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 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. */ 323 enum cpp_bidirectional_level { 324 /* No checking. */ 325 bidirectional_none = 0, 326 /* Only detect unpaired uses of bidirectional control characters. */ 327 bidirectional_unpaired = 1, 328 /* Detect any use of bidirectional control characters. */ 329 bidirectional_any = 2, 330 /* Also warn about UCNs. */ 331 bidirectional_ucn = 4 332 }; 333 334 /* This structure is nested inside struct cpp_reader, and 335 carries all the options visible to the command line. */ 336 struct cpp_options 337 { 338 /* The language we're preprocessing. */ 339 enum c_lang lang; 340 341 /* Nonzero means use extra default include directories for C++. */ 342 unsigned char cplusplus; 343 344 /* Nonzero means handle cplusplus style comments. */ 345 unsigned char cplusplus_comments; 346 347 /* Nonzero means define __OBJC__, treat @ as a special token, use 348 the OBJC[PLUS]_INCLUDE_PATH environment variable, and allow 349 "#import". */ 350 unsigned char objc; 351 352 /* Nonzero means don't copy comments into the output file. */ 353 unsigned char discard_comments; 354 355 /* Nonzero means don't copy comments into the output file during 356 macro expansion. */ 357 unsigned char discard_comments_in_macro_exp; 358 359 /* Nonzero means process the ISO trigraph sequences. */ 360 unsigned char trigraphs; 361 362 /* Nonzero means process the ISO digraph sequences. */ 363 unsigned char digraphs; 364 365 /* Nonzero means to allow hexadecimal floats and LL suffixes. */ 366 unsigned char extended_numbers; 367 368 /* Nonzero means process u/U prefix literals (UTF-16/32). */ 369 unsigned char uliterals; 370 371 /* Nonzero means process u8 prefixed character literals (UTF-8). */ 372 unsigned char utf8_char_literals; 373 374 /* Nonzero means process r/R raw strings. If this is set, uliterals 375 must be set as well. */ 376 unsigned char rliterals; 377 378 /* Nonzero means print names of header files (-H). */ 379 unsigned char print_include_names; 380 381 /* Nonzero means complain about deprecated features. */ 382 unsigned char cpp_warn_deprecated; 383 384 /* Nonzero means warn if slash-star appears in a comment. */ 385 unsigned char warn_comments; 386 387 /* Nonzero means to warn about __DATA__, __TIME__ and __TIMESTAMP__ usage. */ 388 unsigned char warn_date_time; 389 390 /* Nonzero means warn if a user-supplied include directory does not 391 exist. */ 392 unsigned char warn_missing_include_dirs; 393 394 /* Nonzero means warn if there are any trigraphs. */ 395 unsigned char warn_trigraphs; 396 397 /* Nonzero means warn about multicharacter charconsts. */ 398 unsigned char warn_multichar; 399 400 /* Nonzero means warn about various incompatibilities with 401 traditional C. */ 402 unsigned char cpp_warn_traditional; 403 404 /* Nonzero means warn about long long numeric constants. */ 405 unsigned char cpp_warn_long_long; 406 407 /* Nonzero means warn about text after an #endif (or #else). */ 408 unsigned char warn_endif_labels; 409 410 /* Nonzero means warn about implicit sign changes owing to integer 411 promotions. */ 412 unsigned char warn_num_sign_change; 413 414 /* Zero means don't warn about __VA_ARGS__ usage in c89 pedantic mode. 415 Presumably the usage is protected by the appropriate #ifdef. */ 416 unsigned char warn_variadic_macros; 417 418 /* Nonzero means warn about builtin macros that are redefined or 419 explicitly undefined. */ 420 unsigned char warn_builtin_macro_redefined; 421 422 /* Different -Wimplicit-fallthrough= levels. */ 423 unsigned char cpp_warn_implicit_fallthrough; 424 425 /* Nonzero means we should look for header.gcc files that remap file 426 names. */ 427 unsigned char remap; 428 429 /* Zero means dollar signs are punctuation. */ 430 unsigned char dollars_in_ident; 431 432 /* Nonzero means UCNs are accepted in identifiers. */ 433 unsigned char extended_identifiers; 434 435 /* True if we should warn about dollars in identifiers or numbers 436 for this translation unit. */ 437 unsigned char warn_dollars; 438 439 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */ 440 unsigned char warn_undef; 441 442 /* Nonzero means warn if "defined" is encountered in a place other than 443 an #if. */ 444 unsigned char warn_expansion_to_defined; 445 446 /* Nonzero means warn of unused macros from the main file. */ 447 unsigned char warn_unused_macros; 448 449 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */ 450 unsigned char c99; 451 452 /* Nonzero if we are conforming to a specific C or C++ standard. */ 453 unsigned char std; 454 455 /* Nonzero means give all the error messages the ANSI standard requires. */ 456 unsigned char cpp_pedantic; 457 458 /* Nonzero means we're looking at already preprocessed code, so don't 459 bother trying to do macro expansion and whatnot. */ 460 unsigned char preprocessed; 461 462 /* Nonzero means we are going to emit debugging logs during 463 preprocessing. */ 464 unsigned char debug; 465 466 /* Nonzero means we are tracking locations of tokens involved in 467 macro expansion. 1 Means we track the location in degraded mode 468 where we do not track locations of tokens resulting from the 469 expansion of arguments of function-like macro. 