1 /* Definitions for CPP library. 2 Copyright (C) 1995-2013 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 answer; 40 struct _cpp_file; 41 42 /* The first three groups, apart from '=', can appear in preprocessor 43 expressions (+= and -= are used to indicate unary + and - resp.). 44 This allows a lookup table to be implemented in _cpp_parse_expr. 45 46 The first group, to CPP_LAST_EQ, can be immediately followed by an 47 '='. The lexer needs operators ending in '=', like ">>=", to be in 48 the same order as their counterparts without the '=', like ">>". 49 50 See the cpp_operator table optab in expr.c if you change the order or 51 add or remove anything in the first group. */ 52 53 #define TTYPE_TABLE \ 54 OP(EQ, "=") \ 55 OP(NOT, "!") \ 56 OP(GREATER, ">") /* compare */ \ 57 OP(LESS, "<") \ 58 OP(PLUS, "+") /* math */ \ 59 OP(MINUS, "-") \ 60 OP(MULT, "*") \ 61 OP(DIV, "/") \ 62 OP(MOD, "%") \ 63 OP(AND, "&") /* bit ops */ \ 64 OP(OR, "|") \ 65 OP(XOR, "^") \ 66 OP(RSHIFT, ">>") \ 67 OP(LSHIFT, "<<") \ 68 \ 69 OP(COMPL, "~") \ 70 OP(AND_AND, "&&") /* logical */ \ 71 OP(OR_OR, "||") \ 72 OP(QUERY, "?") \ 73 OP(COLON, ":") \ 74 OP(COMMA, ",") /* grouping */ \ 75 OP(OPEN_PAREN, "(") \ 76 OP(CLOSE_PAREN, ")") \ 77 TK(EOF, NONE) \ 78 OP(EQ_EQ, "==") /* compare */ \ 79 OP(NOT_EQ, "!=") \ 80 OP(GREATER_EQ, ">=") \ 81 OP(LESS_EQ, "<=") \ 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(OTHER, LITERAL) /* stray punctuation */ \ 123 \ 124 TK(STRING, LITERAL) /* "string" */ \ 125 TK(WSTRING, LITERAL) /* L"string" */ \ 126 TK(STRING16, LITERAL) /* u"string" */ \ 127 TK(STRING32, LITERAL) /* U"string" */ \ 128 TK(UTF8STRING, LITERAL) /* u8"string" */ \ 129 TK(OBJC_STRING, LITERAL) /* @"string" - Objective-C */ \ 130 TK(HEADER_NAME, LITERAL) /* <stdio.h> in #include */ \ 131 \ 132 TK(CHAR_USERDEF, LITERAL) /* 'char'_suffix - C++-0x */ \ 133 TK(WCHAR_USERDEF, LITERAL) /* L'char'_suffix - C++-0x */ \ 134 TK(CHAR16_USERDEF, LITERAL) /* u'char'_suffix - C++-0x */ \ 135 TK(CHAR32_USERDEF, LITERAL) /* U'char'_suffix - C++-0x */ \ 136 TK(STRING_USERDEF, LITERAL) /* "string"_suffix - C++-0x */ \ 137 TK(WSTRING_USERDEF, LITERAL) /* L"string"_suffix - C++-0x */ \ 138 TK(STRING16_USERDEF, LITERAL) /* u"string"_suffix - C++-0x */ \ 139 TK(STRING32_USERDEF, LITERAL) /* U"string"_suffix - C++-0x */ \ 140 TK(UTF8STRING_USERDEF,LITERAL) /* u8"string"_suffix - C++-0x */ \ 141 \ 142 TK(COMMENT, LITERAL) /* Only if output comments. */ \ 143 /* SPELL_LITERAL happens to DTRT. */ \ 144 TK(MACRO_ARG, NONE) /* Macro argument. */ \ 145 TK(PRAGMA, NONE) /* Only for deferred pragmas. */ \ 146 TK(PRAGMA_EOL, NONE) /* End-of-line for deferred pragmas. */ \ 147 TK(PADDING, NONE) /* Whitespace for -E. */ 148 149 #define OP(e, s) CPP_ ## e, 150 #define TK(e, s) CPP_ ## e, 151 enum cpp_ttype 152 { 153 TTYPE_TABLE 154 N_TTYPES, 155 156 /* Positions in the table. */ 157 CPP_LAST_EQ = CPP_LSHIFT, 158 CPP_FIRST_DIGRAPH = CPP_HASH, 159 CPP_LAST_PUNCTUATOR= CPP_ATSIGN, 160 CPP_LAST_CPP_OP = CPP_LESS_EQ 161 }; 162 #undef OP 163 #undef TK 164 165 /* C language kind, used when calling cpp_create_reader. */ 166 enum c_lang {CLK_GNUC89 = 0, CLK_GNUC99, CLK_GNUC11, 167 CLK_STDC89, CLK_STDC94, CLK_STDC99, CLK_STDC11, 168 CLK_GNUCXX, CLK_CXX98, CLK_GNUCXX11, CLK_CXX11, CLK_ASM}; 169 170 /* Payload of a