. */
590 bool bytes_big_endian;
591
592 /* Nonzero means __STDC__ should have the value 0 in system headers. */
593 unsigned char stdc_0_in_system_headers;
594
595 /* True disables tokenization outside of preprocessing directives. */
596 bool directives_only;
597
598 /* True enables canonicalization of system header file paths. */
599 bool canonical_system_headers;
600
601 /* The maximum depth of the nested #include. */
602 unsigned int max_include_depth;
603
604 cpp_main_search main_search : 8;
605 };
606
607 /* Diagnostic levels. To get a diagnostic without associating a
608 position in the translation unit with it, use cpp_error_with_line
609 with a line number of zero. */
610
611 enum cpp_diagnostic_level {
612 /* Warning, an error with -Werror. */
613 CPP_DL_WARNING = 0,
614 /* Same as CPP_DL_WARNING, except it is not suppressed in system headers. */
615 CPP_DL_WARNING_SYSHDR,
616 /* Warning, an error with -pedantic-errors or -Werror. */
617 CPP_DL_PEDWARN,
618 /* An error. */
619 CPP_DL_ERROR,
620 /* An internal consistency check failed. Prints "internal error: ",
621 otherwise the same as CPP_DL_ERROR. */
622 CPP_DL_ICE,
623 /* An informative note following a warning. */
624 CPP_DL_NOTE,
625 /* A fatal error. */
626 CPP_DL_FATAL
627 };
628
629 /* Warning reason codes. Use a reason code of CPP_W_NONE for unclassified
630 warnings and diagnostics that are not warnings. */
631
632 enum cpp_warning_reason {
633 CPP_W_NONE = 0,
634 CPP_W_DEPRECATED,
635 CPP_W_COMMENTS,
636 CPP_W_MISSING_INCLUDE_DIRS,
637 CPP_W_TRIGRAPHS,
638 CPP_W_MULTICHAR,
639 CPP_W_TRADITIONAL,
640 CPP_W_LONG_LONG,
641 CPP_W_ENDIF_LABELS,
642 CPP_W_NUM_SIGN_CHANGE,
643 CPP_W_VARIADIC_MACROS,
644 CPP_W_BUILTIN_MACRO_REDEFINED,
645 CPP_W_DOLLARS,
646 CPP_W_UNDEF,
647 CPP_W_UNUSED_MACROS,
648 CPP_W_CXX_OPERATOR_NAMES,
649 CPP_W_NORMALIZE,
650 CPP_W_INVALID_PCH,
651 CPP_W_WARNING_DIRECTIVE,
652 CPP_W_LITERAL_SUFFIX,
653 CPP_W_SIZE_T_LITERALS,
654 CPP_W_DATE_TIME,
655 CPP_W_PEDANTIC,
656 CPP_W_C90_C99_COMPAT,
657 CPP_W_C11_C2X_COMPAT,
658 CPP_W_CXX11_COMPAT,
659 CPP_W_EXPANSION_TO_DEFINED,
660 CPP_W_BIDIRECTIONAL
661 };
662
663 /* Callback for header lookup for HEADER, which is the name of a
664 source file. It is used as a method of last resort to find headers
665 that are not otherwise found during the normal include processing.
666 The return value is the malloced name of a header to try and open,
667 if any, or NULL otherwise. This callback is called only if the
668 header is otherwise unfound. */
669 typedef const char *(*missing_header_cb)(cpp_reader *, const char *header, cpp_dir **);
670
671 /* Call backs to cpplib client. */
672 struct cpp_callbacks
673 {
674 /* Called when a new line of preprocessed output is started. */
675 void (*line_change) (cpp_reader *, const cpp_token *, int);
676
677 /* Called when switching to/from a new file.
678 The line_map is for the new file. It is NULL if there is no new file.
679 (In C this happens when done with <built-in>+<command line> and also
680 when done with a main file.) This can be used for resource cleanup. */
681 void (*file_change) (cpp_reader *, const line_map_ordinary *);
682
683 void (*dir_change) (cpp_reader *, const char *);
684 void (*include) (cpp_reader *, location_t, const unsigned char *,
685 const char *, int, const cpp_token **);
686 void (*define) (cpp_reader *, location_t, cpp_hashnode *);
687 void (*undef) (cpp_reader *, location_t, cpp_hashnode *);
688 void (*ident) (cpp_reader *, location_t, const cpp_string *);
689 void (*def_pragma) (cpp_reader *, location_t);
690 int (*valid_pch) (cpp_reader *, const char *, int);
691 void (*read_pch) (cpp_reader *, const char *, int, const char *);
692 missing_header_cb missing_header;
693
694 /* Context-sensitive macro support. Returns macro (if any) that should
695 be expanded. */
696 cpp_hashnode * (*macro_to_expand) (cpp_reader *, const cpp_token *);
697
698 /* Called to emit a diagnostic. This callback receives the
699 translated message. */
700 bool (*diagnostic) (cpp_reader *,
701 enum cpp_diagnostic_level,
702 enum cpp_warning_reason,
703 rich_location *,
704 const char *, va_list *)
705 ATTRIBUTE_FPTR_PRINTF(5,0);
706
707 /* Callbacks for when a macro is expanded, or tested (whether
708 defined or not at the time) in #ifdef, #ifndef or "defined". */
709 void (*used_define) (cpp_reader *, location_t, cpp_hashnode *);
710 void (*used_undef) (cpp_reader *, location_t, cpp_hashnode *);
711 /* Called before #define and #undef or other macro definition
712 changes are processed. */
713 void (*before_define) (cpp_reader *);
714 /* Called whenever a macro is expanded or tested.
