1 /* Definitions for CPP library.
2 Copyright (C) 1995-2021 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.c 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.c. */
196 #define SP_DIGRAPH (1 << 8) /* # or ## token was a digraph. */
197 #define SP_PREV_WHITE (1 << 9) /* If whitespace before a ##
198 operator, or before this token
199 after a # operator. */
200 #define NO_EXPAND (1 << 10) /* Do not macro-expand this token. */
201 #define PRAGMA_OP (1 << 11) /* _Pragma token. */
202
203 /* Specify which field, if any, of the cpp_token union is used. */
204
205 enum cpp_token_fld_kind {
206 CPP_TOKEN_FLD_NODE,
207 CPP_TOKEN_FLD_SOURCE,
208 CPP_TOKEN_FLD_STR,
209 CPP_TOKEN_FLD_ARG_NO,
210 CPP_TOKEN_FLD_TOKEN_NO,
211 CPP_TOKEN_FLD_PRAGMA,
212 CPP_TOKEN_FLD_NONE
213 };
214
215 /* A macro argument in the cpp_token union. */
216 struct GTY(()) cpp_macro_arg {
217 /* Argument number. */
218 unsigned int arg_no;
219 /* The original spelling of the macro argument token. */
220 cpp_hashnode *
221 GTY ((nested_ptr (union tree_node,
222 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
223 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL")))
224 spelling;
225 };
226
227 /* An identifier in the cpp_token union. */
228 struct GTY(()) cpp_identifier {
229 /* The canonical (UTF-8) spelling of the identifier. */
230 cpp_hashnode *
231 GTY ((nested_ptr (union tree_node,
232 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
233 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL")))
234 node;
235 /* The original spelling of the identifier. */
236 cpp_hashnode *
237 GTY ((nested_ptr (union tree_node,
238 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
239 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL")))
240 spelling;
241 };
242
243 /* A preprocessing token. This has been carefully packed and should
244 occupy 16 bytes on 32-bit hosts and 24 bytes on 64-bit hosts. */
245 struct GTY(()) cpp_token {
246
247 /* Location of first char of token, together with range of full token. */
248 location_t src_loc;
249
250 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
251 unsigned short flags; /* flags - see above */
252
253 union cpp_token_u
254 {
255 /* An identifier. */
256 struct cpp_identifier GTY ((tag ("CPP_TOKEN_FLD_NODE"))) node;
257
258 /* Inherit padding from this token. */
259 cpp_token * GTY ((tag ("CPP_TOKEN_FLD_SOURCE"))) source;
260
261 /* A string, or number. */
262 struct cpp_string GTY ((tag ("CPP_TOKEN_FLD_STR"))) str;
263
264 /* Argument no. (and original spelling) for a CPP_MACRO_ARG. */
265 struct cpp_macro_arg GTY ((tag ("CPP_TOKEN_FLD_ARG_NO"))) macro_arg;
266
267 /* Original token no. for a CPP_PASTE (from a sequence of
268 consecutive paste tokens in a macro expansion). */
269 unsigned int GTY ((tag ("CPP_TOKEN_FLD_TOKEN_NO"))) token_no;
270
271 /* Caller-supplied identifier for a CPP_PRAGMA. */
272 unsigned int GTY ((tag ("CPP_TOKEN_FLD_PRAGMA"))) pragma;
273 } GTY ((desc ("cpp_token_val_index (&%1)"))) val;
274 };
275
276 /* Say which field is in use. */
277 extern enum cpp_token_fld_kind cpp_token_val_index (const cpp_token *tok);
278
279 /* A type wide enough to hold any multibyte source character.
280 cpplib's character constant interpreter requires an unsigned type.
281 Also, a typedef for the signed equivalent.
282 The width of this type is capped at 32 bits; there do exist targets
283 where wchar_t is 64 bits, but only in a non-default mode, and there
284 would be no meaningful interpretation for a wchar_t value greater
285 than 2^32 anyway -- the widest wide-character encoding around is
286 ISO 10646, which stops at 2^31. */
287 #if CHAR_BIT * SIZEOF_INT >= 32
288 # define CPPCHAR_SIGNED_T int
289 #elif CHAR_BIT * SIZEOF_LONG >= 32
290 # define CPPCHAR_SIGNED_T long
291 #else
292 # error "Cannot find a least-32-bit signed integer type"
293 #endif
294 typedef unsigned CPPCHAR_SIGNED_T cppchar_t;
295 typedef CPPCHAR_SIGNED_T cppchar_signed_t;
296
297 /* Style of header dependencies to generate. */
298 enum cpp_deps_style { DEPS_NONE = 0, DEPS_USER, DEPS_SYSTEM };
299
300 /* The possible normalization levels, from most restrictive to least. */
301 enum cpp_normalize_level {
302 /* In NFKC. */
303 normalized_KC = 0,
304 /* In NFC. */
305 normalized_C,
306 /* In NFC, except for subsequences where being in NFC would make
307 the identifier invalid. */
308 normalized_identifier_C,
309 /* Not normalized at all. */
310 normalized_none
311 };
312
313 enum cpp_main_search
314 {
315 CMS_none, /* A regular source file. */
316 CMS_header, /* Is a directly-specified header file (eg PCH or
317 header-unit). */
318 CMS_user, /* Search the user INCLUDE path. */
319 CMS_system, /* Search the system INCLUDE path. */
320 };
321
322 /* The possible bidirectional control characters checking levels, from least
323 restrictive to most. */
324 enum cpp_bidirectional_level {
325 /* No checking. */
326 bidirectional_none,
327 /* Only detect unpaired uses of bidirectional control characters. */
328 bidirectional_unpaired,
329 /* Detect any use of bidirectional control characters. */
330 bidirectional_any
331 };
332
333 /* This structure is nested inside struct cpp_reader, and
334 carries all the options visible to the command line. */
335 struct cpp_options
336 {
337 /* The language we're preprocessing. */
338 enum c_lang lang;
339
340 /* Nonzero means use extra default include directories for C++. */
341 unsigned char cplusplus;
342
343 /* Nonzero means handle cplusplus style comments. */
344 unsigned char cplusplus_comments;
345
346 /* Nonzero means define __OBJC__, treat @ as a special token, use
347 the OBJC[PLUS]_INCLUDE_PATH environment variable, and allow
348 "#import". */
349 unsigned char objc;
350
351 /* Nonzero means don't copy comments into the output file. */
352 unsigned char discard_comments;
353
354 /* Nonzero means don't copy comments into the output file during
355 macro expansion. */
356 unsigned char discard_comments_in_macro_exp;
357
358 /* Nonzero means process the ISO trigraph sequences. */
359 unsigned char trigraphs;
360
361 /* Nonzero means process the ISO digraph sequences. */
362 unsigned char digraphs;
363
364 /* Nonzero means to allow hexadecimal floats and LL suffixes. */
365 unsigned char extended_numbers;
366
367 /* Nonzero means process u/U prefix literals (UTF-16/32). */
368 unsigned char uliterals;
369
370 /* Nonzero means process u8 prefixed character literals (UTF-8). */
371 unsigned char utf8_char_literals;
372
373 /* Nonzero means process r/R raw strings. If this is set, uliterals
374 must be set as well. */
375 unsigned char rliterals;
376
377 /* Nonzero means print names of header files (-H). */
378 unsigned char print_include_names;
379
380 /* Nonzero means complain about deprecated features. */
381 unsigned char cpp_warn_deprecated;
382
383 /* Nonzero means warn if slash-star appears in a comment. */
384 unsigned char warn_comments;
385
386 /* Nonzero means to warn about __DATA__, __TIME__ and __TIMESTAMP__ usage. */
387 unsigned char warn_date_time;
388
389 /* Nonzero means warn if a user-supplied include directory does not
390 exist. */
391 unsigned char warn_missing_include_dirs;
392
393 /* Nonzero means warn if there are any trigraphs. */
394 unsigned char warn_trigraphs;
395
396 /* Nonzero means warn about multicharacter charconsts. */
397 unsigned char warn_multichar;
398
399 /* Nonzero means warn about various incompatibilities with
400 traditional C. */
401 unsigned char cpp_warn_traditional;
402
403 /* Nonzero means warn about long long numeric constants. */
404 unsigned char cpp_warn_long_long;
405
406 /* Nonzero means warn about text after an #endif (or #else). */
407 unsigned char warn_endif_labels;
408
409 /* Nonzero means warn about implicit sign changes owing to integer
410 promotions. */
411 unsigned char warn_num_sign_change;
412
413 /* Zero means don't warn about __VA_ARGS__ usage in c89 pedantic mode.