NUMBER, STRING, CHAR or COMMENT token. */ 171 struct GTY(()) cpp_string { 172 unsigned int len; 173 const unsigned char *text; 174 }; 175 176 /* Flags for the cpp_token structure. */ 177 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */ 178 #define DIGRAPH (1 << 1) /* If it was a digraph. */ 179 #define STRINGIFY_ARG (1 << 2) /* If macro argument to be stringified. */ 180 #define PASTE_LEFT (1 << 3) /* If on LHS of a ## operator. */ 181 #define NAMED_OP (1 << 4) /* C++ named operators. */ 182 #define NO_EXPAND (1 << 5) /* Do not macro-expand this token. */ 183 #define BOL (1 << 6) /* Token at beginning of line. */ 184 #define PURE_ZERO (1 << 7) /* Single 0 digit, used by the C++ frontend, 185 set in c-lex.c. */ 186 #define SP_DIGRAPH (1 << 8) /* # or ## token was a digraph. */ 187 #define SP_PREV_WHITE (1 << 9) /* If whitespace before a ## 188 operator, or before this token 189 after a # operator. */ 190 191 /* Specify which field, if any, of the cpp_token union is used. */ 192 193 enum cpp_token_fld_kind { 194 CPP_TOKEN_FLD_NODE, 195 CPP_TOKEN_FLD_SOURCE, 196 CPP_TOKEN_FLD_STR, 197 CPP_TOKEN_FLD_ARG_NO, 198 CPP_TOKEN_FLD_TOKEN_NO, 199 CPP_TOKEN_FLD_PRAGMA, 200 CPP_TOKEN_FLD_NONE 201 }; 202 203 /* A macro argument in the cpp_token union. */ 204 struct GTY(()) cpp_macro_arg { 205 /* Argument number. */ 206 unsigned int arg_no; 207 }; 208 209 /* An identifier in the cpp_token union. */ 210 struct GTY(()) cpp_identifier { 211 /* The canonical (UTF-8) spelling of the identifier. */ 212 cpp_hashnode * 213 GTY ((nested_ptr (union tree_node, 214 "%h ? CPP_HASHNODE (GCC_IDENT_TO_HT_IDENT (%h)) : NULL", 215 "%h ? HT_IDENT_TO_GCC_IDENT (HT_NODE (%h)) : NULL"))) 216 node; 217 }; 218 219 /* A preprocessing token. This has been carefully packed and should 220 occupy 16 bytes on 32-bit hosts and 24 bytes on 64-bit hosts. */ 221 struct GTY(()) cpp_token { 222 source_location src_loc; /* Location of first char of token. */ 223 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */ 224 unsigned short flags; /* flags - see above */ 225 226 union cpp_token_u 227 { 228 /* An identifier. */ 229 struct cpp_identifier GTY ((tag ("CPP_TOKEN_FLD_NODE"))) node; 230 231 /* Inherit padding from this token. */ 232 cpp_token * GTY ((tag ("CPP_TOKEN_FLD_SOURCE"))) source; 233 234 /* A string, or number. */ 235 struct cpp_string GTY ((tag ("CPP_TOKEN_FLD_STR"))) str; 236 237 /* Argument no. for a CPP_MACRO_ARG. */ 238 struct cpp_macro_arg GTY ((tag ("CPP_TOKEN_FLD_ARG_NO"))) macro_arg; 239 240 /* Original token no. for a CPP_PASTE (from a sequence of 241 consecutive paste tokens in a macro expansion). */ 242 unsigned int GTY ((tag ("CPP_TOKEN_FLD_TOKEN_NO"))) token_no; 243 244 /* Caller-supplied identifier for a CPP_PRAGMA. */ 245 unsigned int GTY ((tag ("CPP_TOKEN_FLD_PRAGMA"))) pragma; 246 } GTY ((desc ("cpp_token_val_index (&%1)"))) val; 247 }; 248 249 /* Say which field is in use. */ 250 extern enum cpp_token_fld_kind cpp_token_val_index (cpp_token *tok); 251 252 /* A type wide enough to hold any multibyte source character. 253 cpplib's character constant interpreter requires an unsigned type. 254 Also, a typedef for the signed equivalent. 