715 Second argument is the location of the start of the current expansion. */
716 void (*used) (cpp_reader *, location_t, cpp_hashnode *);
717
718 /* Callback to identify whether an attribute exists. */
719 int (*has_attribute) (cpp_reader *, bool);
720
721 /* Callback to determine whether a built-in function is recognized. */
722 int (*has_builtin) (cpp_reader *);
723
724 /* Callback that can change a user lazy into normal macro. */
725 void (*user_lazy_macro) (cpp_reader *, cpp_macro *, unsigned);
726
727 /* Callback to handle deferred cpp_macros. */
728 cpp_macro *(*user_deferred_macro) (cpp_reader *, location_t, cpp_hashnode *);
729
730 /* Callback to parse SOURCE_DATE_EPOCH from environment. */
731 time_t (*get_source_date_epoch) (cpp_reader *);
732
733 /* Callback for providing suggestions for misspelled directives. */
734 const char *(*get_suggestion) (cpp_reader *, const char *, const char *const *);
735
736 /* Callback for when a comment is encountered, giving the location
737 of the opening slash, a pointer to the content (which is not
738 necessarily 0-terminated), and the length of the content.
739 The content contains the opening slash-star (or slash-slash),
740 and for C-style comments contains the closing star-slash. For
741 C++-style comments it does not include the terminating newline. */
742 void (*comment) (cpp_reader *, location_t, const unsigned char *,
743 size_t);
744
745 /* Callback for filename remapping in __FILE__ and __BASE_FILE__ macro
746 expansions. */
747 const char *(*remap_filename) (const char*);
748
749 /* Maybe translate a #include into something else. Return a
750 cpp_buffer containing the translation if translating. */
751 char *(*translate_include) (cpp_reader *, line_maps *, location_t,
752 const char *path);
753 };
754
755 #ifdef VMS
756 #define INO_T_CPP ino_t ino[3]
757 #elif defined (_AIX) && SIZEOF_INO_T == 4
758 #define INO_T_CPP ino64_t ino
759 #else
760 #define INO_T_CPP ino_t ino
761 #endif
762
763 #if defined (_AIX) && SIZEOF_DEV_T == 4
764 #define DEV_T_CPP dev64_t dev
765 #else
766 #define DEV_T_CPP dev_t dev
767 #endif
768
769 /* Chain of directories to look for include files in. */
770 struct cpp_dir
771 {
772 /* NULL-terminated singly-linked list. */
773 struct cpp_dir *next;
774
775 /* NAME of the directory, NUL-terminated. */
776 char *name;
777 unsigned int len;
778
779 /* One if a system header, two if a system header that has extern
780 "C" guards for C++. */
781 unsigned char sysp;
782
783 /* Is this a user-supplied directory? */
784 bool user_supplied_p;
785
786 /* The canonicalized NAME as determined by lrealpath. This field
787 is only used by hosts that lack reliable inode numbers. */
788 char *canonical_name;
789
790 /* Mapping of file names for this directory for MS-DOS and related
791 platforms. A NULL-terminated array of (from, to) pairs. */
792 const char **name_map;
793
794 /* Routine to construct pathname, given the search path name and the
795 HEADER we are trying to find, return a constructed pathname to
796 try and open. If this is NULL, the constructed pathname is as
797 constructed by append_file_to_dir. */
798 char *(*construct) (const char *header, cpp_dir *dir);
799
800 /* The C front end uses these to recognize duplicated
801 directories in the search path. */
802 INO_T_CPP;
803 DEV_T_CPP;
804 };
805
806 /* The kind of the cpp_macro. */
807 enum cpp_macro_kind {
808 cmk_macro, /* An ISO macro (token expansion). */
809 cmk_assert, /* An assertion. */
810 cmk_traditional /* A traditional macro (text expansion). */
811 };
812
813 /* Each macro definition is recorded in a cpp_macro structure.
814 Variadic macros cannot occur with traditional cpp. */
815 struct GTY(()) cpp_macro {
816 union cpp_parm_u
817 {
818 /* Parameters, if any. If parameter names use extended identifiers,
819 the original spelling of those identifiers, not the canonical
820 UTF-8 spelling, goes here. */
821 cpp_hashnode ** GTY ((tag ("false"),
822 nested_ptr (union tree_node,
823 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
824 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL"),
825 length ("%1.paramc"))) params;
826
827 /* If this is an assertion, the next one in the chain. */
828 cpp_macro *GTY ((tag ("true"))) next;
829 } GTY ((desc ("%1.kind == cmk_assert"))) parm;
830
831 /* Definition line number. */
832 location_t line;
833
834 /* Number of tokens in body, or bytes for traditional macros. */
835 /* Do we really need 2^32-1 range here? */
836 unsigned int count;
837
838 /* Number of parameters. */
839 unsigned short paramc;
840
841 /* Non-zero if this is a user-lazy macro, value provided by user. */
842 unsigned char lazy;
843
844 /* The kind of this macro (ISO, trad or assert) */
845 unsigned kind : 2;
846
847 /* If a function-like macro. */
848 unsigned int fun_like : 1;
849
850 /* If a variadic macro. */
851 unsigned int variadic : 1;
852
853 /* If macro defined in system header. */
854 unsigned int syshdr : 1;
855
856 /* Nonzero if it has been expanded or had its existence tested. */
857 unsigned int used : 1;
858