414 Presumably the usage is protected by the appropriate #ifdef. */
415 unsigned char warn_variadic_macros;
416
417 /* Nonzero means warn about builtin macros that are redefined or
418 explicitly undefined. */
419 unsigned char warn_builtin_macro_redefined;
420
421 /* Different -Wimplicit-fallthrough= levels. */
422 unsigned char cpp_warn_implicit_fallthrough;
423
424 /* Nonzero means we should look for header.gcc files that remap file
425 names. */
426 unsigned char remap;
427
428 /* Zero means dollar signs are punctuation. */
429 unsigned char dollars_in_ident;
430
431 /* Nonzero means UCNs are accepted in identifiers. */
432 unsigned char extended_identifiers;
433
434 /* True if we should warn about dollars in identifiers or numbers
435 for this translation unit. */
436 unsigned char warn_dollars;
437
438 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
439 unsigned char warn_undef;
440
441 /* Nonzero means warn if "defined" is encountered in a place other than
442 an #if. */
443 unsigned char warn_expansion_to_defined;
444
445 /* Nonzero means warn of unused macros from the main file. */
446 unsigned char warn_unused_macros;
447
448 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
449 unsigned char c99;
450
451 /* Nonzero if we are conforming to a specific C or C++ standard. */
452 unsigned char std;
453
454 /* Nonzero means give all the error messages the ANSI standard requires. */
455 unsigned char cpp_pedantic;
456
457 /* Nonzero means we're looking at already preprocessed code, so don't
458 bother trying to do macro expansion and whatnot. */
459 unsigned char preprocessed;
460
461 /* Nonzero means we are going to emit debugging logs during
462 preprocessing. */
463 unsigned char debug;
464
465 /* Nonzero means we are tracking locations of tokens involved in
466 macro expansion. 1 Means we track the location in degraded mode
467 where we do not track locations of tokens resulting from the
468 expansion of arguments of function-like macro. 2 Means we do
469 track all macro expansions. This last option is the one that
470 consumes the highest amount of memory. */
471 unsigned char track_macro_expansion;
472
473 /* Nonzero means handle C++ alternate operator names. */
474 unsigned char operator_names;
475
476 /* Nonzero means warn about use of C++ alternate operator names. */
477 unsigned char warn_cxx_operator_names;
478
479 /* True for traditional preprocessing. */
480 unsigned char traditional;
481
482 /* Nonzero for C++ 2011 Standard user-defined literals. */
483 unsigned char user_literals;
484
485 /* Nonzero means warn when a string or character literal is followed by a
486 ud-suffix which does not beging with an underscore. */
487 unsigned char warn_literal_suffix;
488
489 /* Nonzero means interpret imaginary, fixed-point, or other gnu extension
490 literal number suffixes as user-defined literal number suffixes. */
491 unsigned char ext_numeric_literals;
492
493 /* Nonzero means extended identifiers allow the characters specified
494 in C11. */
495 unsigned char c11_identifiers;
496
497 /* Nonzero for C++ 2014 Standard binary constants. */
498 unsigned char binary_constants;
499
500 /* Nonzero for C++ 2014 Standard digit separators. */
501 unsigned char digit_separators;
502
503 /* Nonzero for C2X decimal floating-point constants. */
504 unsigned char dfp_constants;
505
506 /* Nonzero for C++20 __VA_OPT__ feature. */
507 unsigned char va_opt;
508
509 /* Nonzero for the '::' token. */
510 unsigned char scope;
511
512 /* Nonzero for the '#elifdef' and '#elifndef' directives. */
513 unsigned char elifdef;
514
515 /* Nonzero means tokenize C++20 module directives. */
516 unsigned char module_directives;
517
518 /* Nonzero for C++23 size_t literals. */
519 unsigned char size_t_literals;
520
521 /* Holds the name of the target (execution) character set. */
522 const char *narrow_charset;
523
524 /* Holds the name of the target wide character set. */
525 const char *wide_charset;
526
527 /* Holds the name of the input character set. */
528 const char *input_charset;
529
530 /* The minimum permitted level of normalization before a warning
531 is generated. See enum cpp_normalize_level. */
532 int warn_normalize;
533
534 /* True to warn about precompiled header files we couldn't use. */
535 bool warn_invalid_pch;
536
537 /* True if dependencies should be restored from a precompiled header. */
538 bool restore_pch_deps;
539
540 /* True if warn about differences between C90 and C99. */
541 signed char cpp_warn_c90_c99_compat;
542
543 /* True if warn about differences between C11 and C2X. */
544 signed char cpp_warn_c11_c2x_compat;
545
546 /* True if warn about differences between C++98 and C++11. */
547 bool cpp_warn_cxx11_compat;
548
549 /* Nonzero if bidirectional control characters checking is on. See enum
550 cpp_bidirectional_level. */
551 unsigned char cpp_warn_bidirectional;
552
553 /* Dependency generation. */
554 struct
555 {
556 /* Style of header dependencies to generate. */
557 enum cpp_deps_style style;
558
559 /* Assume missing files are generated files. */
560 bool missing_files;
561
562 /* Generate phony targets for each dependency apart from the first
563 one. */
564 bool phony_targets;
565
566 /* Generate dependency info for modules. */
567 bool modules;
568
569 /* If true, no dependency is generated on the main file. */
570 bool ignore_main_file;
571
572 /* If true, intend to use the preprocessor output (e.g., for compilation)
573 in addition to the dependency info. */
574 bool need_preprocessor_output;
575 } deps;
576
577 /* Target-specific features set by the front end or client. */
578
579 /* Precision for target CPP arithmetic, target characters, target
580 ints and target wide characters, respectively. */
581 size_t precision, char_precision, int_precision, wchar_precision;
582
583 /* True means chars (wide chars) are unsigned. */