255 The width of this type is capped at 32 bits; there do exist targets 256 where wchar_t is 64 bits, but only in a non-default mode, and there 257 would be no meaningful interpretation for a wchar_t value greater 258 than 2^32 anyway -- the widest wide-character encoding around is 259 ISO 10646, which stops at 2^31. */ 260 #if CHAR_BIT * SIZEOF_INT >= 32 261 # define CPPCHAR_SIGNED_T int 262 #elif CHAR_BIT * SIZEOF_LONG >= 32 263 # define CPPCHAR_SIGNED_T long 264 #else 265 # error "Cannot find a least-32-bit signed integer type" 266 #endif 267 typedef unsigned CPPCHAR_SIGNED_T cppchar_t; 268 typedef CPPCHAR_SIGNED_T cppchar_signed_t; 269 270 /* Style of header dependencies to generate. */ 271 enum cpp_deps_style { DEPS_NONE = 0, DEPS_USER, DEPS_SYSTEM }; 272 273 /* The possible normalization levels, from most restrictive to least. */ 274 enum cpp_normalize_level { 275 /* In NFKC. */ 276 normalized_KC = 0, 277 /* In NFC. */ 278 normalized_C, 279 /* In NFC, except for subsequences where being in NFC would make 280 the identifier invalid. */ 281 normalized_identifier_C, 282 /* Not normalized at all. */ 283 normalized_none 284 }; 285 286 /* This structure is nested inside struct cpp_reader, and 287 carries all the options visible to the command line. */ 288 struct cpp_options 289 { 290 /* Characters between tab stops. */ 291 unsigned int tabstop; 292 293 /* The language we're preprocessing. */ 294 enum c_lang lang; 295 296 /* Nonzero means use extra default include directories for C++. */ 297 unsigned char cplusplus; 298 299 /* Nonzero means handle cplusplus style comments. */ 300 unsigned char cplusplus_comments; 301 302 /* Nonzero means define __OBJC__, treat @ as a special token, use 303 the OBJC[PLUS]_INCLUDE_PATH environment variable, and allow 304 "#import". */ 305 unsigned char objc; 306 307 /* Nonzero means don't copy comments into the output file. */ 308 unsigned char discard_comments; 309 310 /* Nonzero means don't copy comments into the output file during 311 macro expansion. */ 312 unsigned char discard_comments_in_macro_exp; 313 314 /* Nonzero means process the ISO trigraph sequences. */ 315 unsigned char trigraphs; 316 317 /* Nonzero means process the ISO digraph sequences. */ 318 unsigned char digraphs; 319 320 /* Nonzero means to allow hexadecimal floats and LL suffixes. */ 321 unsigned char extended_numbers; 322 323 /* Nonzero means process u/U prefix literals (UTF-16/32). */ 324 unsigned char uliterals; 325 326 /* Nonzero means process r/R raw strings. If this is set, uliterals 327 must be set as well. */ 328 unsigned char rliterals; 329 330 /* Nonzero means print names of header files (-H). */ 331 unsigned char print_include_names; 332 333 /* Nonzero means complain about deprecated features. */ 334 unsigned char cpp_warn_deprecated; 335 336 /* Nonzero means warn if slash-star appears in a comment. */ 337 unsigned char warn_comments; 338 339 /* Nonzero means warn if a user-supplied include directory does not 340 exist. */ 341 unsigned char warn_missing_include_dirs; 342 343 /* Nonzero means warn if there are any trigraphs. */ 344 unsigned char warn_trigraphs; 345 346 /* Nonzero means warn about multicharacter charconsts. */ 347 unsigned char warn_multichar; 348 349 /* Nonzero means warn about various incompatibilities with 350 traditional C. */ 351 unsigned char cpp_warn_traditional; 352 353 /* Nonzero means warn about long long numeric constants. */ 354 unsigned char cpp_warn_long_long; 