2 Means we do 470 track all macro expansions. This last option is the one that 471 consumes the highest amount of memory. */ 472 unsigned char track_macro_expansion; 473 474 /* Nonzero means handle C++ alternate operator names. */ 475 unsigned char operator_names; 476 477 /* Nonzero means warn about use of C++ alternate operator names. */ 478 unsigned char warn_cxx_operator_names; 479 480 /* True for traditional preprocessing. */ 481 unsigned char traditional; 482 483 /* Nonzero for C++ 2011 Standard user-defined literals. */ 484 unsigned char user_literals; 485 486 /* Nonzero means warn when a string or character literal is followed by a 487 ud-suffix which does not beging with an underscore. */ 488 unsigned char warn_literal_suffix; 489 490 /* Nonzero means interpret imaginary, fixed-point, or other gnu extension 491 literal number suffixes as user-defined literal number suffixes. */ 492 unsigned char ext_numeric_literals; 493 494 /* Nonzero means extended identifiers allow the characters specified 495 in C11. */ 496 unsigned char c11_identifiers; 497 498 /* Nonzero for C++ 2014 Standard binary constants. */ 499 unsigned char binary_constants; 500 501 /* Nonzero for C++ 2014 Standard digit separators. */ 502 unsigned char digit_separators; 503 504 /* Nonzero for C2X decimal floating-point constants. */ 505 unsigned char dfp_constants; 506 507 /* Nonzero for C++20 __VA_OPT__ feature. */ 508 unsigned char va_opt; 509 510 /* Nonzero for the '::' token. */ 511 unsigned char scope; 512 513 /* Nonzero for the '#elifdef' and '#elifndef' directives. */ 514 unsigned char elifdef; 515 516 /* Nonzero means tokenize C++20 module directives. */ 517 unsigned char module_directives; 518 519 /* Nonzero for C++23 size_t literals. */ 520 unsigned char size_t_literals; 521 522 /* Holds the name of the target (execution) character set. */ 523 const char *narrow_charset; 524 525 /* Holds the name of the target wide character set. */ 526 const char *wide_charset; 527 528 /* Holds the name of the input character set. */ 529 const char *input_charset; 530 531 /* The minimum permitted level of normalization before a warning 532 is generated. See enum cpp_normalize_level. */ 533 int warn_normalize; 534 535 /* True to warn about precompiled header files we couldn't use. */ 536 bool warn_invalid_pch; 537 538 /* True if dependencies should be restored from a precompiled header. */ 539 bool restore_pch_deps; 540 541 /* True if warn about differences between C90 and C99. */ 542 signed char cpp_warn_c90_c99_compat; 543 544 /* True if warn about differences between C11 and C2X. */ 545 signed char cpp_warn_c11_c2x_compat; 546 547 /* True if warn about differences between C++98 and C++11. */ 548 bool cpp_warn_cxx11_compat; 549 550 /* Nonzero if bidirectional control characters checking is on. See enum 551 cpp_bidirectional_level. */ 552 unsigned char cpp_warn_bidirectional; 553 554 /* Dependency generation. */ 555 struct 556 { 557 /* Style of header dependencies to generate. */ 558 enum cpp_deps_style style; 559 560 /* Assume missing files are generated files. */ 561 bool missing_files; 562 563 /* Generate phony targets for each dependency apart from the first 564 one. */ 565 bool phony_targets; 566 567 /* Generate dependency info for modules. */ 568 bool modules; 569 570 /* If true, no dependency is generated on the main file. */ 571 bool ignore_main_file; 572 573 /* If true, intend to use the preprocessor output (e.g., for compilation) 574 in addition to the dependency info. */ 575 bool need_preprocessor_output; 576 } deps; 577 578 /* Target-specific features set by the front end or client. */ 579 580 /* Precision for target CPP arithmetic, target characters, target 581 ints and target wide characters, respectively. */ 582 size_t precision, char_precision, int_precision, wchar_precision; 583 584 /* True means chars (wide chars) are unsigned. */ 585 bool unsigned_char, unsigned_wchar; 586 587 /* True if the most significant byte in a word has the lowest 588 address in memory. */ 589 bool bytes_big_endian; 590 591 /* Nonzero means __STDC__ should have the value 0 in system headers. */ 592 unsigned char stdc_0_in_system_headers; 593 594 /* True disables tokenization outside of preprocessing directives. */ 595 bool directives_only; 596 597 /* True enables canonicalization of system header file paths. */ 598 bool canonical_system_headers; 599 600 /* The maximum depth of the nested #include. */ 601 unsigned int max_include_depth; 602 603 cpp_main_search main_search : 8; 604 }; 605 606 /* Diagnostic levels. To get a diagnostic without associating a 607 position in the translation unit with it, use cpp_error_with_line 608 with a line number of zero. */ 609 610 enum cpp_diagnostic_level { 611 /* Warning, an error with -Werror. */ 612 CPP_DL_WARNING = 0, 613 /* Same as CPP_DL_WARNING, except it is not suppressed in system headers. */ 614 CPP_DL_WARNING_SYSHDR, 615 /* Warning, an error with -pedantic-errors or -Werror. */ 616 CPP_DL_PEDWARN, 617 /* An error. */ 618 CPP_DL_ERROR, 619 /* An internal consistency check failed. Prints "internal error: ", 620 otherwise the same as CPP_DL_ERROR. */ 621 CPP_DL_ICE, 622 /* An informative note following a warning. */ 623 CPP_DL_NOTE, 624 /* A fatal error. */ 625 CPP_DL_FATAL 626 }; 627 628 /* Warning reason codes. Use a reason code of CPP_W_NONE for unclassified 629 warnings and diagnostics that are not warnings. */ 630 631 enum cpp_warning_reason { 632 CPP_W_NONE = 0, 633 CPP_W_DEPRECATED, 634 CPP_W_COMMENTS, 635 CPP_W_MISSING_INCLUDE_DIRS, 636 CPP_W_TRIGRAPHS, 637 CPP_W_MULTICHAR, 638 CPP_W_TRADITIONAL, 639 CPP_W_LONG_LONG, 640 CPP_W_ENDIF_LABELS, 641 CPP_W_NUM_SIGN_CHANGE, 642 CPP_W_VARIADIC_MACROS, 643 CPP_W_BUILTIN_MACRO_REDEFINED, 644 CPP_W_DOLLARS, 645 CPP_W_UNDEF, 646 CPP_W_UNUSED_MACROS, 647 CPP_W_CXX_OPERATOR_NAMES, 648 CPP_W_NORMALIZE, 649 CPP_W_INVALID_PCH, 650 CPP_W_WARNING_DIRECTIVE, 651 CPP_W_LITERAL_SUFFIX, 652 CPP_W_SIZE_T_LITERALS, 653 CPP_W_DATE_TIME, 654 CPP_W_PEDANTIC, 655 CPP_W_C90_C99_COMPAT, 656 CPP_W_C11_C2X_COMPAT, 657 CPP_W_CXX11_COMPAT, 658 CPP_W_EXPANSION_TO_DEFINED, 659 CPP_W_BIDIRECTIONAL 660 }; 661 662 /* Callback for header lookup for HEADER, which is the name of a 663 source file. It is used as a method of last resort to find headers 664 that are not otherwise found during the normal include processing. 