859 /* Indicate whether the tokens include extra CPP_PASTE tokens at the
860 end to track invalid redefinitions with consecutive CPP_PASTE
861 tokens. */
862 unsigned int extra_tokens : 1;
863
864 /* Imported C++20 macro (from a header unit). */
865 unsigned int imported_p : 1;
866
867 /* 0 bits spare (32-bit). 32 on 64-bit target. */
868
869 union cpp_exp_u
870 {
871 /* Trailing array of replacement tokens (ISO), or assertion body value. */
872 cpp_token GTY ((tag ("false"), length ("%1.count"))) tokens[1];
873
874 /* Pointer to replacement text (traditional). See comment at top
875 of cpptrad.c for how traditional function-like macros are
876 encoded. */
877 const unsigned char *GTY ((tag ("true"))) text;
878 } GTY ((desc ("%1.kind == cmk_traditional"))) exp;
879 };
880
881 /* Poisoned identifiers are flagged NODE_POISONED. NODE_OPERATOR (C++
882 only) indicates an identifier that behaves like an operator such as
883 "xor". NODE_DIAGNOSTIC is for speed in lex_token: it indicates a
884 diagnostic may be required for this node. Currently this only
885 applies to __VA_ARGS__, poisoned identifiers, and -Wc++-compat
886 warnings about NODE_OPERATOR. */
887
888 /* Hash node flags. */
889 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */
890 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */
891 #define NODE_DIAGNOSTIC (1 << 2) /* Possible diagnostic when lexed. */
892 #define NODE_WARN (1 << 3) /* Warn if redefined or undefined. */
893 #define NODE_DISABLED (1 << 4) /* A disabled macro. */
894 #define NODE_USED (1 << 5) /* Dumped with -dU. */
895 #define NODE_CONDITIONAL (1 << 6) /* Conditional macro */
896 #define NODE_WARN_OPERATOR (1 << 7) /* Warn about C++ named operator. */
897 #define NODE_MODULE (1 << 8) /* C++-20 module-related name. */
898
899 /* Different flavors of hash node. */
900 enum node_type
901 {
902 NT_VOID = 0, /* Maybe an assert? */
903 NT_MACRO_ARG, /* A macro arg. */
904 NT_USER_MACRO, /* A user macro. */
905 NT_BUILTIN_MACRO, /* A builtin macro. */
906 NT_MACRO_MASK = NT_USER_MACRO /* Mask for either macro kind. */
907 };
908
909 /* Different flavors of builtin macro. _Pragma is an operator, but we
910 handle it with the builtin code for efficiency reasons. */
911 enum cpp_builtin_type
912 {
913 BT_SPECLINE = 0, /* `__LINE__' */
914 BT_DATE, /* `__DATE__' */
915 BT_FILE, /* `__FILE__' */
916 BT_FILE_NAME, /* `__FILE_NAME__' */
917 BT_BASE_FILE, /* `__BASE_FILE__' */
918 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
919 BT_TIME, /* `__TIME__' */
920 BT_STDC, /* `__STDC__' */
921 BT_PRAGMA, /* `_Pragma' operator */
922 BT_TIMESTAMP, /* `__TIMESTAMP__' */
923 BT_COUNTER, /* `__COUNTER__' */
924 BT_HAS_ATTRIBUTE, /* `__has_attribute(x)' */
925 BT_HAS_STD_ATTRIBUTE, /* `__has_c_attribute(x)' */
926 BT_HAS_BUILTIN, /* `__has_builtin(x)' */
927 BT_HAS_INCLUDE, /* `__has_include(x)' */
928 BT_HAS_INCLUDE_NEXT /* `__has_include_next(x)' */
929 };
930
931 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE))
932 #define HT_NODE(NODE) (&(NODE)->ident)
933 #define NODE_LEN(NODE) HT_LEN (HT_NODE (NODE))
934 #define NODE_NAME(NODE) HT_STR (HT_NODE (NODE))
935
936 /* The common part of an identifier node shared amongst all 3 C front
937 ends. Also used to store CPP identifiers, which are a superset of
938 identifiers in the grammatical sense. */
939
940 union GTY(()) _cpp_hashnode_value {
941 /* Assert (maybe NULL) */
942 cpp_macro * GTY((tag ("NT_VOID"))) answers;
943 /* Macro (maybe NULL) */
944 cpp_macro * GTY((tag ("NT_USER_MACRO"))) macro;
945 /* Code for a builtin macro. */
946 enum cpp_builtin_type GTY ((tag ("NT_BUILTIN_MACRO"))) builtin;
947 /* Macro argument index. */
948 unsigned short GTY ((tag ("NT_MACRO_ARG"))) arg_index;
949 };
950
951 struct GTY(()) cpp_hashnode {
952 struct ht_identifier ident;
953 unsigned int is_directive : 1;
954 unsigned int directive_index : 7; /* If is_directive,
955 then index into directive table.
956 Otherwise, a NODE_OPERATOR. */
957 unsigned int rid_code : 8; /* Rid code - for front ends. */
958 unsigned int flags : 9; /* CPP flags. */
959 ENUM_BITFIELD(node_type) type : 2; /* CPP node type. */
960
961 /* 5 bits spare. */
962
963 /* The deferred cookie is applicable to NT_USER_MACRO or NT_VOID.
964 The latter for when a macro had a prevailing undef.
965 On a 64-bit system there would be 32-bits of padding to the value
966 field. So placing the deferred index here is not costly. */
967 unsigned deferred; /* Deferred cookie */
968
969 union _cpp_hashnode_value GTY ((desc ("%1.type"))) value;
970 };
971
972 /* A class for iterating through the source locations within a
973 string token (before escapes are interpreted, and before
974 concatenation). */
975
976 class cpp_string_location_reader {
977 public:
978 cpp_string_location_reader (location_t src_loc,
979 line_maps *line_table);
980
981 source_range get_next ();
982
983 private:
984 location_t m_loc;
985 int m_offset_per_column;
986 };
987
988 /* A class for storing the source ranges of all of the characters within
989 a string literal, after escapes are interpreted, and after
990 concatenation.