355 356 /* Nonzero means warn about text after an #endif (or #else). */ 357 unsigned char warn_endif_labels; 358 359 /* Nonzero means warn about implicit sign changes owing to integer 360 promotions. */ 361 unsigned char warn_num_sign_change; 362 363 /* Zero means don't warn about __VA_ARGS__ usage in c89 pedantic mode. 364 Presumably the usage is protected by the appropriate #ifdef. */ 365 unsigned char warn_variadic_macros; 366 367 /* Nonzero means warn about builtin macros that are redefined or 368 explicitly undefined. */ 369 unsigned char warn_builtin_macro_redefined; 370 371 /* Nonzero means we should look for header.gcc files that remap file 372 names. */ 373 unsigned char remap; 374 375 /* Zero means dollar signs are punctuation. */ 376 unsigned char dollars_in_ident; 377 378 /* Nonzero means UCNs are accepted in identifiers. */ 379 unsigned char extended_identifiers; 380 381 /* True if we should warn about dollars in identifiers or numbers 382 for this translation unit. */ 383 unsigned char warn_dollars; 384 385 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */ 386 unsigned char warn_undef; 387 388 /* Nonzero means warn of unused macros from the main file. */ 389 unsigned char warn_unused_macros; 390 391 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */ 392 unsigned char c99; 393 394 /* Nonzero if we are conforming to a specific C or C++ standard. */ 395 unsigned char std; 396 397 /* Nonzero means give all the error messages the ANSI standard requires. */ 398 unsigned char cpp_pedantic; 399 400 /* Nonzero means we're looking at already preprocessed code, so don't 401 bother trying to do macro expansion and whatnot. */ 402 unsigned char preprocessed; 403 404 /* Nonzero means we are going to emit debugging logs during 405 preprocessing. */ 406 unsigned char debug; 407 408 /* Nonzero means we are tracking locations of tokens involved in 409 macro expansion. 1 Means we track the location in degraded mode 410 where we do not track locations of tokens resulting from the 411 expansion of arguments of function-like macro. 2 Means we do 412 track all macro expansions. This last option is the one that 413 consumes the highest amount of memory. */ 414 unsigned char track_macro_expansion; 415 416 /* Nonzero means handle C++ alternate operator names. */ 417 unsigned char operator_names; 418 419 /* Nonzero means warn about use of C++ alternate operator names. */ 420 unsigned char warn_cxx_operator_names; 421 422 /* True for traditional preprocessing. */ 423 unsigned char traditional; 424 425 /* Nonzero for C++ 2011 Standard user-defnied literals. */ 426 unsigned char user_literals; 427 428 /* Nonzero means warn when a string or character literal is followed by a 429 ud-suffix which does not beging with an underscore. */ 430 unsigned char warn_literal_suffix; 431 432 /* Nonzero means interpret imaginary, fixed-point, or other gnu extension 433 literal number suffixes as user-defined literal number suffixes. */ 434 unsigned char ext_numeric_literals; 435 436 /* Holds the name of the target (execution) character set. */ 437 const char *narrow_charset; 438 439 /* Holds the name of the target wide character set. */ 440 const char *wide_charset; 441 442 /* Holds the name of the input character set. */ 443 const char *input_charset; 444 445 /* The