584 bool unsigned_char, unsigned_wchar;
585
586 /* True if the most significant byte in a word has the lowest
587 address in memory. */
588 bool bytes_big_endian;
589
590 /* Nonzero means __STDC__ should have the value 0 in system headers. */
591 unsigned char stdc_0_in_system_headers;
592
593 /* True disables tokenization outside of preprocessing directives. */
594 bool directives_only;
595
596 /* True enables canonicalization of system header file paths. */
597 bool canonical_system_headers;
598
599 /* The maximum depth of the nested #include. */
600 unsigned int max_include_depth;
601
602 cpp_main_search main_search : 8;
603 };
604
605 /* Diagnostic levels. To get a diagnostic without associating a
606 position in the translation unit with it, use cpp_error_with_line
607 with a line number of zero. */
608
609 enum cpp_diagnostic_level {
610 /* Warning, an error with -Werror. */
611 CPP_DL_WARNING = 0,
612 /* Same as CPP_DL_WARNING, except it is not suppressed in system headers. */
613 CPP_DL_WARNING_SYSHDR,
614 /* Warning, an error with -pedantic-errors or -Werror. */
615 CPP_DL_PEDWARN,
616 /* An error. */
617 CPP_DL_ERROR,
618 /* An internal consistency check failed. Prints "internal error: ",
619 otherwise the same as CPP_DL_ERROR. */
620 CPP_DL_ICE,
621 /* An informative note following a warning. */
622 CPP_DL_NOTE,
623 /* A fatal error. */
624 CPP_DL_FATAL
625 };
626
627 /* Warning reason codes. Use a reason code of CPP_W_NONE for unclassified
628 warnings and diagnostics that are not warnings. */
629
630 enum cpp_warning_reason {
631 CPP_W_NONE = 0,
632 CPP_W_DEPRECATED,
633 CPP_W_COMMENTS,
634 CPP_W_MISSING_INCLUDE_DIRS,
635 CPP_W_TRIGRAPHS,
636 CPP_W_MULTICHAR,
637 CPP_W_TRADITIONAL,
638 CPP_W_LONG_LONG,
639 CPP_W_ENDIF_LABELS,
640 CPP_W_NUM_SIGN_CHANGE,
641 CPP_W_VARIADIC_MACROS,
642 CPP_W_BUILTIN_MACRO_REDEFINED,
643 CPP_W_DOLLARS,
644 CPP_W_UNDEF,
645 CPP_W_UNUSED_MACROS,
646 CPP_W_CXX_OPERATOR_NAMES,
647 CPP_W_NORMALIZE,
648 CPP_W_INVALID_PCH,
649 CPP_W_WARNING_DIRECTIVE,
650 CPP_W_LITERAL_SUFFIX,
651 CPP_W_SIZE_T_LITERALS,
652 CPP_W_DATE_TIME,
653 CPP_W_PEDANTIC,
654 CPP_W_C90_C99_COMPAT,
655 CPP_W_C11_C2X_COMPAT,
656 CPP_W_CXX11_COMPAT,
657 CPP_W_EXPANSION_TO_DEFINED,
658 CPP_W_BIDIRECTIONAL
659 };
660
661 /* Callback for header lookup for HEADER, which is the name of a
662 source file. It is used as a method of last resort to find headers
663 that are not otherwise found during the normal include processing.
664 The return value is the malloced name of a header to try and open,
665 if any, or NULL otherwise. This callback is called only if the
666 header is otherwise unfound. */
667 typedef const char *(*missing_header_cb)(cpp_reader *, const char *header, cpp_dir **);
668
669 /* Call backs to cpplib client. */
670 struct cpp_callbacks
671 {
672 /* Called when a new line of preprocessed output is started. */
673 void (*line_change) (cpp_reader *, const cpp_token *, int);
674
675 /* Called when switching to/from a new file.
676 The line_map is for the new file. It is NULL if there is no new file.
677 (In C this happens when done with <built-in>+<command line> and also
678 when done with a main file.) This can be used for resource cleanup. */
679 void (*file_change) (cpp_reader *, const line_map_ordinary *);
680
681 void (*dir_change) (cpp_reader *, const char *);
682 void (*include) (cpp_reader *, location_t, const unsigned char *,
683 const char *, int, const cpp_token **);
684 void (*define) (cpp_reader *, location_t, cpp_hashnode *);
685 void (*undef) (cpp_reader *, location_t, cpp_hashnode *);
686 void (*ident) (cpp_reader *, location_t, const cpp_string *);
687 void (*def_pragma) (cpp_reader *, location_t);
688 int (*valid_pch) (cpp_reader *, const char *, int);
689 void (*read_pch) (cpp_reader *, const char *, int, const char *);
690 missing_header_cb missing_header;
691
692 /* Context-sensitive macro support. Returns macro (if any) that should
693 be expanded. */
694 cpp_hashnode * (*macro_to_expand) (cpp_reader *, const cpp_token *);
695
696 /* Called to emit a diagnostic. This callback receives the
697 translated message. */
698 bool (*diagnostic) (cpp_reader *,
699 enum cpp_diagnostic_level,
700 enum cpp_warning_reason,
701 rich_location *,
702 const char *, va_list *)
703 ATTRIBUTE_FPTR_PRINTF(5,0);
704
705 /* Callbacks for when a macro is expanded, or tested (whether
706 defined or not at the time) in #ifdef, #ifndef or "defined". */
707 void (*used_define) (cpp_reader *, location_t, cpp_hashnode *);
708 void (*used_undef) (cpp_reader *, location_t, cpp_hashnode *);
709 /* Called before #define and #undef or other macro definition
710 changes are processed. */
711 void (*before_define) (cpp_reader *);
712 /* Called whenever a macro is expanded or tested.
713 Second argument is the location of the start of the current expansion. */
714 void (*used) (cpp_reader *, location_t, cpp_hashnode *);
715
716 /* Callback to identify whether an attribute exists. */
717 int (*has_attribute) (cpp_reader *, bool);
718
719 /* Callback to determine whether a built-in function is recognized. */
720 int (*has_builtin) (cpp_reader *);
721
722 /* Callback that can change a user lazy into normal macro. */
723 void (*user_lazy_macro) (cpp_reader *, cpp_macro *, unsigned);
724
725 /* Callback to handle deferred cpp_macros. */
726 cpp_macro *(*user_deferred_macro) (cpp_reader *, location_t, cpp_hashnode *);
727
728 /* Callback to parse SOURCE_DATE_EPOCH from environment. */
729 time_t (*get_source_date_epoch) (cpp_reader *);
730
731 /* Callback for providing suggestions for misspelled directives. */
732 const char *(*get_suggestion) (cpp_reader *, const char *, const char *const *);
733
734 /* Callback for when a comment is encountered, giving the location
735 of the opening slash, a pointer to the content (which is not
736 necessarily 0-terminated), and the length of the content.