665 The return value is the malloced name of a header to try and open, 666 if any, or NULL otherwise. This callback is called only if the 667 header is otherwise unfound. */ 668 typedef const char *(*missing_header_cb)(cpp_reader *, const char *header, cpp_dir **); 669 670 /* Call backs to cpplib client. */ 671 struct cpp_callbacks 672 { 673 /* Called when a new line of preprocessed output is started. */ 674 void (*line_change) (cpp_reader *, const cpp_token *, int); 675 676 /* Called when switching to/from a new file. 677 The line_map is for the new file. It is NULL if there is no new file. 678 (In C this happens when done with <built-in>+<command line> and also 679 when done with a main file.) This can be used for resource cleanup. */ 680 void (*file_change) (cpp_reader *, const line_map_ordinary *); 681 682 void (*dir_change) (cpp_reader *, const char *); 683 void (*include) (cpp_reader *, location_t, const unsigned char *, 684 const char *, int, const cpp_token **); 685 void (*define) (cpp_reader *, location_t, cpp_hashnode *); 686 void (*undef) (cpp_reader *, location_t, cpp_hashnode *); 687 void (*ident) (cpp_reader *, location_t, const cpp_string *); 688 void (*def_pragma) (cpp_reader *, location_t); 689 int (*valid_pch) (cpp_reader *, const char *, int); 690 void (*read_pch) (cpp_reader *, const char *, int, const char *); 691 missing_header_cb missing_header; 692 693 /* Context-sensitive macro support. Returns macro (if any) that should 694 be expanded. */ 695 cpp_hashnode * (*macro_to_expand) (cpp_reader *, const cpp_token *); 696 697 /* Called to emit a diagnostic. This callback receives the 698 translated message. */ 699 bool (*diagnostic) (cpp_reader *, 700 enum cpp_diagnostic_level, 701 enum cpp_warning_reason, 702 rich_location *, 703 const char *, va_list *) 704 ATTRIBUTE_FPTR_PRINTF(5,0); 705 706 /* Callbacks for when a macro is expanded, or tested (whether 707 defined or not at the time) in #ifdef, #ifndef or "defined". */ 708 void (*used_define) (cpp_reader *, location_t, cpp_hashnode *); 709 void (*used_undef) (cpp_reader *, location_t, cpp_hashnode *); 710 /* Called before #define and #undef or other macro definition 711 changes are processed. */ 712 void (*before_define) (cpp_reader *); 713 /* Called whenever a macro is expanded or tested. 714 Second argument is the location of the start of the current expansion. */ 715 void (*used) (cpp_reader *, location_t, cpp_hashnode *); 716 717 /* Callback to identify whether an attribute exists. */ 718 int (*has_attribute) (cpp_reader *, bool); 719 720 /* Callback to determine whether a built-in function is recognized. */ 721 int (*has_builtin) (cpp_reader *); 722 723 /* Callback that can change a user lazy into normal macro. */ 724 void (*user_lazy_macro) (cpp_reader *, cpp_macro *, unsigned); 725 726 /* Callback to handle deferred cpp_macros. */ 727 cpp_macro *(*user_deferred_macro) (cpp_reader *, location_t, cpp_hashnode *); 728 729 /* Callback to parse SOURCE_DATE_EPOCH from environment. */ 730 time_t (*get_source_date_epoch) (cpp_reader *); 731 732 /* Callback for providing suggestions for misspelled directives. */ 733 const char *(*get_suggestion) (cpp_reader *, const char *, const char *const *); 734 735 /* Callback for when a comment is encountered, giving the location 736 of the opening slash, a pointer to the content (which is not 737 necessarily 0-terminated), and the length of the content. 738 The content contains the opening slash-star (or slash-slash), 739 and for C-style comments contains the closing star-slash. For 740 C++-style comments it does not include the terminating newline. */ 741 void (*comment) (cpp_reader *, location_t, const unsigned char *, 742 size_t); 743 744 /* Callback for filename remapping in __FILE__ and __BASE_FILE__ macro 745 expansions. */ 746 const char *(*remap_filename) (const char*); 747 748 /* Maybe translate a #include into something else. Return a 749 cpp_buffer containing the translation if translating. */ 750 char *(*translate_include) (cpp_reader *, line_maps *, location_t, 751 const char *path); 752 }; 753 754 #ifdef VMS 755 #define INO_T_CPP ino_t ino[3] 756 #elif defined (_AIX) && SIZEOF_INO_T == 4 757 #define INO_T_CPP ino64_t ino 758 #else 759 #define INO_T_CPP ino_t ino 760 #endif 761 762 #if defined (_AIX) && SIZEOF_DEV_T == 4 763 #define DEV_T_CPP dev64_t dev 764 #else 765 #define DEV_T_CPP dev_t dev 766 #endif 767 768 /* Chain of directories to look for include files in. */ 769 struct cpp_dir 770 { 771 /* NULL-terminated singly-linked list. */ 772 struct cpp_dir *next; 773 774 /* NAME of the directory, NUL-terminated. */ 775 char *name; 776 unsigned int len; 777 778 /* One if a system header, two if a system header that has extern 779 "C" guards for C++. */ 780 unsigned char sysp; 781 782 /* Is this a user-supplied directory? */ 783 bool user_supplied_p; 784 785 /* The canonicalized NAME as determined by lrealpath. This field 786 is only used by hosts that lack reliable inode numbers. */ 787 char *canonical_name; 788 789 /* Mapping of file names for this directory for MS-DOS and related 790 platforms. A NULL-terminated array of (from, to) pairs. */ 791 const char **name_map; 792 793 /* Routine to construct pathname, given the search path name and the 794 HEADER we are trying to find, return a constructed pathname to 795 try and open. If this is NULL, the constructed pathname is as 796 constructed by append_file_to_dir. */ 797 char *(*construct) (const char *header, cpp_dir *dir); 798 799 /* The C front end uses these to recognize duplicated 800 directories in the search path. */ 801 INO_T_CPP; 802 DEV_T_CPP; 803 }; 804 805 /* The kind of the cpp_macro. */ 806 enum cpp_macro_kind { 807 cmk_macro, /* An ISO macro (token expansion). */ 808 cmk_assert, /* An assertion. */ 809 cmk_traditional /* A traditional macro (text expansion). */ 810 }; 811 812 /* Each macro definition is recorded in a cpp_macro structure. 