991
992 This is not GTY-marked, as instances are intended to be temporary. */
993
994 class cpp_substring_ranges
995 {
996 public:
997 cpp_substring_ranges ();
998 ~cpp_substring_ranges ();
999
get_num_ranges()1000 int get_num_ranges () const { return m_num_ranges; }
get_range(int idx)1001 source_range get_range (int idx) const
1002 {
1003 linemap_assert (idx < m_num_ranges);
1004 return m_ranges[idx];
1005 }
1006
1007 void add_range (source_range range);
1008 void add_n_ranges (int num, cpp_string_location_reader &loc_reader);
1009
1010 private:
1011 source_range *m_ranges;
1012 int m_num_ranges;
1013 int m_alloc_ranges;
1014 };
1015
1016 /* Call this first to get a handle to pass to other functions.
1017
1018 If you want cpplib to manage its own hashtable, pass in a NULL
1019 pointer. Otherwise you should pass in an initialized hash table
1020 that cpplib will share; this technique is used by the C front
1021 ends. */
1022 extern cpp_reader *cpp_create_reader (enum c_lang, struct ht *,
1023 class line_maps *);
1024
1025 /* Reset the cpp_reader's line_map. This is only used after reading a
1026 PCH file. */
1027 extern void cpp_set_line_map (cpp_reader *, class line_maps *);
1028
1029 /* Call this to change the selected language standard (e.g. because of
1030 command line options). */
1031 extern void cpp_set_lang (cpp_reader *, enum c_lang);
1032
1033 /* Set the include paths. */
1034 extern void cpp_set_include_chains (cpp_reader *, cpp_dir *, cpp_dir *, int);
1035
1036 /* Provide src:dst pair for __FILE__ remapping. */
1037 extern void add_cpp_remap_path (const char *);
1038
1039 /* Call these to get pointers to the options, callback, and deps
1040 structures for a given reader. These pointers are good until you
1041 call cpp_finish on that reader. You can either edit the callbacks
1042 through the pointer returned from cpp_get_callbacks, or set them
1043 with cpp_set_callbacks. */
1044 extern cpp_options *cpp_get_options (cpp_reader *) ATTRIBUTE_PURE;
1045 extern cpp_callbacks *cpp_get_callbacks (cpp_reader *) ATTRIBUTE_PURE;
1046 extern void cpp_set_callbacks (cpp_reader *, cpp_callbacks *);
1047 extern class mkdeps *cpp_get_deps (cpp_reader *) ATTRIBUTE_PURE;
1048
1049 extern const char *cpp_probe_header_unit (cpp_reader *, const char *file,
1050 bool angle_p, location_t);
1051
1052 /* Call these to get name data about the various compile-time
1053 charsets. */
1054 extern const char *cpp_get_narrow_charset_name (cpp_reader *) ATTRIBUTE_PURE;
1055 extern const char *cpp_get_wide_charset_name (cpp_reader *) ATTRIBUTE_PURE;
1056
1057 /* This function reads the file, but does not start preprocessing. It
1058 returns the name of the original file; this is the same as the
1059 input file, except for preprocessed input. This will generate at
1060 least one file change callback, and possibly a line change callback
1061 too. If there was an error opening the file, it returns NULL. */
1062 extern const char *cpp_read_main_file (cpp_reader *, const char *,
1063 bool injecting = false);
1064 extern location_t cpp_main_loc (const cpp_reader *);
1065
1066 /* Adjust for the main file to be an include. */
1067 extern void cpp_retrofit_as_include (cpp_reader *);
1068
1069 /* Set up built-ins with special behavior. Use cpp_init_builtins()
1070 instead unless your know what you are doing. */
1071 extern void cpp_init_special_builtins (cpp_reader *);
1072
1073 /* Set up built-ins like __FILE__. */
1074 extern void cpp_init_builtins (cpp_reader *, int);
1075
1076 /* This is called after options have been parsed, and partially
1077 processed. */
1078 extern void cpp_post_options (cpp_reader *);
1079
1080 /* Set up translation to the target character set. */
1081 extern void cpp_init_iconv (cpp_reader *);
1082
1083 /* Call this to finish preprocessing. If you requested dependency
1084 generation, pass an open stream to write the information to,
1085 otherwise NULL. It is your responsibility to close the stream. */
1086 extern void cpp_finish (cpp_reader *, FILE *deps_stream);
1087
1088 /* Call this to release the handle at the end of preprocessing. Any
1089 use of the handle after this function returns is invalid. */
1090 extern void cpp_destroy (cpp_reader *);
1091
1092 extern unsigned int cpp_token_len (const cpp_token *);
1093 extern unsigned char *cpp_token_as_text (cpp_reader *, const cpp_token *);
1094 extern unsigned char *cpp_spell_token (cpp_reader *, const cpp_token *,
1095 unsigned char *, bool);
1096 extern void cpp_register_pragma (cpp_reader *, const char *, const char *,
1097 void (*) (cpp_reader *), bool);
1098 extern void cpp_register_deferred_pragma (cpp_reader *, const char *,
1099 const char *, unsigned, bool, bool);