737 The content contains the opening slash-star (or slash-slash),
738 and for C-style comments contains the closing star-slash. For
739 C++-style comments it does not include the terminating newline. */
740 void (*comment) (cpp_reader *, location_t, const unsigned char *,
741 size_t);
742
743 /* Callback for filename remapping in __FILE__ and __BASE_FILE__ macro
744 expansions. */
745 const char *(*remap_filename) (const char*);
746
747 /* Maybe translate a #include into something else. Return a
748 cpp_buffer containing the translation if translating. */
749 char *(*translate_include) (cpp_reader *, line_maps *, location_t,
750 const char *path);
751 };
752
753 #ifdef VMS
754 #define INO_T_CPP ino_t ino[3]
755 #else
756 #define INO_T_CPP ino_t ino
757 #endif
758
759 /* Chain of directories to look for include files in. */
760 struct cpp_dir
761 {
762 /* NULL-terminated singly-linked list. */
763 struct cpp_dir *next;
764
765 /* NAME of the directory, NUL-terminated. */
766 char *name;
767 unsigned int len;
768
769 /* One if a system header, two if a system header that has extern
770 "C" guards for C++. */
771 unsigned char sysp;
772
773 /* Is this a user-supplied directory? */
774 bool user_supplied_p;
775
776 /* The canonicalized NAME as determined by lrealpath. This field
777 is only used by hosts that lack reliable inode numbers. */
778 char *canonical_name;
779
780 /* Mapping of file names for this directory for MS-DOS and related
781 platforms. A NULL-terminated array of (from, to) pairs. */
782 const char **name_map;
783
784 /* Routine to construct pathname, given the search path name and the
785 HEADER we are trying to find, return a constructed pathname to
786 try and open. If this is NULL, the constructed pathname is as
787 constructed by append_file_to_dir. */
788 char *(*construct) (const char *header, cpp_dir *dir);
789
790 /* The C front end uses these to recognize duplicated
791 directories in the search path. */
792 INO_T_CPP;
793 dev_t dev;
794 };
795
796 /* The kind of the cpp_macro. */
797 enum cpp_macro_kind {
798 cmk_macro, /* An ISO macro (token expansion). */
799 cmk_assert, /* An assertion. */
800 cmk_traditional /* A traditional macro (text expansion). */
801 };
802
803 /* Each macro definition is recorded in a cpp_macro structure.
804 Variadic macros cannot occur with traditional cpp. */
805 struct GTY(()) cpp_macro {
806 union cpp_parm_u
807 {
808 /* Parameters, if any. If parameter names use extended identifiers,
809 the original spelling of those identifiers, not the canonical
810 UTF-8 spelling, goes here. */
811 cpp_hashnode ** GTY ((tag ("false"),
812 nested_ptr (union tree_node,
813 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL",
814 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL"),
815 length ("%1.paramc"))) params;
816
817 /* If this is an assertion, the next one in the chain. */
818 cpp_macro *GTY ((tag ("true"))) next;
819 } GTY ((desc ("%1.kind == cmk_assert"))) parm;
820
821 /* Definition line number. */
822 location_t line;
823
824 /* Number of tokens in body, or bytes for traditional macros. */
825 /* Do we really need 2^32-1 range here? */
826 unsigned int count;
827
828 /* Number of parameters. */
829 unsigned short paramc;
830
831 /* Non-zero if this is a user-lazy macro, value provided by user. */
832 unsigned char lazy;
833
834 /* The kind of this macro (ISO, trad or assert) */
835 unsigned kind : 2;
836
837 /* If a function-like macro. */
838 unsigned int fun_like : 1;
839
840 /* If a variadic macro. */
841 unsigned int variadic : 1;
842
843 /* If macro defined in system header. */
844 unsigned int syshdr : 1;
845
846 /* Nonzero if it has been expanded or had its existence tested. */
847 unsigned int used : 1;
848
849 /* Indicate whether the tokens include extra CPP_PASTE tokens at the
850 end to track invalid redefinitions with consecutive CPP_PASTE
851 tokens. */
852 unsigned int extra_tokens : 1;
853
854 /* Imported C++20 macro (from a header unit). */
855 unsigned int imported_p : 1;
856
857 /* 0 bits spare (32-bit). 32 on 64-bit target. */
858
859 union cpp_exp_u
860 {
861 /* Trailing array of replacement tokens (ISO), or assertion body value. */
862 cpp_token GTY ((tag ("false"), length ("%1.count"))) tokens[1];
863
864 /* Pointer to replacement text (traditional). See comment at top
865 of cpptrad.c for how traditional function-like macros are
866 encoded. */
867 const unsigned char *GTY ((tag ("true"))) text;
868 } GTY ((desc ("%1.kind == cmk_traditional"))) exp;
869 };
870
871 /* Poisoned identifiers are flagged NODE_POISONED. NODE_OPERATOR (C++
872 only) indicates an identifier that behaves like an operator such as
873 "xor". NODE_DIAGNOSTIC is for speed in lex_token: it indicates a
874 diagnostic may be required for this node. Currently this only
875 applies to __VA_ARGS__, poisoned identifiers, and -Wc++-compat
876 warnings about NODE_OPERATOR. */
877
878 /* Hash node flags. */
879 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */
880 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */
881 #define NODE_DIAGNOSTIC (1 << 2) /* Possible diagnostic when lexed. */
882 #define NODE_WARN (1 << 3) /* Warn if redefined or undefined. */
883 #define NODE_DISABLED (1 << 4) /* A disabled macro. */
884 #define NODE_USED (1 << 5) /* Dumped with -dU. */
885 #define NODE_CONDITIONAL (1 << 6) /* Conditional macro */
886 #define NODE_WARN_OPERATOR (1 << 7) /* Warn about C++ named operator. */
887 #define NODE_MODULE (1 << 8) /* C++-20 module-related name. */
888
889 /* Different flavors of hash node. */
890 enum node_type
891 {
892 NT_VOID = 0, /* Maybe an assert? */
893 NT_MACRO_ARG, /* A macro arg. */
894 NT_USER_MACRO, /* A user macro. */
895 NT_BUILTIN_MACRO, /* A builtin macro. */
896 NT_MACRO_MASK = NT_USER_MACRO /* Mask for either macro kind. */
897 };
898
899 /* Different flavors of builtin macro. _Pragma is an operator, but we
900 handle it with the builtin code for efficiency reasons. */
901 enum cpp_builtin_type
902 {
903 BT_SPECLINE = 0, /* `__LINE__' */
904 BT_DATE, /* `__DATE__' */
905 BT_FILE, /* `__FILE__' */
906 BT_FILE_NAME, /* `__FILE_NAME__' */
907 BT_BASE_FILE, /* `__BASE_FILE__' */
908 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
909 BT_TIME, /* `__TIME__' */
910 BT_STDC, /* `__STDC__' */
911 BT_PRAGMA, /* `_Pragma' operator */
912 BT_TIMESTAMP, /* `__TIMESTAMP__' */
913 BT_COUNTER, /* `__COUNTER__' */
914 BT_HAS_ATTRIBUTE, /* `__has_attribute(x)' */
915 BT_HAS_STD_ATTRIBUTE, /* `__has_c_attribute(x)' */
916 BT_HAS_BUILTIN, /* `__has_builtin(x)' */
917 BT_HAS_INCLUDE, /* `__has_include(x)' */
918 BT_HAS_INCLUDE_NEXT /* `__has_include_next(x)' */
919 };
920
921 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE))
922 #define HT_NODE(NODE) (&(NODE)->ident)
923 #define NODE_LEN(NODE) HT_LEN (HT_NODE (NODE))
924 #define NODE_NAME(NODE) HT_STR (HT_NODE (NODE))
925
926 /* The common part of an identifier node shared amongst all 3 C front
927 ends. Also used to store CPP identifiers, which are a superset of
928 identifiers in the grammatical sense. */
929
930 union GTY(()) _cpp_hashnode_value {
931 /* Assert (maybe NULL) */
932 cpp_macro * GTY((tag ("NT_VOID"))) answers;
933 /* Macro (maybe NULL) */
934 cpp_macro * GTY((tag ("NT_USER_MACRO"))) macro;
935 /* Code for a builtin macro. */
936 enum cpp_builtin_type GTY ((tag ("NT_BUILTIN_MACRO"))) builtin;
937 /* Macro argument index. */
938 unsigned short GTY ((tag ("NT_MACRO_ARG"))) arg_index;
939 };
940
941 struct GTY(()) cpp_hashnode {
942 struct ht_identifier ident;
943 unsigned int is_directive : 1;
944 unsigned int directive_index : 7; /* If is_directive,
945 then index into directive table.