813 Variadic macros cannot occur with traditional cpp. */ 814 struct GTY(()) cpp_macro { 815 union cpp_parm_u 816 { 817 /* Parameters, if any. If parameter names use extended identifiers, 818 the original spelling of those identifiers, not the canonical 819 UTF-8 spelling, goes here. */ 820 cpp_hashnode ** GTY ((tag ("false"), 821 nested_ptr (union tree_node, 822 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL", 823 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL"), 824 length ("%1.paramc"))) params; 825 826 /* If this is an assertion, the next one in the chain. */ 827 cpp_macro *GTY ((tag ("true"))) next; 828 } GTY ((desc ("%1.kind == cmk_assert"))) parm; 829 830 /* Definition line number. */ 831 location_t line; 832 833 /* Number of tokens in body, or bytes for traditional macros. */ 834 /* Do we really need 2^32-1 range here? */ 835 unsigned int count; 836 837 /* Number of parameters. */ 838 unsigned short paramc; 839 840 /* Non-zero if this is a user-lazy macro, value provided by user. */ 841 unsigned char lazy; 842 843 /* The kind of this macro (ISO, trad or assert) */ 844 unsigned kind : 2; 845 846 /* If a function-like macro. */ 847 unsigned int fun_like : 1; 848 849 /* If a variadic macro. */ 850 unsigned int variadic : 1; 851 852 /* If macro defined in system header. */ 853 unsigned int syshdr : 1; 854 855 /* Nonzero if it has been expanded or had its existence tested. */ 856 unsigned int used : 1; 857 858 /* Indicate whether the tokens include extra CPP_PASTE tokens at the 859 end to track invalid redefinitions with consecutive CPP_PASTE 860 tokens. */ 861 unsigned int extra_tokens : 1; 862 863 /* Imported C++20 macro (from a header unit). */ 864 unsigned int imported_p : 1; 865 866 /* 0 bits spare (32-bit). 32 on 64-bit target. */ 867 868 union cpp_exp_u 869 { 870 /* Trailing array of replacement tokens (ISO), or assertion body value. */ 871 cpp_token GTY ((tag ("false"), length ("%1.count"))) tokens[1]; 872 873 /* Pointer to replacement text (traditional). See comment at top 874 of cpptrad.c for how traditional function-like macros are 875 encoded. */ 876 const unsigned char *GTY ((tag ("true"))) text; 877 } GTY ((desc ("%1.kind == cmk_traditional"))) exp; 878 }; 879 880 /* Poisoned identifiers are flagged NODE_POISONED. NODE_OPERATOR (C++ 881 only) indicates an identifier that behaves like an operator such as 882 "xor". NODE_DIAGNOSTIC is for speed in lex_token: it indicates a 883 diagnostic may be required for this node. Currently this only 884 applies to __VA_ARGS__, poisoned identifiers, and -Wc++-compat 885 warnings about NODE_OPERATOR. */ 886 887 /* Hash node flags. */ 888 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */ 889 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */ 890 #define NODE_DIAGNOSTIC (1 << 2) /* Possible diagnostic when lexed. */ 891 #define NODE_WARN (1 << 3) /* Warn if redefined or undefined. */ 892 #define NODE_DISABLED (1 << 4) /* A disabled macro. */ 893 #define NODE_USED (1 << 5) /* Dumped with -dU. */ 894 #define NODE_CONDITIONAL (1 << 6) /* Conditional macro */ 895 #define NODE_WARN_OPERATOR (1 << 7) /* Warn about C++ named operator. */ 896 #define NODE_MODULE (1 << 8) /* C++-20 module-related name. */ 897 898 /* Different flavors of hash node. */ 899 enum node_type 900 { 901 NT_VOID = 0, /* Maybe an assert? */ 902 NT_MACRO_ARG, /* A macro arg. */ 903 NT_USER_MACRO, /* A user macro. */ 904 NT_BUILTIN_MACRO, /* A builtin macro. */ 905 NT_MACRO_MASK = NT_USER_MACRO /* Mask for either macro kind. */ 906 }; 907 908 /* Different flavors of builtin macro. _Pragma is an operator, but we 909 handle it with the builtin code for efficiency reasons. */ 910 enum cpp_builtin_type 911 { 912 BT_SPECLINE = 0, /* `__LINE__' */ 913 BT_DATE, /* `__DATE__' */ 914 BT_FILE, /* `__FILE__' */ 915 BT_FILE_NAME, /* `__FILE_NAME__' */ 916 BT_BASE_FILE, /* `__BASE_FILE__' */ 917 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */ 918 BT_TIME, /* `__TIME__' */ 919 BT_STDC, /* `__STDC__' */ 920 BT_PRAGMA, /* `_Pragma' operator */ 921 BT_TIMESTAMP, /* `__TIMESTAMP__' */ 922 BT_COUNTER, /* `__COUNTER__' */ 923 BT_HAS_ATTRIBUTE, /* `__has_attribute(x)' */ 924 BT_HAS_STD_ATTRIBUTE, /* `__has_c_attribute(x)' */ 925 BT_HAS_BUILTIN, /* `__has_builtin(x)' */ 926 BT_HAS_INCLUDE, /* `__has_include(x)' */ 927 BT_HAS_INCLUDE_NEXT /* `__has_include_next(x)' */ 928 }; 929 930 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE)) 931 #define HT_NODE(NODE) (&(NODE)->ident) 932 #define NODE_LEN(NODE) HT_LEN (HT_NODE (NODE)) 933 #define NODE_NAME(NODE) HT_STR (HT_NODE (NODE)) 934 935 /* The common part of an identifier node shared amongst all 3 C front 936 ends. Also used to store CPP identifiers, which are a superset of 937 identifiers in the grammatical sense. */ 938 939 union GTY(()) _cpp_hashnode_value { 940 /* Assert (maybe NULL) */ 941 cpp_macro * GTY((tag ("NT_VOID"))) answers; 942 /* Macro (maybe NULL) */ 943 cpp_macro * GTY((tag ("NT_USER_MACRO"))) macro; 944 /* Code for a builtin macro. */ 945 enum cpp_builtin_type GTY ((tag ("NT_BUILTIN_MACRO"))) builtin; 946 /* Macro argument index. */ 947 unsigned short GTY ((tag ("NT_MACRO_ARG"))) arg_index; 948 }; 949 950 struct GTY(()) cpp_hashnode { 951 struct ht_identifier ident; 952 unsigned int is_directive : 1; 953 unsigned int directive_index : 7; /* If is_directive, 954 then index into directive table. 955 Otherwise, a NODE_OPERATOR. */ 956 unsigned int rid_code : 8; /* Rid code - for front ends. */ 957 unsigned int flags : 9; /* CPP flags. */ 958 ENUM_BITFIELD(node_type) type : 2; /* CPP node type. */ 959 960 /* 5 bits spare. */ 961 962 /* The deferred cookie is applicable to NT_USER_MACRO or NT_VOID. 963 The latter for when a macro had a prevailing undef. 964 On a 64-bit system there would be 32-bits of padding to the value 965 field. So placing the deferred index here is not costly. */ 966 unsigned deferred; /* Deferred cookie */ 967 968 union _cpp_hashnode_value GTY ((desc ("%1.type"))) value; 969 }; 970 971 /* A class for iterating through the source locations within a 972 string token (before escapes are interpreted, and before 973 concatenation). */ 974 975 class cpp_string_location_reader { 976 public: 977 cpp_string_location_reader (location_t src_loc, 978 line_maps *line_table); 979 980 source_range get_next (); 981 982 private: 983 location_t m_loc; 984 int m_offset_per_column; 985 }; 986 987 /* A class for storing the source ranges of all of the characters within 988 a string literal, after escapes are interpreted, and after 989 concatenation. 