minimum permitted level of normalization before a warning 446 is generated. */ 447 enum cpp_normalize_level warn_normalize; 448 449 /* True to warn about precompiled header files we couldn't use. */ 450 bool warn_invalid_pch; 451 452 /* True if dependencies should be restored from a precompiled header. */ 453 bool restore_pch_deps; 454 455 /* Dependency generation. */ 456 struct 457 { 458 /* Style of header dependencies to generate. */ 459 enum cpp_deps_style style; 460 461 /* Assume missing files are generated files. */ 462 bool missing_files; 463 464 /* Generate phony targets for each dependency apart from the first 465 one. */ 466 bool phony_targets; 467 468 /* If true, no dependency is generated on the main file. */ 469 bool ignore_main_file; 470 471 /* If true, intend to use the preprocessor output (e.g., for compilation) 472 in addition to the dependency info. */ 473 bool need_preprocessor_output; 474 } deps; 475 476 /* Target-specific features set by the front end or client. */ 477 478 /* Precision for target CPP arithmetic, target characters, target 479 ints and target wide characters, respectively. */ 480 size_t precision, char_precision, int_precision, wchar_precision; 481 482 /* True means chars (wide chars) are unsigned. */ 483 bool unsigned_char, unsigned_wchar; 484 485 /* True if the most significant byte in a word has the lowest 486 address in memory. */ 487 bool bytes_big_endian; 488 489 /* Nonzero means __STDC__ should have the value 0 in system headers. */ 490 unsigned char stdc_0_in_system_headers; 491 492 /* True disables tokenization outside of preprocessing directives. */ 493 bool directives_only; 494 495 /* True enables canonicalization of system header file paths. */ 496 bool canonical_system_headers; 497 }; 498 499 /* Callback for header lookup for HEADER, which is the name of a 500 source file. It is used as a method of last resort to find headers 501 that are not otherwise found during the normal include processing. 502 The return value is the malloced name of a header to try and open, 503 if any, or NULL otherwise. This callback is called only if the 504 header is otherwise unfound. */ 505 typedef const char *(*missing_header_cb)(cpp_reader *, const char *header, cpp_dir **); 506 507 /* Call backs to cpplib client. */ 508 struct cpp_callbacks 509 { 510 /* Called when a new line of preprocessed output is started. */ 511 void (*line_change) (cpp_reader *, const cpp_token *, int); 512 513 /* Called when switching to/from a new file. 514 The line_map is for the new file. It is NULL if there is no new file. 515 (In C this happens when done with <built-in>+<command line> and also 516 when done with a main file.) This can be used for resource cleanup. */ 517 void (*file_change) (cpp_reader *, const struct line_map *); 518 519 void (*dir_change) (cpp_reader *, const char *); 520 void (*include) (cpp_reader *, source_location, const unsigned char *, 521 const char *, int, const cpp_token **); 522 void (*define) (cpp_reader *, source_location, cpp_hashnode *); 523 void (*undef) (cpp_reader *, source_location, cpp_hashnode *); 524 void (*ident) (cpp_reader *, source_location, const cpp_string *); 525 void (*def_pragma) (cpp_reader *, source_location); 526 int (*valid_pch) (cpp_reader *, const char *, int); 527 void (*read_pch) (cpp_reader *, const