1100 extern int cpp_avoid_paste (cpp_reader *, const cpp_token *,
1101 const cpp_token *);
1102 extern const cpp_token *cpp_get_token (cpp_reader *);
1103 extern const cpp_token *cpp_get_token_with_location (cpp_reader *,
1104 location_t *);
cpp_user_macro_p(const cpp_hashnode * node)1105 inline bool cpp_user_macro_p (const cpp_hashnode *node)
1106 {
1107 return node->type == NT_USER_MACRO;
1108 }
cpp_builtin_macro_p(const cpp_hashnode * node)1109 inline bool cpp_builtin_macro_p (const cpp_hashnode *node)
1110 {
1111 return node->type == NT_BUILTIN_MACRO;
1112 }
cpp_macro_p(const cpp_hashnode * node)1113 inline bool cpp_macro_p (const cpp_hashnode *node)
1114 {
1115 return node->type & NT_MACRO_MASK;
1116 }
1117 inline cpp_macro *cpp_set_deferred_macro (cpp_hashnode *node,
1118 cpp_macro *forced = NULL)
1119 {
1120 cpp_macro *old = node->value.macro;
1121
1122 node->value.macro = forced;
1123 node->type = NT_USER_MACRO;
1124 node->flags &= ~NODE_USED;
1125
1126 return old;
1127 }
1128 cpp_macro *cpp_get_deferred_macro (cpp_reader *, cpp_hashnode *, location_t);
1129
1130 /* Returns true if NODE is a function-like user macro. */
cpp_fun_like_macro_p(cpp_hashnode * node)1131 inline bool cpp_fun_like_macro_p (cpp_hashnode *node)
1132 {
1133 return cpp_user_macro_p (node) && node->value.macro->fun_like;
1134 }
1135
1136 extern const unsigned char *cpp_macro_definition (cpp_reader *, cpp_hashnode *);
1137 extern const unsigned char *cpp_macro_definition (cpp_reader *, cpp_hashnode *,
1138 const cpp_macro *);
cpp_macro_definition_location(cpp_hashnode * node)1139 inline location_t cpp_macro_definition_location (cpp_hashnode *node)
1140 {
1141 const cpp_macro *macro = node->value.macro;
1142 return macro ? macro->line : 0;
1143 }
1144 /* Return an idempotent time stamp (possibly from SOURCE_DATE_EPOCH). */
1145 enum class CPP_time_kind
1146 {
1147 FIXED = -1, /* Fixed time via source epoch. */
1148 DYNAMIC = -2, /* Dynamic via time(2). */
1149 UNKNOWN = -3 /* Wibbly wobbly, timey wimey. */
1150 };
1151 extern CPP_time_kind cpp_get_date (cpp_reader *, time_t *);
1152
1153 extern void _cpp_backup_tokens (cpp_reader *, unsigned int);
1154 extern const cpp_token *cpp_peek_token (cpp_reader *, int);
1155
1156 /* Evaluate a CPP_*CHAR* token. */
1157 extern cppchar_t cpp_interpret_charconst (cpp_reader *, const cpp_token *,
1158 unsigned int *, int *);
1159 /* Evaluate a vector of CPP_*STRING* tokens. */
1160 extern bool cpp_interpret_string (cpp_reader *,
1161 const cpp_string *, size_t,
1162 cpp_string *, enum cpp_ttype);
1163 extern const char *cpp_interpret_string_ranges (cpp_reader *pfile,
1164 const cpp_string *from,
1165 cpp_string_location_reader *,
1166 size_t count,
1167 cpp_substring_ranges *out,
1168 enum cpp_ttype type);
1169 extern bool cpp_interpret_string_notranslate (cpp_reader *,
1170 const cpp_string *, size_t,
1171 cpp_string *, enum cpp_ttype);
1172
1173 /* Convert a host character constant to the execution character set. */
1174 extern cppchar_t cpp_host_to_exec_charset (cpp_reader *, cppchar_t);
1175
1176 /* Used to register macros and assertions, perhaps from the command line.
1177 The text is the same as the command line argument. */
1178 extern void cpp_define (cpp_reader *, const char *);
1179 extern void cpp_define_unused (cpp_reader *, const char *);
1180 extern void cpp_define_formatted (cpp_reader *pfile,
1181 const char *fmt, ...) ATTRIBUTE_PRINTF_2;
1182 extern void cpp_define_formatted_unused (cpp_reader *pfile,
1183 const char *fmt,
1184 ...) ATTRIBUTE_PRINTF_2;
1185 extern void cpp_assert (cpp_reader *, const char *);
1186 extern void cpp_undef (cpp_reader *, const char *);
1187 extern void cpp_unassert (cpp_reader *, const char *);
1188
1189 /* Mark a node as a lazily defined macro. */
1190 extern void cpp_define_lazily (cpp_reader *, cpp_hashnode *node, unsigned N);
1191
1192 /* Undefine all macros and assertions. */
1193 extern void cpp_undef_all (cpp_reader *);
1194
1195 extern cpp_buffer *cpp_push_buffer (cpp_reader *, const unsigned char *,
1196 size_t, int);
1197 extern int cpp_defined (cpp_reader *, const unsigned char *, int);
1198
1199 /* A preprocessing number. Code assumes that any unused high bits of
1200 the double integer are set to zero. */
1201
1202 /* This type has to be equal to unsigned HOST_WIDE_INT, see
1203 gcc/c-family/c-lex.cc. */
1204 typedef uint64_t cpp_num_part;
1205 typedef struct cpp_num cpp_num;
1206 struct cpp_num
1207 {
1208 cpp_num_part high;
1209 cpp_num_part low;
1210 bool unsignedp; /* True if value should be treated as unsigned. */
1211 bool overflow; /* True if the most recent calculation overflowed. */
1212 };
1213
1214 /* cpplib provides two interfaces for interpretation of preprocessing
1215 numbers.