990 991 This is not GTY-marked, as instances are intended to be temporary. */ 992 993 class cpp_substring_ranges 994 { 995 public: 996 cpp_substring_ranges (); 997 ~cpp_substring_ranges (); 998 999 int get_num_ranges () const { return m_num_ranges; } 1000 source_range get_range (int idx) const 1001 { 1002 linemap_assert (idx < m_num_ranges); 1003 return m_ranges[idx]; 1004 } 1005 1006 void add_range (source_range range); 1007 void add_n_ranges (int num, cpp_string_location_reader &loc_reader); 1008 1009 private: 1010 source_range *m_ranges; 1011 int m_num_ranges; 1012 int m_alloc_ranges; 1013 }; 1014 1015 /* Call this first to get a handle to pass to other functions. 1016 1017 If you want cpplib to manage its own hashtable, pass in a NULL 1018 pointer. Otherwise you should pass in an initialized hash table 1019 that cpplib will share; this technique is used by the C front 1020 ends. */ 1021 extern cpp_reader *cpp_create_reader (enum c_lang, struct ht *, 1022 class line_maps *); 1023 1024 /* Reset the cpp_reader's line_map. This is only used after reading a 1025 PCH file. */ 1026 extern void cpp_set_line_map (cpp_reader *, class line_maps *); 1027 1028 /* Call this to change the selected language standard (e.g. because of 1029 command line options). */ 1030 extern void cpp_set_lang (cpp_reader *, enum c_lang); 1031 1032 /* Set the include paths. */ 1033 extern void cpp_set_include_chains (cpp_reader *, cpp_dir *, cpp_dir *, int); 1034 1035 /* Provide src:dst pair for __FILE__ remapping. */ 1036 extern void add_cpp_remap_path (const char *); 1037 1038 /* Call these to get pointers to the options, callback, and deps 1039 structures for a given reader. These pointers are good until you 1040 call cpp_finish on that reader. You can either edit the callbacks 1041 through the pointer returned from cpp_get_callbacks, or set them 1042 with cpp_set_callbacks. */ 1043 extern cpp_options *cpp_get_options (cpp_reader *) ATTRIBUTE_PURE; 1044 extern cpp_callbacks *cpp_get_callbacks (cpp_reader *) ATTRIBUTE_PURE; 1045 extern void cpp_set_callbacks (cpp_reader *, cpp_callbacks *); 1046 extern class mkdeps *cpp_get_deps (cpp_reader *) ATTRIBUTE_PURE; 1047 1048 extern const char *cpp_probe_header_unit (cpp_reader *, const char *file, 1049 bool angle_p, location_t); 1050 1051 /* Call these to get name data about the various compile-time 1052 charsets. */ 1053 extern const char *cpp_get_narrow_charset_name (cpp_reader *) ATTRIBUTE_PURE; 1054 extern const char *cpp_get_wide_charset_name (cpp_reader *) ATTRIBUTE_PURE; 1055 1056 /* This function reads the file, but does not start preprocessing. It 1057 returns the name of the original file; this is the same as the 1058 input file, except for preprocessed input. This will generate at 1059 least one file change callback, and possibly a line change callback 1060 too. If there was an error opening the file, it returns NULL. */ 1061 extern const char *cpp_read_main_file (cpp_reader *, const char *, 1062 bool injecting = false); 1063 extern location_t cpp_main_loc (const cpp_reader *); 1064 1065 /* Adjust for the main file to be an include. */ 1066 extern void cpp_retrofit_as_include (cpp_reader *); 1067 1068 /* Set up built-ins with special behavior. Use cpp_init_builtins() 1069 instead unless your know what you are doing. */ 1070 extern void cpp_init_special_builtins (cpp_reader *); 1071 1072 /* Set up built-ins like __FILE__. */ 1073 extern void cpp_init_builtins (cpp_reader *, int); 1074 1075 /* This is called after options have been parsed, and partially 1076 processed. */ 1077 extern void cpp_post_options (cpp_reader *); 1078 1079 /* Set up translation to the target character set. */ 1080 extern void cpp_init_iconv (cpp_reader *); 1081 1082 /* Call this to finish preprocessing. If you requested dependency 1083 generation, pass an open stream to write the information to, 1084 otherwise NULL. It is your responsibility to close the stream. */ 1085 extern void cpp_finish (cpp_reader *, FILE *deps_stream); 1086 1087 /* Call this to release the handle at the end of preprocessing. Any 1088 use of the handle after this function returns is invalid. */ 1089 extern void cpp_destroy (cpp_reader *); 1090 1091 extern unsigned int cpp_token_len (const cpp_token *); 1092 extern unsigned char *cpp_token_as_text (cpp_reader *, const cpp_token *); 1093 extern unsigned char *cpp_spell_token (cpp_reader *, const cpp_token *, 1094 unsigned char *, bool); 1095 extern void cpp_register_pragma (cpp_reader *, const char *, const char *, 1096 void (*) (cpp_reader *), bool); 1097 extern void cpp_register_deferred_pragma (cpp_reader *, const char *, 1098 const char *, unsigned, bool, bool); 1099 extern int cpp_avoid_paste (cpp_reader *, const cpp_token *, 1100 const cpp_token *); 1101 extern const cpp_token *cpp_get_token (cpp_reader *); 1102 extern const cpp_token *cpp_get_token_with_location (cpp_reader *, 1103 location_t *); 1104 inline bool cpp_user_macro_p (const cpp_hashnode *node) 1105 { 1106 return node->type == NT_USER_MACRO; 1107 } 1108 inline bool cpp_builtin_macro_p (const cpp_hashnode *node) 1109 { 1110 return node->type == NT_BUILTIN_MACRO; 1111 } 1112 inline bool cpp_macro_p (const cpp_hashnode *node) 1113 { 1114 return node->type & NT_MACRO_MASK; 1115 } 1116 inline cpp_macro *cpp_set_deferred_macro (cpp_hashnode *node, 1117 cpp_macro *forced = NULL) 1118 { 1119 cpp_macro *old = node->value.macro; 1120 1121 node->value.macro = forced; 1122 node->type = NT_USER_MACRO; 1123 node->flags &= ~NODE_USED; 1124 1125 return old; 1126 } 1127 cpp_macro *cpp_get_deferred_macro (cpp_reader *, cpp_hashnode *, location_t); 1128 1129 /* Returns true if NODE is a function-like user macro. */ 1130 inline bool cpp_fun_like_macro_p (cpp_hashnode *node) 1131 { 1132 return cpp_user_macro_p (node) && node->value.macro->fun_like; 1133 } 1134 1135 extern const unsigned char *cpp_macro_definition (cpp_reader *, cpp_hashnode *); 1136 extern const unsigned char *cpp_macro_definition (cpp_reader *, cpp_hashnode *, 1137 const cpp_macro *); 1138 inline location_t cpp_macro_definition_location (cpp_hashnode *node) 1139 { 1140 const cpp_macro *macro = node->value.macro; 1141 return macro ? macro->line : 0; 1142 } 1143 /* Return an idempotent time stamp (possibly from SOURCE_DATE_EPOCH). */ 1144 enum class CPP_time_kind 1145 { 1146 FIXED = -1, /* Fixed time via source epoch. */ 1147 DYNAMIC = -2, /* Dynamic via time(2). */ 1148 UNKNOWN = -3 /* Wibbly wobbly, timey wimey. */ 1149 }; 1150 extern CPP_time_kind cpp_get_date (cpp_reader *, time_t *); 1151 1152 extern void _cpp_backup_tokens (cpp_reader *, unsigned int); 1153 extern const cpp_token *cpp_peek_token (cpp_reader *, int); 1154 1155 /* Evaluate a CPP_*CHAR* token. */ 1156 extern cppchar_t cpp_interpret_charconst (cpp_reader *, const cpp_token *, 1157 unsigned int *, int *); 1158 /* Evaluate a vector of CPP_*STRING* tokens. */ 1159 extern bool cpp_interpret_string (cpp_reader *, 1160 const cpp_string *, size_t, 1161 cpp_string *, enum cpp_ttype); 1162 extern const char *cpp_interpret_string_ranges (cpp_reader *pfile, 1163 const cpp_string *from, 1164 cpp_string_location_reader *, 1165 size_t count, 1166 cpp_substring_ranges *out, 1167 enum cpp_ttype type); 1168 extern bool cpp_interpret_string_notranslate (cpp_reader *, 1169 const cpp_string *, size_t, 1170 cpp_string *, enum cpp_ttype); 1171 1172 /* Convert a host character constant to the execution character set. */ 1173 extern cppchar_t cpp_host_to_exec_charset (cpp_reader *, cppchar_t); 1174 1175 /* Used to register macros and assertions, perhaps from the command line. 1176 The text is the same as the command line argument. */ 1177 extern void cpp_define (cpp_reader *, const char *); 1178 extern void cpp_define_unused (cpp_reader *, const char *); 1179 extern void cpp_define_formatted (cpp_reader *pfile, 1180 const char *fmt, ...) ATTRIBUTE_PRINTF_2; 1181 extern void cpp_define_formatted_unused (cpp_reader *pfile, 1182 const char *fmt, 1183 ...) ATTRIBUTE_PRINTF_2; 1184 extern void cpp_assert (cpp_reader *, const char *); 1185 extern void cpp_undef (cpp_reader *, const char *); 1186 extern void cpp_unassert (cpp_reader *, const char *); 1187 1188 /* Mark a node as a lazily defined macro. */ 1189 extern void cpp_define_lazily (cpp_reader *, cpp_hashnode *node, unsigned N); 1190 1191 /* Undefine all macros and assertions. */ 1192 extern void cpp_undef_all (cpp_reader *); 1193 1194 extern cpp_buffer *cpp_push_buffer (cpp_reader *, const unsigned char *, 1195 size_t, int); 1196 extern int cpp_defined (cpp_reader *, const unsigned char *, int); 1197 1198 /* A preprocessing number. Code assumes that any unused high bits of 1199 the double integer are set to zero. */ 1200 1201 /* This type has to be equal to unsigned HOST_WIDE_INT, see 1202 gcc/c-family/c-lex.cc. */ 1203 typedef uint64_t cpp_num_part; 1204 typedef struct cpp_num cpp_num; 1205 struct cpp_num 1206 { 1207 cpp_num_part high; 1208 cpp_num_part low; 1209 bool unsignedp; /* True if value should be treated as unsigned. */ 1210 bool overflow; /* True if the most recent calculation overflowed. */ 1211 }; 1212 1213 /* cpplib provides two interfaces for interpretation of preprocessing 1214 numbers. 1215 1216 cpp_classify_number categorizes numeric constants according to 1217 their field (integer, floating point, or invalid), radix (decimal, 1218 octal, hexadecimal), and type suffixes. */ 1219 1220 #define CPP_N_CATEGORY 0x000F 1221 #define CPP_N_INVALID 0x0000 1222 #define CPP_N_INTEGER 0x0001 1223 #define CPP_N_FLOATING 0x0002 1224 1225 #define CPP_N_WIDTH 0x00F0 1226 #define CPP_N_SMALL 0x0010 /* int, float, short _Fract/Accum */ 1227 #define CPP_N_MEDIUM 0x0020 /* long, double, long _Fract/_Accum. */ 1228 #define CPP_N_LARGE 0x0040 /* long long, long double, 1229 long long _Fract/Accum. */ 1230 1231 #define CPP_N_WIDTH_MD 0xF0000 /* machine defined. */ 1232 #define CPP_N_MD_W 0x10000 1233 #define CPP_N_MD_Q 0x20000 1234 1235 #define CPP_N_RADIX 0x0F00 1236 #define CPP_N_DECIMAL 0x0100 1237 #define CPP_N_HEX 0x0200 1238 #define CPP_N_OCTAL 0x0400 1239 #define CPP_N_BINARY 0x0800 1240 1241 #define CPP_N_UNSIGNED 0x1000 /* Properties. */ 1242 #define CPP_N_IMAGINARY 0x2000 1243 #define CPP_N_DFLOAT 0x4000 1244 #define CPP_N_DEFAULT 0x8000 1245 1246 #define CPP_N_FRACT 0x100000 /* Fract types. */ 1247 #define CPP_N_ACCUM 0x200000 /* Accum types. */ 1248 #define CPP_N_FLOATN 0x400000 /* _FloatN types. */ 1249 #define CPP_N_FLOATNX 0x800000 /* _FloatNx types. */ 1250 1251 #define CPP_N_USERDEF 0x1000000 /* C++11 user-defined literal. */ 1252 1253 #define CPP_N_SIZE_T 0x2000000 /* C++23 size_t literal. */ 1254 1255 #define CPP_N_WIDTH_FLOATN_NX 0xF0000000 /* _FloatN / _FloatNx value 1256 of N, divided by 16. */ 1257 #define CPP_FLOATN_SHIFT 24 1258 #define CPP_FLOATN_MAX 0xF0 1259 1260 /* Classify a CPP_NUMBER token. The return value is a combination of 1261 the flags from the above sets. */ 1262 extern unsigned cpp_classify_number (cpp_reader *, const cpp_token *, 1263 const char **, location_t); 1264 1265 /* Return the classification flags for a float suffix. */ 1266 extern unsigned int cpp_interpret_float_suffix (cpp_reader *, const char *, 1267 size_t); 1268 1269 /* Return the classification flags for an int suffix. */ 1270 extern unsigned int cpp_interpret_int_suffix (cpp_reader *, const char *, 1271 size_t); 1272 1273 /* Evaluate a token classified as category CPP_N_INTEGER. */ 1274 extern cpp_num cpp_interpret_integer (cpp_reader *, const cpp_token *, 1275 unsigned int); 1276 1277 /* Sign extend a number, with PRECISION significant bits and all 1278 others assumed clear, to fill out a cpp_num structure. */ 1279 cpp_num cpp_num_sign_extend (cpp_num, size_t); 1280 1281 /* Output a diagnostic of some kind. */ 1282 extern bool cpp_error (cpp_reader *, enum cpp_diagnostic_level, 1283 const char *msgid, ...) 