char *, int, const char *); 528 missing_header_cb missing_header; 529 530 /* Context-sensitive macro support. Returns macro (if any) that should 531 be expanded. */ 532 cpp_hashnode * (*macro_to_expand) (cpp_reader *, const cpp_token *); 533 534 /* Called to emit a diagnostic. This callback receives the 535 translated message. */ 536 bool (*error) (cpp_reader *, int, int, source_location, unsigned int, 537 const char *, va_list *) 538 ATTRIBUTE_FPTR_PRINTF(6,0); 539 540 /* Callbacks for when a macro is expanded, or tested (whether 541 defined or not at the time) in #ifdef, #ifndef or "defined". */ 542 void (*used_define) (cpp_reader *, source_location, cpp_hashnode *); 543 void (*used_undef) (cpp_reader *, source_location, cpp_hashnode *); 544 /* Called before #define and #undef or other macro definition 545 changes are processed. */ 546 void (*before_define) (cpp_reader *); 547 /* Called whenever a macro is expanded or tested. 548 Second argument is the location of the start of the current expansion. */ 549 void (*used) (cpp_reader *, source_location, cpp_hashnode *); 550 551 /* Callback that can change a user builtin into normal macro. */ 552 bool (*user_builtin_macro) (cpp_reader *, cpp_hashnode *); 553 }; 554 555 #ifdef VMS 556 #define INO_T_CPP ino_t ino[3] 557 #else 558 #define INO_T_CPP ino_t ino 559 #endif 560 561 /* Chain of directories to look for include files in. */ 562 struct cpp_dir 563 { 564 /* NULL-terminated singly-linked list. */ 565 struct cpp_dir *next; 566 567 /* NAME of the directory, NUL-terminated. */ 568 char *name; 569 unsigned int len; 570 571 /* One if a system header, two if a system header that has extern 572 "C" guards for C++. */ 573 unsigned char sysp; 574 575 /* Is this a user-supplied directory? */ 576 bool user_supplied_p; 577 578 /* The canonicalized NAME as determined by lrealpath. This field 579 is only used by hosts that lack reliable inode numbers. */ 580 char *canonical_name; 581 582 /* Mapping of file names for this directory for MS-DOS and related 583 platforms. A NULL-terminated array of (from, to) pairs. */ 584 const char **name_map; 585 586 /* Routine to construct pathname, given the search path name and the 587 HEADER we are trying to find, return a constructed pathname to 588 try and open. If this is NULL, the constructed pathname is as 589 constructed by append_file_to_dir. */ 590 char *(*construct) (const char *header, cpp_dir *dir); 591 592 /* The C front end uses these to recognize duplicated 593 directories in the search path. */ 594 INO_T_CPP; 595 dev_t dev; 596 }; 597 598 /* The structure of a node in the hash table. The hash table has 599 entries for all identifiers: either macros defined by #define 600 commands (type NT_MACRO), assertions created with #assert 601 (NT_ASSERTION), or neither of the above (NT_VOID). Builtin macros 602 like __LINE__ are flagged NODE_BUILTIN. Poisoned identifiers are 603 flagged NODE_POISONED. NODE_OPERATOR (C++ only) indicates an 604 identifier that behaves like an operator such as "xor". 