946 Otherwise, a NODE_OPERATOR. */
947 unsigned int rid_code : 8; /* Rid code - for front ends. */
948 unsigned int flags : 9; /* CPP flags. */
949 ENUM_BITFIELD(node_type) type : 2; /* CPP node type. */
950
951 /* 5 bits spare. */
952
953 /* The deferred cookie is applicable to NT_USER_MACRO or NT_VOID.
954 The latter for when a macro had a prevailing undef.
955 On a 64-bit system there would be 32-bits of padding to the value
956 field. So placing the deferred index here is not costly. */
957 unsigned deferred; /* Deferred cookie */
958
959 union _cpp_hashnode_value GTY ((desc ("%1.type"))) value;
960 };
961
962 /* A class for iterating through the source locations within a
963 string token (before escapes are interpreted, and before
964 concatenation). */
965
966 class cpp_string_location_reader {
967 public:
968 cpp_string_location_reader (location_t src_loc,
969 line_maps *line_table);
970
971 source_range get_next ();
972
973 private:
974 location_t m_loc;
975 int m_offset_per_column;
976 };
977
978 /* A class for storing the source ranges of all of the characters within
979 a string literal, after escapes are interpreted, and after
980 concatenation.
981
982 This is not GTY-marked, as instances are intended to be temporary. */
983
984 class cpp_substring_ranges
985 {
986 public:
987 cpp_substring_ranges ();
988 ~cpp_substring_ranges ();
989
get_num_ranges()990 int get_num_ranges () const { return m_num_ranges; }
get_range(int idx)991 source_range get_range (int idx) const
992 {
993 linemap_assert (idx < m_num_ranges);
994 return m_ranges[idx];
995 }
996
997 void add_range (source_range range);
998 void add_n_ranges (int num, cpp_string_location_reader &loc_reader);
999
1000 private:
1001 source_range *m_ranges;
1002 int m_num_ranges;
1003 int m_alloc_ranges;
1004 };
1005
1006 /* Call this first to get a handle to pass to other functions.
1007
1008 If you want cpplib to manage its own hashtable, pass in a NULL
1009 pointer. Otherwise you should pass in an initialized hash table
1010 that cpplib will share; this technique is used by the C front
1011 ends. */
1012 extern cpp_reader *cpp_create_reader (enum c_lang, struct ht *,
1013 class line_maps *);
1014
1015 /* Reset the cpp_reader's line_map. This is only used after reading a
1016 PCH file. */
1017 extern void cpp_set_line_map (cpp_reader *, class line_maps *);
1018
1019 /* Call this to change the selected language standard (e.g. because of
1020 command line options). */
1021 extern void cpp_set_lang (cpp_reader *, enum c_lang);
1022
1023 /* Set the include paths. */
1024 extern void cpp_set_include_chains (cpp_reader *, cpp_dir *, cpp_dir *, int);
1025
1026 /* Call these to get pointers to the options, callback, and deps
1027 structures for a given reader. These pointers are good until you
1028 call cpp_finish on that reader. You can either edit the callbacks
1029 through the pointer returned from cpp_get_callbacks, or set them
1030 with cpp_set_callbacks. */
1031 extern cpp_options *cpp_get_options (cpp_reader *) ATTRIBUTE_PURE;
1032 extern cpp_callbacks *cpp_get_callbacks (cpp_reader *) ATTRIBUTE_PURE;
1033 extern void cpp_set_callbacks (cpp_reader *, cpp_callbacks *);
1034 extern class mkdeps *cpp_get_deps (cpp_reader *) ATTRIBUTE_PURE;
1035
1036 extern const char *cpp_probe_header_unit (cpp_reader *, const char *file,
1037 bool angle_p, location_t);
1038
1039 /* Call these to get name data about the various compile-time
1040 charsets. */
1041 extern const char *cpp_get_narrow_charset_name (cpp_reader *) ATTRIBUTE_PURE;
1042 extern const char *cpp_get_wide_charset_name (cpp_reader *) ATTRIBUTE_PURE;
1043
1044 /* This function reads the file, but does not start preprocessing. It
1045 returns the name of the original file; this is the same as the
1046 input file, except for preprocessed input. This will generate at
1047 least one file change callback, and possibly a line change callback
1048 too. If there was an error opening the file, it returns NULL. */
1049 extern const char *cpp_read_main_file (cpp_reader *, const char *,
1050 bool injecting = false);
1051 extern location_t cpp_main_loc (const cpp_reader *);
1052
1053 /* Adjust for the main file to be an include. */
1054 extern void cpp_retrofit_as_include (cpp_reader *);
1055
1056 /* Set up built-ins with special behavior. Use cpp_init_builtins()
1057 instead unless your know what you are doing. */
1058 extern void cpp_init_special_builtins (cpp_reader *);
1059
1060 /* Set up built-ins like __FILE__. */
1061 extern void cpp_init_builtins (cpp_reader *, int);
1062
1063 /* This is called after options have been parsed, and partially
1064 processed. */
1065 extern void cpp_post_options (cpp_reader *);
1066
1067 /* Set up translation to the target character set. */
1068 extern void cpp_init_iconv (cpp_reader *);
1069
1070 /* Call this to finish preprocessing. If you requested dependency
1071 generation, pass an open stream to write the information to,
1072 otherwise NULL. It is your responsibility to close the stream. */
1073 extern void cpp_finish (cpp_reader *, FILE *deps_stream);
1074
1075 /* Call this to release the handle at the end of preprocessing. Any
1076 use of the handle after this function returns is invalid. */
1077 extern void cpp_destroy (cpp_reader *);
1078
1079 extern unsigned int cpp_token_len (const cpp_token *);
1080 extern unsigned char *cpp_token_as_text (cpp_reader *, const cpp_token *);
1081 extern unsigned char *cpp_spell_token (cpp_reader *, const cpp_token *,
1082 unsigned char *, bool);
1083 extern void cpp_register_pragma (cpp_reader *, const char *, const char *,
1084 void (*) (cpp_reader *), bool);
1085 extern void cpp_register_deferred_pragma (cpp_reader *, const char *,
1086 const char *, unsigned, bool, bool);
1087 extern int cpp_avoid_paste (cpp_reader *, const cpp_token *,
1088 const cpp_token *);
1089 extern const cpp_token *cpp_get_token (cpp_reader *);
1090 extern const cpp_token *cpp_get_token_with_location (cpp_reader *,
1091 location_t *);
cpp_user_macro_p(const cpp_hashnode * node)1092 inline bool cpp_user_macro_p (const cpp_hashnode *node)
1093 {
1094 return node->type == NT_USER_MACRO;
1095 }
cpp_builtin_macro_p(const cpp_hashnode * node)1096 inline bool cpp_builtin_macro_p (const cpp_hashnode *node)
1097 {
1098 return node->type == NT_BUILTIN_MACRO;
1099 }
cpp_macro_p(const cpp_hashnode * node)1100 inline bool cpp_macro_p (const cpp_hashnode *node)
1101 {
1102 return node->type & NT_MACRO_MASK;
1103 }
1104 inline cpp_macro *cpp_set_deferred_macro (cpp_hashnode *node,
1105 cpp_macro *forced = NULL)
1106 {