1216
1217 cpp_classify_number categorizes numeric constants according to
1218 their field (integer, floating point, or invalid), radix (decimal,
1219 octal, hexadecimal), and type suffixes. */
1220
1221 #define CPP_N_CATEGORY 0x000F
1222 #define CPP_N_INVALID 0x0000
1223 #define CPP_N_INTEGER 0x0001
1224 #define CPP_N_FLOATING 0x0002
1225
1226 #define CPP_N_WIDTH 0x00F0
1227 #define CPP_N_SMALL 0x0010 /* int, float, short _Fract/Accum */
1228 #define CPP_N_MEDIUM 0x0020 /* long, double, long _Fract/_Accum. */
1229 #define CPP_N_LARGE 0x0040 /* long long, long double,
1230 long long _Fract/Accum. */
1231
1232 #define CPP_N_WIDTH_MD 0xF0000 /* machine defined. */
1233 #define CPP_N_MD_W 0x10000
1234 #define CPP_N_MD_Q 0x20000
1235
1236 #define CPP_N_RADIX 0x0F00
1237 #define CPP_N_DECIMAL 0x0100
1238 #define CPP_N_HEX 0x0200
1239 #define CPP_N_OCTAL 0x0400
1240 #define CPP_N_BINARY 0x0800
1241
1242 #define CPP_N_UNSIGNED 0x1000 /* Properties. */
1243 #define CPP_N_IMAGINARY 0x2000
1244 #define CPP_N_DFLOAT 0x4000
1245 #define CPP_N_DEFAULT 0x8000
1246
1247 #define CPP_N_FRACT 0x100000 /* Fract types. */
1248 #define CPP_N_ACCUM 0x200000 /* Accum types. */
1249 #define CPP_N_FLOATN 0x400000 /* _FloatN types. */
1250 #define CPP_N_FLOATNX 0x800000 /* _FloatNx types. */
1251
1252 #define CPP_N_USERDEF 0x1000000 /* C++11 user-defined literal. */
1253
1254 #define CPP_N_SIZE_T 0x2000000 /* C++23 size_t literal. */
1255
1256 #define CPP_N_WIDTH_FLOATN_NX 0xF0000000 /* _FloatN / _FloatNx value
1257 of N, divided by 16. */
1258 #define CPP_FLOATN_SHIFT 24
1259 #define CPP_FLOATN_MAX 0xF0
1260
1261 /* Classify a CPP_NUMBER token. The return value is a combination of
1262 the flags from the above sets. */
1263 extern unsigned cpp_classify_number (cpp_reader *, const cpp_token *,
1264 const char **, location_t);
1265
1266 /* Return the classification flags for a float suffix. */
1267 extern unsigned int cpp_interpret_float_suffix (cpp_reader *, const char *,
1268 size_t);
1269
1270 /* Return the classification flags for an int suffix. */
1271 extern unsigned int cpp_interpret_int_suffix (cpp_reader *, const char *,
1272 size_t);
1273
1274 /* Evaluate a token classified as category CPP_N_INTEGER. */
1275 extern cpp_num cpp_interpret_integer (cpp_reader *, const cpp_token *,
1276 unsigned int);
1277
1278 /* Sign extend a number, with PRECISION significant bits and all
1279 others assumed clear, to fill out a cpp_num structure. */
1280 cpp_num cpp_num_sign_extend (cpp_num, size_t);
1281
1282 /* Output a diagnostic of some kind. */
1283 extern bool cpp_error (cpp_reader *, enum cpp_diagnostic_level,
1284 const char *msgid, ...)
1285 ATTRIBUTE_PRINTF_3;
1286 extern bool cpp_warning (cpp_reader *, enum cpp_warning_reason,
1287 const char *msgid, ...)
1288 ATTRIBUTE_PRINTF_3;
1289 extern bool cpp_pedwarning (cpp_reader *, enum cpp_warning_reason,
1290 const char *msgid, ...)
1291 ATTRIBUTE_PRINTF_3;
1292 extern bool cpp_warning_syshdr (cpp_reader *, enum cpp_warning_reason reason,
1293 const char *msgid, ...)
1294 ATTRIBUTE_PRINTF_3;
1295
1296 /* As their counterparts above, but use RICHLOC. */
1297 extern bool cpp_warning_at (cpp_reader *, enum cpp_warning_reason,
1298 rich_location *richloc, const char *msgid, ...)
1299 ATTRIBUTE_PRINTF_4;
1300 extern bool cpp_pedwarning_at (cpp_reader *, enum cpp_warning_reason,
1301 rich_location *richloc, const char *msgid, ...)
1302 ATTRIBUTE_PRINTF_4;
1303
1304 /* Output a diagnostic with "MSGID: " preceding the
1305 error string of errno. No location is printed. */
1306 extern bool cpp_errno (cpp_reader *, enum cpp_diagnostic_level,
1307 const char *msgid);
1308 /* Similarly, but with "FILENAME: " instead of "MSGID: ", where
1309 the filename is not localized. */
1310 extern bool cpp_errno_filename (cpp_reader *, enum cpp_diagnostic_level,
1311 const char *filename, location_t loc);
1312
1313 /* Same as cpp_error, except additionally specifies a position as a
1314 (translation unit) physical line and physical column. If the line is
1315 zero, then no location is printed. */
1316 extern bool cpp_error_with_line (cpp_reader *, enum cpp_diagnostic_level,
1317 location_t, unsigned,
1318 const char *msgid, ...)
1319 ATTRIBUTE_PRINTF_5;
1320 extern bool cpp_warning_with_line (cpp_reader *, enum cpp_warning_reason,
1321 location_t, unsigned,
1322 const char *msgid, ...)
1323 ATTRIBUTE_PRINTF_5;
1324 extern bool cpp_pedwarning_with_line (cpp_reader *, enum cpp_warning_reason,
1325 location_t, unsigned,
1326 const char *msgid, ...)