605 NODE_DIAGNOSTIC is for speed in lex_token: it indicates a 606 diagnostic may be required for this node. Currently this only 607 applies to __VA_ARGS__, poisoned identifiers, and -Wc++-compat 608 warnings about NODE_OPERATOR. */ 609 610 /* Hash node flags. */ 611 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */ 612 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */ 613 #define NODE_BUILTIN (1 << 2) /* Builtin macro. */ 614 #define NODE_DIAGNOSTIC (1 << 3) /* Possible diagnostic when lexed. */ 615 #define NODE_WARN (1 << 4) /* Warn if redefined or undefined. */ 616 #define NODE_DISABLED (1 << 5) /* A disabled macro. */ 617 #define NODE_MACRO_ARG (1 << 6) /* Used during #define processing. */ 618 #define NODE_USED (1 << 7) /* Dumped with -dU. */ 619 #define NODE_CONDITIONAL (1 << 8) /* Conditional macro */ 620 #define NODE_WARN_OPERATOR (1 << 9) /* Warn about C++ named operator. */ 621 622 /* Different flavors of hash node. */ 623 enum node_type 624 { 625 NT_VOID = 0, /* No definition yet. */ 626 NT_MACRO, /* A macro of some form. */ 627 NT_ASSERTION /* Predicate for #assert. */ 628 }; 629 630 /* Different flavors of builtin macro. _Pragma is an operator, but we 631 handle it with the builtin code for efficiency reasons. */ 632 enum cpp_builtin_type 633 { 634 BT_SPECLINE = 0, /* `__LINE__' */ 635 BT_DATE, /* `__DATE__' */ 636 BT_FILE, /* `__FILE__' */ 637 BT_BASE_FILE, /* `__BASE_FILE__' */ 638 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */ 639 BT_TIME, /* `__TIME__' */ 640 BT_STDC, /* `__STDC__' */ 641 BT_PRAGMA, /* `_Pragma' operator */ 642 BT_TIMESTAMP, /* `__TIMESTAMP__' */ 643 BT_COUNTER, /* `__COUNTER__' */ 644 BT_FIRST_USER, /* User defined builtin macros. */ 645 BT_LAST_USER = BT_FIRST_USER + 31 646 }; 647 648 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE)) 649 #define HT_NODE(NODE) ((ht_identifier *) (NODE)) 650 #define NODE_LEN(NODE) HT_LEN (&(NODE)->ident) 651 #define NODE_NAME(NODE) HT_STR (&(NODE)->ident) 652 653 /* Specify which field, if any, of the union is used. */ 654 655 enum { 656 NTV_MACRO, 657 NTV_ANSWER, 658 NTV_BUILTIN, 659 NTV_ARGUMENT, 660 NTV_NONE 661 }; 662 663 #define CPP_HASHNODE_VALUE_IDX(HNODE) \ 664 ((HNODE.flags & NODE_MACRO_ARG) ? NTV_ARGUMENT \ 665 : HNODE.type == NT_MACRO ? ((HNODE.flags & NODE_BUILTIN) \ 666 ? NTV_BUILTIN : NTV_MACRO) \ 667 : HNODE.type == NT_ASSERTION ? NTV_ANSWER \ 668 : NTV_NONE) 669 670 /* The common part of an identifier node shared amongst all 3 C front 671 ends. Also used to store CPP identifiers, which are a superset of 672 identifiers in the grammatical sense. */ 673 674 union GTY(()) _cpp_hashnode_value { 675 /* If a macro. */ 676 cpp_macro * GTY((tag ("NTV_MACRO"))) macro; 677 /* Answers to an assertion. */ 678 struct answer * GTY ((tag ("NTV_ANSWER"))) answers; 679 /* Code for a builtin macro. */ 680 enum cpp_builtin_type GTY ((tag ("NTV_BUILTIN"))) builtin; 681 /* Macro argument index. */ 682 unsigned short GTY ((tag ("NTV_ARGUMENT"))) arg_index; 683 }; 684 685 struct GTY(()) cpp_hashnode { 686 struct ht_identifier ident; 687 unsigned int is_directive : 1; 688 unsigned int directive_index : 7; /* If is_directive, 689 then index into directive table. 690 Otherwise, a NODE_OPERATOR. */ 691 unsigned char rid_code; /* Rid code - for front ends. */ 692 ENUM_BITFIELD(node_type) type : 6; /* CPP node type. */ 693 unsigned int flags : 10; /* CPP flags. */ 694 695 union _cpp_hashnode_value GTY ((desc ("CPP_HASHNODE_VALUE_IDX (%1)"))) value; 696 }; 697 698 /* Call this first to get a handle to pass to other functions. 