1107 cpp_macro *old = node->value.macro;
1108
1109 node->value.macro = forced;
1110 node->type = NT_USER_MACRO;
1111 node->flags &= ~NODE_USED;
1112
1113 return old;
1114 }
1115 cpp_macro *cpp_get_deferred_macro (cpp_reader *, cpp_hashnode *, location_t);
1116
1117 /* Returns true if NODE is a function-like user macro. */
cpp_fun_like_macro_p(cpp_hashnode * node)1118 inline bool cpp_fun_like_macro_p (cpp_hashnode *node)
1119 {
1120 return cpp_user_macro_p (node) && node->value.macro->fun_like;
1121 }
1122
1123 extern const unsigned char *cpp_macro_definition (cpp_reader *, cpp_hashnode *);
1124 extern const unsigned char *cpp_macro_definition (cpp_reader *, cpp_hashnode *,
1125 const cpp_macro *);
cpp_macro_definition_location(cpp_hashnode * node)1126 inline location_t cpp_macro_definition_location (cpp_hashnode *node)
1127 {
1128 const cpp_macro *macro = node->value.macro;
1129 return macro ? macro->line : 0;
1130 }
1131 /* Return an idempotent time stamp (possibly from SOURCE_DATE_EPOCH). */
1132 enum class CPP_time_kind
1133 {
1134 FIXED = -1, /* Fixed time via source epoch. */
1135 DYNAMIC = -2, /* Dynamic via time(2). */
1136 UNKNOWN = -3 /* Wibbly wobbly, timey wimey. */
1137 };
1138 extern CPP_time_kind cpp_get_date (cpp_reader *, time_t *);
1139
1140 extern void _cpp_backup_tokens (cpp_reader *, unsigned int);
1141 extern const cpp_token *cpp_peek_token (cpp_reader *, int);
1142
1143 /* Evaluate a CPP_*CHAR* token. */
1144 extern cppchar_t cpp_interpret_charconst (cpp_reader *, const cpp_token *,
1145 unsigned int *, int *);
1146 /* Evaluate a vector of CPP_*STRING* tokens. */
1147 extern bool cpp_interpret_string (cpp_reader *,
1148 const cpp_string *, size_t,
1149 cpp_string *, enum cpp_ttype);
1150 extern const char *cpp_interpret_string_ranges (cpp_reader *pfile,
1151 const cpp_string *from,
1152 cpp_string_location_reader *,
1153 size_t count,
1154 cpp_substring_ranges *out,
1155 enum cpp_ttype type);
1156 extern bool cpp_interpret_string_notranslate (cpp_reader *,
1157 const cpp_string *, size_t,
1158 cpp_string *, enum cpp_ttype);
1159
1160 /* Convert a host character constant to the execution character set. */
1161 extern cppchar_t cpp_host_to_exec_charset (cpp_reader *, cppchar_t);
1162
1163 /* Used to register macros and assertions, perhaps from the command line.
1164 The text is the same as the command line argument. */
1165 extern void cpp_define (cpp_reader *, const char *);
1166 extern void cpp_define_unused (cpp_reader *, const char *);
1167 extern void cpp_define_formatted (cpp_reader *pfile,
1168 const char *fmt, ...) ATTRIBUTE_PRINTF_2;
1169 extern void cpp_define_formatted_unused (cpp_reader *pfile,
1170 const char *fmt,
1171 ...) ATTRIBUTE_PRINTF_2;
1172 extern void cpp_assert (cpp_reader *, const char *);
1173 extern void cpp_undef (cpp_reader *, const char *);
1174 extern void cpp_unassert (cpp_reader *, const char *);
1175
1176 /* Mark a node as a lazily defined macro. */
1177 extern void cpp_define_lazily (cpp_reader *, cpp_hashnode *node, unsigned N);
1178
1179 /* Undefine all macros and assertions. */
1180 extern void cpp_undef_all (cpp_reader *);
1181
1182 extern cpp_buffer *cpp_push_buffer (cpp_reader *, const unsigned char *,
1183 size_t, int);
1184 extern int cpp_defined (cpp_reader *, const unsigned char *, int);
1185
1186 /* A preprocessing number. Code assumes that any unused high bits of
1187 the double integer are set to zero. */
1188
1189 /* This type has to be equal to unsigned HOST_WIDE_INT, see
1190 gcc/c-family/c-lex.c. */
1191 typedef uint64_t cpp_num_part;
1192 typedef struct cpp_num cpp_num;
1193 struct cpp_num
1194 {
1195 cpp_num_part high;
1196 cpp_num_part low;
1197 bool unsignedp; /* True if value should be treated as unsigned. */
1198 bool overflow; /* True if the most recent calculation overflowed. */
1199 };
1200
1201 /* cpplib provides two interfaces for interpretation of preprocessing
1202 numbers.
1203
1204 cpp_classify_number categorizes numeric constants according to
1205 their field (integer, floating point, or invalid), radix (decimal,
1206 octal, hexadecimal), and type suffixes. */
1207
1208 #define CPP_N_CATEGORY 0x000F
1209 #define CPP_N_INVALID 0x0000
1210 #define CPP_N_INTEGER 0x0001
1211 #define CPP_N_FLOATING 0x0002
1212
1213 #define CPP_N_WIDTH 0x00F0
1214 #define CPP_N_SMALL 0x0010 /* int, float, short _Fract/Accum */
1215 #define CPP_N_MEDIUM 0x0020 /* long, double, long _Fract/_Accum. */
1216 #define CPP_N_LARGE 0x0040 /* long long, long double,
1217 long long _Fract/Accum. */
1218
1219 #define CPP_N_WIDTH_MD 0xF0000 /* machine defined. */
1220 #define CPP_N_MD_W 0x10000
1221 #define CPP_N_MD_Q 0x20000
1222
1223 #define CPP_N_RADIX 0x0F00
1224 #define CPP_N_DECIMAL 0x0100
1225 #define CPP_N_HEX 0x0200
1226 #define CPP_N_OCTAL 0x0400
1227 #define CPP_N_BINARY 0x0800
1228
1229 #define CPP_N_UNSIGNED 0x1000 /* Properties. */
1230 #define CPP_N_IMAGINARY 0x2000
1231 #define CPP_N_DFLOAT 0x4000
1232 #define CPP_N_DEFAULT 0x8000
1233
1234 #define CPP_N_FRACT 0x100000 /* Fract types. */
1235 #define CPP_N_ACCUM 0x200000 /* Accum types. */
1236 #define CPP_N_FLOATN 0x400000 /* _FloatN types. */
1237 #define CPP_N_FLOATNX 0x800000 /* _FloatNx types. */
1238
1239 #define CPP_N_USERDEF 0x1000000 /* C++11 user-defined literal. */
1240
1241 #define CPP_N_SIZE_T 0x2000000 /* C++23 size_t literal. */
1242
1243 #define CPP_N_WIDTH_FLOATN_NX 0xF0000000 /* _FloatN / _FloatNx value
1244 of N, divided by 16. */
1245 #define CPP_FLOATN_SHIFT 24
1246 #define CPP_FLOATN_MAX 0xF0
1247
1248 /* Classify a CPP_NUMBER token. The return value is a combination of
1249 the flags from the above sets. */
1250 extern unsigned cpp_classify_number (cpp_reader *, const cpp_token *,
1251 const char **, location_t);
1252
1253 /* Return the classification flags for a float suffix. */
1254 extern unsigned int cpp_interpret_float_suffix (cpp_reader *, const char *,
1255 size_t);
1256
1257 /* Return the classification flags for an int suffix. */
1258 extern unsigned int cpp_interpret_int_suffix (cpp_reader *, const char *,
1259 size_t);
1260
1261 /* Evaluate a token classified as category CPP_N_INTEGER. */
1262 extern cpp_num cpp_interpret_integer (cpp_reader *, const cpp_token *,
1263 unsigned int);
1264
1265 /* Sign extend a number, with PRECISION significant bits and all
1266 others assumed clear, to fill out a cpp_num structure. */
1267 cpp_num cpp_num_sign_extend (cpp_num, size_t);
1268
1269 /* Output a diagnostic of some kind. */
1270 extern bool cpp_error (cpp_reader *, enum cpp_diagnostic_level,
1271 const char *msgid, ...)