1327 ATTRIBUTE_PRINTF_5;
1328 extern bool cpp_warning_with_line_syshdr (cpp_reader *, enum cpp_warning_reason,
1329 location_t, unsigned,
1330 const char *msgid, ...)
1331 ATTRIBUTE_PRINTF_5;
1332
1333 extern bool cpp_error_at (cpp_reader * pfile, enum cpp_diagnostic_level,
1334 location_t src_loc, const char *msgid, ...)
1335 ATTRIBUTE_PRINTF_4;
1336
1337 extern bool cpp_error_at (cpp_reader * pfile, enum cpp_diagnostic_level,
1338 rich_location *richloc, const char *msgid, ...)
1339 ATTRIBUTE_PRINTF_4;
1340
1341 /* In lex.cc */
1342 extern int cpp_ideq (const cpp_token *, const char *);
1343 extern void cpp_output_line (cpp_reader *, FILE *);
1344 extern unsigned char *cpp_output_line_to_string (cpp_reader *,
1345 const unsigned char *);
1346 extern const unsigned char *cpp_alloc_token_string
1347 (cpp_reader *, const unsigned char *, unsigned);
1348 extern void cpp_output_token (const cpp_token *, FILE *);
1349 extern const char *cpp_type2name (enum cpp_ttype, unsigned char flags);
1350 /* Returns the value of an escape sequence, truncated to the correct
1351 target precision. PSTR points to the input pointer, which is just
1352 after the backslash. LIMIT is how much text we have. WIDE is true
1353 if the escape sequence is part of a wide character constant or
1354 string literal. Handles all relevant diagnostics. */
1355 extern cppchar_t cpp_parse_escape (cpp_reader *, const unsigned char ** pstr,
1356 const unsigned char *limit, int wide);
1357
1358 /* Structure used to hold a comment block at a given location in the
1359 source code. */
1360
1361 typedef struct
1362 {
1363 /* Text of the comment including the terminators. */
1364 char *comment;
1365
1366 /* source location for the given comment. */
1367 location_t sloc;
1368 } cpp_comment;
1369
1370 /* Structure holding all comments for a given cpp_reader. */
1371
1372 typedef struct
1373 {
1374 /* table of comment entries. */
1375 cpp_comment *entries;
1376
1377 /* number of actual entries entered in the table. */
1378 int count;
1379
1380 /* number of entries allocated currently. */
1381 int allocated;
1382 } cpp_comment_table;
1383
1384 /* Returns the table of comments encountered by the preprocessor. This
1385 table is only populated when pfile->state.save_comments is true. */
1386 extern cpp_comment_table *cpp_get_comments (cpp_reader *);
1387
1388 /* In hash.c */
1389
1390 /* Lookup an identifier in the hashtable. Puts the identifier in the
1391 table if it is not already there. */
1392 extern cpp_hashnode *cpp_lookup (cpp_reader *, const unsigned char *,
1393 unsigned int);
1394
1395 typedef int (*cpp_cb) (cpp_reader *, cpp_hashnode *, void *);
1396 extern void cpp_forall_identifiers (cpp_reader *, cpp_cb, void *);
1397
1398 /* In macro.cc */
1399 extern void cpp_scan_nooutput (cpp_reader *);
1400 extern int cpp_sys_macro_p (cpp_reader *);
1401 extern unsigned char *cpp_quote_string (unsigned char *, const unsigned char *,
1402 unsigned int);
1403 extern bool cpp_compare_macros (const cpp_macro *macro1,
1404 const cpp_macro *macro2);
1405
1406 /* In files.cc */
1407 extern bool cpp_included (cpp_reader *, const char *);
1408 extern bool cpp_included_before (cpp_reader *, const char *, location_t);
1409 extern void cpp_make_system_header (cpp_reader *, int, int);
1410 extern bool cpp_push_include (cpp_reader *, const char *);
1411 extern bool cpp_push_default_include (cpp_reader *, const char *);
1412 extern void cpp_change_file (cpp_reader *, enum lc_reason, const char *);
1413 extern const char *cpp_get_path (struct _cpp_file *);
1414 extern cpp_dir *cpp_get_dir (struct _cpp_file *);
1415 extern cpp_buffer *cpp_get_buffer (cpp_reader *);
1416 extern struct _cpp_file *cpp_get_file (cpp_buffer *);
1417 extern cpp_buffer *cpp_get_prev (cpp_buffer *);
1418 extern void cpp_clear_file_cache (cpp_reader *);
1419
1420 /* cpp_get_converted_source returns the contents of the given file, as it exists
1421 after cpplib has read it and converted it from the input charset to the
1422 source charset. Return struct will be zero-filled if the data could not be
1423 read for any reason. The data starts at the DATA pointer, but the TO_FREE
1424 pointer is what should be passed to free(), as there may be an offset. */
1425 struct cpp_converted_source
1426 {
1427 char *to_free;
1428 char *data;
1429 size_t len;
1430 };
1431 cpp_converted_source cpp_get_converted_source (const char *fname,
1432 const char *input_charset);
1433
1434 /* In pch.cc */
1435 struct save_macro_data;
1436 extern int cpp_save_state (cpp_reader *, FILE *);
1437 extern int cpp_write_pch_deps (cpp_reader *, FILE *);
1438 extern int cpp_write_pch_state (cpp_reader *, FILE *);
1439 extern int cpp_valid_state (cpp_reader *, const char *, int);
1440 extern void cpp_prepare_state (cpp_reader *, struct save_macro_data **);
1441 extern int cpp_read_state (cpp_reader *, const char *, FILE *,
1442 struct save_macro_data *);
1443