1284 ATTRIBUTE_PRINTF_3; 1285 extern bool cpp_warning (cpp_reader *, enum cpp_warning_reason, 1286 const char *msgid, ...) 1287 ATTRIBUTE_PRINTF_3; 1288 extern bool cpp_pedwarning (cpp_reader *, enum cpp_warning_reason, 1289 const char *msgid, ...) 1290 ATTRIBUTE_PRINTF_3; 1291 extern bool cpp_warning_syshdr (cpp_reader *, enum cpp_warning_reason reason, 1292 const char *msgid, ...) 1293 ATTRIBUTE_PRINTF_3; 1294 1295 /* As their counterparts above, but use RICHLOC. */ 1296 extern bool cpp_warning_at (cpp_reader *, enum cpp_warning_reason, 1297 rich_location *richloc, const char *msgid, ...) 1298 ATTRIBUTE_PRINTF_4; 1299 extern bool cpp_pedwarning_at (cpp_reader *, enum cpp_warning_reason, 1300 rich_location *richloc, const char *msgid, ...) 1301 ATTRIBUTE_PRINTF_4; 1302 1303 /* Output a diagnostic with "MSGID: " preceding the 1304 error string of errno. No location is printed. */ 1305 extern bool cpp_errno (cpp_reader *, enum cpp_diagnostic_level, 1306 const char *msgid); 1307 /* Similarly, but with "FILENAME: " instead of "MSGID: ", where 1308 the filename is not localized. */ 1309 extern bool cpp_errno_filename (cpp_reader *, enum cpp_diagnostic_level, 1310 const char *filename, location_t loc); 1311 1312 /* Same as cpp_error, except additionally specifies a position as a 1313 (translation unit) physical line and physical column. If the line is 1314 zero, then no location is printed. */ 1315 extern bool cpp_error_with_line (cpp_reader *, enum cpp_diagnostic_level, 1316 location_t, unsigned, 1317 const char *msgid, ...) 1318 ATTRIBUTE_PRINTF_5; 1319 extern bool cpp_warning_with_line (cpp_reader *, enum cpp_warning_reason, 1320 location_t, unsigned, 1321 const char *msgid, ...) 1322 ATTRIBUTE_PRINTF_5; 1323 extern bool cpp_pedwarning_with_line (cpp_reader *, enum cpp_warning_reason, 1324 location_t, unsigned, 1325 const char *msgid, ...) 1326 ATTRIBUTE_PRINTF_5; 1327 extern bool cpp_warning_with_line_syshdr (cpp_reader *, enum cpp_warning_reason, 1328 location_t, unsigned, 1329 const char *msgid, ...) 1330 ATTRIBUTE_PRINTF_5; 1331 1332 extern bool cpp_error_at (cpp_reader * pfile, enum cpp_diagnostic_level, 1333 location_t src_loc, const char *msgid, ...) 1334 ATTRIBUTE_PRINTF_4; 1335 1336 extern bool cpp_error_at (cpp_reader * pfile, enum cpp_diagnostic_level, 1337 rich_location *richloc, const char *msgid, ...) 1338 ATTRIBUTE_PRINTF_4; 1339 1340 /* In lex.cc */ 1341 extern int cpp_ideq (const cpp_token *, const char *); 1342 extern void cpp_output_line (cpp_reader *, FILE *); 1343 extern unsigned char *cpp_output_line_to_string (cpp_reader *, 1344 const unsigned char *); 1345 extern const unsigned char *cpp_alloc_token_string 1346 (cpp_reader *, const unsigned char *, unsigned); 1347 extern void cpp_output_token (const cpp_token *, FILE *); 1348 extern const char *cpp_type2name (enum cpp_ttype, unsigned char flags); 1349 /* Returns the value of an escape sequence, truncated to the correct 1350 target precision. PSTR points to the input pointer, which is just 1351 after the backslash. LIMIT is how much text we have. WIDE is true 1352 if the escape sequence is part of a wide character constant or 1353 string literal. Handles all relevant diagnostics. */ 1354 extern cppchar_t cpp_parse_escape (cpp_reader *, const unsigned char ** pstr, 1355 const unsigned char *limit, int wide); 1356 1357 /* Structure used to hold a comment block at a given location in the 1358 source code. */ 1359 1360 typedef struct 1361 { 1362 /* Text of the comment including the terminators. */ 1363 char *comment; 1364 1365 /* source location for the given comment. */ 1366 location_t sloc; 1367 } cpp_comment; 1368 1369 /* Structure holding all comments for a given cpp_reader. */ 1370 1371 typedef struct 1372 { 1373 /* table of comment entries. */ 1374 cpp_comment *entries; 1375 1376 /* number of actual entries entered in the table. */ 1377 int count; 1378 1379 /* number of entries allocated currently. */ 1380 int allocated; 1381 } cpp_comment_table; 1382 1383 /* Returns the table of comments encountered by the preprocessor. This 1384 table is only populated when pfile->state.save_comments is true. */ 1385 extern cpp_comment_table *cpp_get_comments (cpp_reader *); 1386 1387 /* In hash.c */ 1388 1389 /* Lookup an identifier in the hashtable. Puts the identifier in the 1390 table if it is not already there. */ 1391 extern cpp_hashnode *cpp_lookup (cpp_reader *, const unsigned char *, 1392 unsigned int); 1393 1394 typedef int (*cpp_cb) (cpp_reader *, cpp_hashnode *, void *); 1395 extern void cpp_forall_identifiers (cpp_reader *, cpp_cb, void *); 1396 1397 /* In macro.cc */ 1398 extern void cpp_scan_nooutput (cpp_reader *); 1399 extern int cpp_sys_macro_p (cpp_reader *); 1400 extern unsigned char *cpp_quote_string (unsigned char *, const unsigned char *, 1401 unsigned int); 1402 extern bool cpp_compare_macros (const cpp_macro *macro1, 1403 const cpp_macro *macro2); 1404 1405 /* In files.cc */ 1406 extern bool cpp_included (cpp_reader *, const char *); 1407 extern bool cpp_included_before (cpp_reader *, const char *, location_t); 1408 extern void cpp_make_system_header (cpp_reader *, int, int); 1409 extern bool cpp_push_include (cpp_reader *, const char *); 1410 extern bool cpp_push_default_include (cpp_reader *, const char *); 1411 extern void cpp_change_file (cpp_reader *, enum lc_reason, const char *); 1412 extern const char *cpp_get_path (struct _cpp_file *); 1413 extern cpp_dir *cpp_get_dir (struct _cpp_file *); 1414 extern cpp_buffer *cpp_get_buffer (cpp_reader *); 1415 extern struct _cpp_file *cpp_get_file (cpp_buffer *); 1416 extern cpp_buffer *cpp_get_prev (cpp_buffer *); 1417 extern void cpp_clear_file_cache (cpp_reader *); 1418 1419 /* cpp_get_converted_source returns the contents of the given file, as it exists 1420 after cpplib has read it and converted it from the input charset to the 1421 source charset. Return struct will be zero-filled if the data could not be 1422 read for any reason. The data starts at the DATA pointer, but the TO_FREE 1423 pointer is what should be passed to free(), as there may be an offset. */ 1424 struct cpp_converted_source 1425 { 1426 char *to_free; 1427 char *data; 1428 size_t len; 1429 }; 1430 cpp_converted_source cpp_get_converted_source (const char *fname, 1431 const char *input_charset); 1432 1433 /* In pch.cc */ 1434 struct save_macro_data; 1435 extern int cpp_save_state (cpp_reader *, FILE *); 1436 extern int cpp_write_pch_deps (cpp_reader *, FILE *); 1437 extern int cpp_write_pch_state (cpp_reader *, FILE *); 1438 extern int cpp_valid_state (cpp_reader *, const char *, int); 1439 extern void cpp_prepare_state (cpp_reader *, struct save_macro_data **); 1440 extern int cpp_read_state (cpp_reader *, const char *, FILE *, 1441 struct save_macro_data *); 1442 1443 /* In lex.cc */ 1444 extern void cpp_force_token_locations (cpp_reader *, location_t); 1445 extern void cpp_stop_forcing_token_locations (cpp_reader *); 1446 enum CPP_DO_task 1447 { 1448 CPP_DO_print, 1449 CPP_DO_location, 1450 CPP_DO_token 1451 }; 1452 1453 extern void cpp_directive_only_process (cpp_reader *pfile, 1454 void *data, 1455 void (*cb) (cpp_reader *, 1456 CPP_DO_task, 1457 void *data, ...)); 1458 1459 /* In expr.cc */ 1460 extern enum cpp_ttype cpp_userdef_string_remove_type 1461 (enum cpp_ttype type); 1462 extern enum cpp_ttype cpp_userdef_string_add_type 1463 (enum cpp_ttype type); 1464 extern enum cpp_ttype cpp_userdef_char_remove_type 1465 (enum cpp_ttype type); 1466 extern enum cpp_ttype cpp_userdef_char_add_type 1467 (enum cpp_ttype type); 1468 extern bool cpp_userdef_string_p 1469 (enum cpp_ttype type); 1470 extern bool cpp_userdef_char_p 1471 (enum cpp_ttype type); 1472 extern const char * cpp_get_userdef_suffix 1473 (const cpp_token *); 1474 1475 /* In charset.cc */ 1476 1477 /* The result of attempting to decode a run of UTF-8 bytes. */ 1478 1479 struct cpp_decoded_char 1480 { 1481 const char *m_start_byte; 1482 const char *m_next_byte; 1483 1484 bool m_valid_ch; 1485 cppchar_t m_ch; 1486 }; 1487 1488 /* Information for mapping between code points and display columns. 1489 1490 This is a tabstop value, along with a callback for getting the 1491 widths of characters. Normally this callback is cpp_wcwidth, but we 1492 support other schemes for escaping non-ASCII unicode as a series of 1493 ASCII chars when printing the user's source code in diagnostic-show-locus.cc 1494 1495 For example, consider: 1496 - the Unicode character U+03C0 "GREEK SMALL LETTER PI" (UTF-8: 0xCF 0x80) 1497 - the Unicode character U+1F642 "SLIGHTLY SMILING FACE" 1498 (UTF-8: 0xF0 0x9F 0x99 0x82) 1499 - the byte 0xBF (a stray trailing byte of a UTF-8 character) 1500 Normally U+03C0 would occupy one display column, U+1F642 1501 would occupy two display columns, and the stray byte would be 1502 printed verbatim as one display column. 1503 1504 However when escaping them as unicode code points as "<U+03C0>" 1505 and "<U+1F642>" they occupy 8 and 9 display columns respectively, 1506 and when escaping them as bytes as "<CF><80>" and "<F0><9F><99><82>" 1507 they occupy 8 and 16 display columns respectively. In both cases 1508 the stray byte is escaped to <BF> as 4 display columns. */ 1509 1510 struct cpp_char_column_policy 1511 { 1512 cpp_char_column_policy (int tabstop, 1513 int (*width_cb) (cppchar_t c)) 1514 : m_tabstop (tabstop), 1515 m_undecoded_byte_width (1), 1516 m_width_cb (width_cb) 1517 {} 1518 1519 int m_tabstop; 1520 /* Width in display columns of a stray byte that isn't decodable 1521 as UTF-8. */ 1522 int m_undecoded_byte_width; 1523 int (*m_width_cb) (cppchar_t c); 1524 }; 1525 1526 /* A class to manage the state while converting a UTF-8 sequence to cppchar_t 1527 and computing the display width one character at a time. */ 1528 class cpp_display_width_computation { 1529 public: 1530 cpp_display_width_computation (const char *data, int data_length, 1531 const cpp_char_column_policy &policy); 1532 const char *next_byte () const { return m_next; } 1533 int bytes_processed () const { return m_next - m_begin; } 1534 int bytes_left () const { return m_bytes_left; } 1535 bool done () const { return !bytes_left (); } 1536 int display_cols_processed () const { return m_display_cols; } 1537 1538 int process_next_codepoint (cpp_decoded_char *out); 1539 int advance_display_cols (int n); 1540 1541 private: 1542 const char *const m_begin; 1543 const char *m_next; 1544 size_t m_bytes_left; 1545 const cpp_char_column_policy &m_policy; 1546 int m_display_cols; 1547 }; 1548 1549 /* Convenience functions that are simple use cases for class 1550 cpp_display_width_computation. Tab characters will be expanded to spaces 1551 as determined by POLICY.m_tabstop, and non-printable-ASCII characters 1552 will be escaped as per POLICY. */ 1553 1554 int cpp_byte_column_to_display_column (const char *data, int data_length, 1555 int column, 1556 const cpp_char_column_policy &policy); 1557 inline int cpp_display_width (const char *data, int data_length, 1558 const cpp_char_column_policy &policy) 1559 { 1560 return cpp_byte_column_to_display_column (data, data_length, data_length, 1561 policy); 1562 } 1563 int cpp_display_column_to_byte_column (const char *data, int data_length, 1564 int display_col, 1565 const cpp_char_column_policy &policy); 1566 int cpp_wcwidth (cppchar_t c); 1567 1568 bool cpp_input_conversion_is_trivial (const char *input_charset); 1569 int cpp_check_utf8_bom (const char *data, size_t data_length); 1570 1571 #endif /* ! LIBCPP_CPPLIB_H */ 1572