1272 ATTRIBUTE_PRINTF_3;
1273 extern bool cpp_warning (cpp_reader *, enum cpp_warning_reason,
1274 const char *msgid, ...)
1275 ATTRIBUTE_PRINTF_3;
1276 extern bool cpp_pedwarning (cpp_reader *, enum cpp_warning_reason,
1277 const char *msgid, ...)
1278 ATTRIBUTE_PRINTF_3;
1279 extern bool cpp_warning_syshdr (cpp_reader *, enum cpp_warning_reason reason,
1280 const char *msgid, ...)
1281 ATTRIBUTE_PRINTF_3;
1282
1283 /* As their counterparts above, but use RICHLOC. */
1284 extern bool cpp_warning_at (cpp_reader *, enum cpp_warning_reason,
1285 rich_location *richloc, const char *msgid, ...)
1286 ATTRIBUTE_PRINTF_4;
1287 extern bool cpp_pedwarning_at (cpp_reader *, enum cpp_warning_reason,
1288 rich_location *richloc, const char *msgid, ...)
1289 ATTRIBUTE_PRINTF_4;
1290
1291 /* Output a diagnostic with "MSGID: " preceding the
1292 error string of errno. No location is printed. */
1293 extern bool cpp_errno (cpp_reader *, enum cpp_diagnostic_level,
1294 const char *msgid);
1295 /* Similarly, but with "FILENAME: " instead of "MSGID: ", where
1296 the filename is not localized. */
1297 extern bool cpp_errno_filename (cpp_reader *, enum cpp_diagnostic_level,
1298 const char *filename, location_t loc);
1299
1300 /* Same as cpp_error, except additionally specifies a position as a
1301 (translation unit) physical line and physical column. If the line is
1302 zero, then no location is printed. */
1303 extern bool cpp_error_with_line (cpp_reader *, enum cpp_diagnostic_level,
1304 location_t, unsigned,
1305 const char *msgid, ...)
1306 ATTRIBUTE_PRINTF_5;
1307 extern bool cpp_warning_with_line (cpp_reader *, enum cpp_warning_reason,
1308 location_t, unsigned,
1309 const char *msgid, ...)
1310 ATTRIBUTE_PRINTF_5;
1311 extern bool cpp_pedwarning_with_line (cpp_reader *, enum cpp_warning_reason,
1312 location_t, unsigned,
1313 const char *msgid, ...)
1314 ATTRIBUTE_PRINTF_5;
1315 extern bool cpp_warning_with_line_syshdr (cpp_reader *, enum cpp_warning_reason,
1316 location_t, unsigned,
1317 const char *msgid, ...)
1318 ATTRIBUTE_PRINTF_5;
1319
1320 extern bool cpp_error_at (cpp_reader * pfile, enum cpp_diagnostic_level,
1321 location_t src_loc, const char *msgid, ...)
1322 ATTRIBUTE_PRINTF_4;
1323
1324 extern bool cpp_error_at (cpp_reader * pfile, enum cpp_diagnostic_level,
1325 rich_location *richloc, const char *msgid, ...)
1326 ATTRIBUTE_PRINTF_4;
1327
1328 /* In lex.c */
1329 extern int cpp_ideq (const cpp_token *, const char *);
1330 extern void cpp_output_line (cpp_reader *, FILE *);
1331 extern unsigned char *cpp_output_line_to_string (cpp_reader *,
1332 const unsigned char *);
1333 extern const unsigned char *cpp_alloc_token_string
1334 (cpp_reader *, const unsigned char *, unsigned);
1335 extern void cpp_output_token (const cpp_token *, FILE *);
1336 extern const char *cpp_type2name (enum cpp_ttype, unsigned char flags);
1337 /* Returns the value of an escape sequence, truncated to the correct
1338 target precision. PSTR points to the input pointer, which is just
1339 after the backslash. LIMIT is how much text we have. WIDE is true
1340 if the escape sequence is part of a wide character constant or
1341 string literal. Handles all relevant diagnostics. */
1342 extern cppchar_t cpp_parse_escape (cpp_reader *, const unsigned char ** pstr,
1343 const unsigned char *limit, int wide);
1344
1345 /* Structure used to hold a comment block at a given location in the
1346 source code. */
1347
1348 typedef struct
1349 {
1350 /* Text of the comment including the terminators. */
1351 char *comment;
1352
1353 /* source location for the given comment. */
1354 location_t sloc;
1355 } cpp_comment;
1356
1357 /* Structure holding all comments for a given cpp_reader. */
1358
1359 typedef struct
1360 {
1361 /* table of comment entries. */
1362 cpp_comment *entries;
1363
1364 /* number of actual entries entered in the table. */
1365 int count;
1366
1367 /* number of entries allocated currently. */
1368 int allocated;
1369 } cpp_comment_table;
1370
1371 /* Returns the table of comments encountered by the preprocessor. This
1372 table is only populated when pfile->state.save_comments is true. */
1373 extern cpp_comment_table *cpp_get_comments (cpp_reader *);
1374
1375 /* In hash.c */
1376
1377 /* Lookup an identifier in the hashtable. Puts the identifier in the
1378 table if it is not already there. */
1379 extern cpp_hashnode *cpp_lookup (cpp_reader *, const unsigned char *,
1380 unsigned int);
1381
1382 typedef int (*cpp_cb) (cpp_reader *, cpp_hashnode *, void *);
1383 extern void cpp_forall_identifiers (cpp_reader *, cpp_cb, void *);
1384
1385 /* In macro.c */
1386 extern void cpp_scan_nooutput (cpp_reader *);
1387 extern int cpp_sys_macro_p (cpp_reader *);
1388 extern unsigned char *cpp_quote_string (unsigned char *, const unsigned char *,
1389 unsigned int);
1390 extern bool cpp_compare_macros (const cpp_macro *macro1,
1391 const cpp_macro *macro2);
1392
1393 /* In files.c */
1394 extern bool cpp_included (cpp_reader *, const char *);
1395 extern bool cpp_included_before (cpp_reader *, const char *, location_t);
1396 extern void cpp_make_system_header (cpp_reader *, int, int);
1397 extern bool cpp_push_include (cpp_reader *, const char *);
1398 extern bool cpp_push_default_include (cpp_reader *, const char *);
1399 extern void cpp_change_file (cpp_reader *, enum lc_reason, const char *);
1400 extern const char *cpp_get_path (struct _cpp_file *);
1401 extern cpp_dir *cpp_get_dir (struct _cpp_file *);
1402 extern cpp_buffer *cpp_get_buffer (cpp_reader *);
1403 extern struct _cpp_file *cpp_get_file (cpp_buffer *);
1404 extern cpp_buffer *cpp_get_prev (cpp_buffer *);
1405 extern void cpp_clear_file_cache (cpp_reader *);
1406
1407 /* cpp_get_converted_source returns the contents of the given file, as it exists
1408 after cpplib has read it and converted it from the input charset to the