1444 /* In lex.cc */
1445 extern void cpp_force_token_locations (cpp_reader *, location_t);
1446 extern void cpp_stop_forcing_token_locations (cpp_reader *);
1447 enum CPP_DO_task
1448 {
1449 CPP_DO_print,
1450 CPP_DO_location,
1451 CPP_DO_token
1452 };
1453
1454 extern void cpp_directive_only_process (cpp_reader *pfile,
1455 void *data,
1456 void (*cb) (cpp_reader *,
1457 CPP_DO_task,
1458 void *data, ...));
1459
1460 /* In expr.cc */
1461 extern enum cpp_ttype cpp_userdef_string_remove_type
1462 (enum cpp_ttype type);
1463 extern enum cpp_ttype cpp_userdef_string_add_type
1464 (enum cpp_ttype type);
1465 extern enum cpp_ttype cpp_userdef_char_remove_type
1466 (enum cpp_ttype type);
1467 extern enum cpp_ttype cpp_userdef_char_add_type
1468 (enum cpp_ttype type);
1469 extern bool cpp_userdef_string_p
1470 (enum cpp_ttype type);
1471 extern bool cpp_userdef_char_p
1472 (enum cpp_ttype type);
1473 extern const char * cpp_get_userdef_suffix
1474 (const cpp_token *);
1475
1476 /* In charset.cc */
1477
1478 /* The result of attempting to decode a run of UTF-8 bytes. */
1479
1480 struct cpp_decoded_char
1481 {
1482 const char *m_start_byte;
1483 const char *m_next_byte;
1484
1485 bool m_valid_ch;
1486 cppchar_t m_ch;
1487 };
1488
1489 /* Information for mapping between code points and display columns.
1490
1491 This is a tabstop value, along with a callback for getting the
1492 widths of characters. Normally this callback is cpp_wcwidth, but we
1493 support other schemes for escaping non-ASCII unicode as a series of
1494 ASCII chars when printing the user's source code in diagnostic-show-locus.cc
1495
1496 For example, consider:
1497 - the Unicode character U+03C0 "GREEK SMALL LETTER PI" (UTF-8: 0xCF 0x80)
1498 - the Unicode character U+1F642 "SLIGHTLY SMILING FACE"
1499 (UTF-8: 0xF0 0x9F 0x99 0x82)
1500 - the byte 0xBF (a stray trailing byte of a UTF-8 character)
1501 Normally U+03C0 would occupy one display column, U+1F642
1502 would occupy two display columns, and the stray byte would be
1503 printed verbatim as one display column.
1504
1505 However when escaping them as unicode code points as "<U+03C0>"
1506 and "<U+1F642>" they occupy 8 and 9 display columns respectively,
1507 and when escaping them as bytes as "<CF><80>" and "<F0><9F><99><82>"
1508 they occupy 8 and 16 display columns respectively. In both cases
1509 the stray byte is escaped to <BF> as 4 display columns. */
1510
1511 struct cpp_char_column_policy
1512 {
cpp_char_column_policycpp_char_column_policy1513 cpp_char_column_policy (int tabstop,
1514 int (*width_cb) (cppchar_t c))
1515 : m_tabstop (tabstop),
1516 m_undecoded_byte_width (1),
1517 m_width_cb (width_cb)
1518 {}
1519
1520 int m_tabstop;
1521 /* Width in display columns of a stray byte that isn't decodable
1522 as UTF-8. */
1523 int m_undecoded_byte_width;
1524 int (*m_width_cb) (cppchar_t c);
1525 };
1526
1527 /* A class to manage the state while converting a UTF-8 sequence to cppchar_t
1528 and computing the display width one character at a time. */
1529 class cpp_display_width_computation {
1530 public:
1531 cpp_display_width_computation (const char *data, int data_length,
1532 const cpp_char_column_policy &policy);
next_byte()1533 const char *next_byte () const { return m_next; }
bytes_processed()1534 int bytes_processed () const { return m_next - m_begin; }
bytes_left()1535 int bytes_left () const { return m_bytes_left; }
done()1536 bool done () const { return !bytes_left (); }
display_cols_processed()1537 int display_cols_processed () const { return m_display_cols; }
1538
1539 int process_next_codepoint (cpp_decoded_char *out);
1540 int advance_display_cols (int n);
1541
1542 private:
1543 const char *const m_begin;
1544 const char *m_next;
1545 size_t m_bytes_left;
1546 const cpp_char_column_policy &m_policy;
1547 int m_display_cols;
1548 };
1549
1550 /* Convenience functions that are simple use cases for class
1551 cpp_display_width_computation. Tab characters will be expanded to spaces
1552 as determined by POLICY.m_tabstop, and non-printable-ASCII characters
1553 will be escaped as per POLICY. */
1554
1555 int cpp_byte_column_to_display_column (const char *data, int data_length,
1556 int column,
1557 const cpp_char_column_policy &policy);
cpp_display_width(const char * data,int data_length,const cpp_char_column_policy & policy)1558 inline int cpp_display_width (const char *data, int data_length,
1559 const cpp_char_column_policy &policy)
1560 {
1561 return cpp_byte_column_to_display_column (data, data_length, data_length,
1562 policy);
1563 }
1564 int cpp_display_column_to_byte_column (const char *data, int data_length,
1565 int display_col,
1566 const cpp_char_column_policy &policy);
1567 int cpp_wcwidth (cppchar_t c);
1568
1569 bool cpp_input_conversion_is_trivial (const char *input_charset);
1570 int cpp_check_utf8_bom (const char *data, size_t data_length);
1571
1572 #endif /* ! LIBCPP_CPPLIB_H */
1573