1409 source charset. Return struct will be zero-filled if the data could not be
1410 read for any reason. The data starts at the DATA pointer, but the TO_FREE
1411 pointer is what should be passed to free(), as there may be an offset. */
1412 struct cpp_converted_source
1413 {
1414 char *to_free;
1415 char *data;
1416 size_t len;
1417 };
1418 cpp_converted_source cpp_get_converted_source (const char *fname,
1419 const char *input_charset);
1420
1421 /* In pch.c */
1422 struct save_macro_data;
1423 extern int cpp_save_state (cpp_reader *, FILE *);
1424 extern int cpp_write_pch_deps (cpp_reader *, FILE *);
1425 extern int cpp_write_pch_state (cpp_reader *, FILE *);
1426 extern int cpp_valid_state (cpp_reader *, const char *, int);
1427 extern void cpp_prepare_state (cpp_reader *, struct save_macro_data **);
1428 extern int cpp_read_state (cpp_reader *, const char *, FILE *,
1429 struct save_macro_data *);
1430
1431 /* In lex.c */
1432 extern void cpp_force_token_locations (cpp_reader *, location_t);
1433 extern void cpp_stop_forcing_token_locations (cpp_reader *);
1434 enum CPP_DO_task
1435 {
1436 CPP_DO_print,
1437 CPP_DO_location,
1438 CPP_DO_token
1439 };
1440
1441 extern void cpp_directive_only_process (cpp_reader *pfile,
1442 void *data,
1443 void (*cb) (cpp_reader *,
1444 CPP_DO_task,
1445 void *data, ...));
1446
1447 /* In expr.c */
1448 extern enum cpp_ttype cpp_userdef_string_remove_type
1449 (enum cpp_ttype type);
1450 extern enum cpp_ttype cpp_userdef_string_add_type
1451 (enum cpp_ttype type);
1452 extern enum cpp_ttype cpp_userdef_char_remove_type
1453 (enum cpp_ttype type);
1454 extern enum cpp_ttype cpp_userdef_char_add_type
1455 (enum cpp_ttype type);
1456 extern bool cpp_userdef_string_p
1457 (enum cpp_ttype type);
1458 extern bool cpp_userdef_char_p
1459 (enum cpp_ttype type);
1460 extern const char * cpp_get_userdef_suffix
1461 (const cpp_token *);
1462
1463 /* In charset.c */
1464
1465 /* The result of attempting to decode a run of UTF-8 bytes. */
1466
1467 struct cpp_decoded_char
1468 {
1469 const char *m_start_byte;
1470 const char *m_next_byte;
1471
1472 bool m_valid_ch;
1473 cppchar_t m_ch;
1474 };
1475
1476 /* Information for mapping between code points and display columns.
1477
1478 This is a tabstop value, along with a callback for getting the
1479 widths of characters. Normally this callback is cpp_wcwidth, but we
1480 support other schemes for escaping non-ASCII unicode as a series of
1481 ASCII chars when printing the user's source code in diagnostic-show-locus.c
1482
1483 For example, consider:
1484 - the Unicode character U+03C0 "GREEK SMALL LETTER PI" (UTF-8: 0xCF 0x80)
1485 - the Unicode character U+1F642 "SLIGHTLY SMILING FACE"
1486 (UTF-8: 0xF0 0x9F 0x99 0x82)
1487 - the byte 0xBF (a stray trailing byte of a UTF-8 character)
1488 Normally U+03C0 would occupy one display column, U+1F642
1489 would occupy two display columns, and the stray byte would be
1490 printed verbatim as one display column.
1491
1492 However when escaping them as unicode code points as "<U+03C0>"
1493 and "<U+1F642>" they occupy 8 and 9 display columns respectively,
1494 and when escaping them as bytes as "<CF><80>" and "<F0><9F><99><82>"
1495 they occupy 8 and 16 display columns respectively. In both cases
1496 the stray byte is escaped to <BF> as 4 display columns. */
1497
1498 struct cpp_char_column_policy
1499 {
cpp_char_column_policycpp_char_column_policy1500 cpp_char_column_policy (int tabstop,
1501 int (*width_cb) (cppchar_t c))
1502 : m_tabstop (tabstop),
1503 m_undecoded_byte_width (1),
1504 m_width_cb (width_cb)
1505 {}
1506
1507 int m_tabstop;
1508 /* Width in display columns of a stray byte that isn't decodable
1509 as UTF-8. */
1510 int m_undecoded_byte_width;
1511 int (*m_width_cb) (cppchar_t c);
1512 };
1513
1514 /* A class to manage the state while converting a UTF-8 sequence to cppchar_t
1515 and computing the display width one character at a time. */
1516 class cpp_display_width_computation {
1517 public:
1518 cpp_display_width_computation (const char *data, int data_length,
1519 const cpp_char_column_policy &policy);
next_byte()1520 const char *next_byte () const { return m_next; }
bytes_processed()1521 int bytes_processed () const { return m_next - m_begin; }
bytes_left()1522 int bytes_left () const { return m_bytes_left; }
done()1523 bool done () const { return !bytes_left (); }
display_cols_processed()1524 int display_cols_processed () const { return m_display_cols; }
1525
1526 int process_next_codepoint (cpp_decoded_char *out);
1527 int advance_display_cols (int n);
1528
1529 private:
1530 const char *const m_begin;
1531 const char *m_next;
1532 size_t m_bytes_left;
1533 const cpp_char_column_policy &m_policy;
1534 int m_display_cols;
1535 };
1536
1537 /* Convenience functions that are simple use cases for class
1538 cpp_display_width_computation. Tab characters will be expanded to spaces
1539 as determined by POLICY.m_tabstop, and non-printable-ASCII characters
1540 will be escaped as per POLICY. */
1541
1542 int cpp_byte_column_to_display_column (const char *data, int data_length,
1543 int column,
1544 const cpp_char_column_policy &policy);
cpp_display_width(const char * data,int data_length,const cpp_char_column_policy & policy)1545 inline int cpp_display_width (const char *data, int data_length,
1546 const cpp_char_column_policy &policy)
1547 {
1548 return cpp_byte_column_to_display_column (data, data_length, data_length,
1549 policy);
1550 }
1551 int cpp_display_column_to_byte_column (const char *data, int data_length,
1552 int display_col,
1553 const cpp_char_column_policy &policy);
1554 int cpp_wcwidth (cppchar_t c);
1555
1556 bool cpp_input_conversion_is_trivial (const char *input_charset);
1557 int cpp_check_utf8_bom (const char *data, size_t data_length);
1558
1559 #endif /* ! LIBCPP_CPPLIB_H */
1560