699 700 If you want cpplib to manage its own hashtable, pass in a NULL 701 pointer. Otherwise you should pass in an initialized hash table 702 that cpplib will share; this technique is used by the C front 703 ends. */ 704 extern cpp_reader *cpp_create_reader (enum c_lang, struct ht *, 705 struct line_maps *); 706 707 /* Reset the cpp_reader's line_map. This is only used after reading a 708 PCH file. */ 709 extern void cpp_set_line_map (cpp_reader *, struct line_maps *); 710 711 /* Call this to change the selected language standard (e.g. because of 712 command line options). */ 713 extern void cpp_set_lang (cpp_reader *, enum c_lang); 714 715 /* Set the include paths. */ 716 extern void cpp_set_include_chains (cpp_reader *, cpp_dir *, cpp_dir *, int); 717 718 /* Call these to get pointers to the options, callback, and deps 719 structures for a given reader. These pointers are good until you 720 call cpp_finish on that reader. You can either edit the callbacks 721 through the pointer returned from cpp_get_callbacks, or set them 722 with cpp_set_callbacks. */ 723 extern cpp_options *cpp_get_options (cpp_reader *); 724 extern cpp_callbacks *cpp_get_callbacks (cpp_reader *); 725 extern void cpp_set_callbacks (cpp_reader *, cpp_callbacks *); 726 extern struct deps *cpp_get_deps (cpp_reader *); 727 728 /* This function reads the file, but does not start preprocessing. It 729 returns the name of the original file; this is the same as the 730 input file, except for preprocessed input. This will generate at 731 least one file change callback, and possibly a line change callback 732 too. If there was an error opening the file, it returns NULL. */ 733 extern const char *cpp_read_main_file (cpp_reader *, const char *); 734 735 /* Set up built-ins with special behavior. Use cpp_init_builtins() 736 instead unless your know what you are doing. */ 737 extern void cpp_init_special_builtins (cpp_reader *); 738 739 /* Set up built-ins like __FILE__. */ 740 extern void cpp_init_builtins (cpp_reader *, int); 741 742 /* This is called after options have been parsed, and partially 743 processed. */ 744 extern void cpp_post_options (cpp_reader *); 745 746 /* Set up translation to the target character set. */ 747 extern void cpp_init_iconv (cpp_reader *); 748 749 /* Call this to finish preprocessing. If you requested dependency 750 generation, pass an open stream to write the information to, 751 otherwise NULL. It is your responsibility to close the stream. */ 752 extern void cpp_finish (cpp_reader *, FILE *deps_stream); 753 754 /* Call this to release the handle at the end of preprocessing. Any 755 use of the handle after this function returns is invalid. */ 756 extern void cpp_destroy (cpp_reader *); 757 758 extern unsigned int cpp_token_len (const cpp_token *); 759 extern unsigned char *cpp_token_as_text (cpp_reader *, const cpp_token *); 760 extern unsigned char *cpp_spell_token (cpp_reader *, const cpp_token *, 761 unsigned char *, bool); 762 extern void cpp_register_pragma (cpp_reader *, const char *, const char *, 763 void (*) (cpp_reader *), bool); 764 extern void cpp_register_deferred_pragma (cpp_reader *, const char *, 765 const char *, unsigned, bool, bool); 766 extern int cpp_avoid_paste (cpp_reader *, const cpp_token *, 767 const cpp_token *); 768 extern const cpp_token *cpp_get_token (cpp_reader *); 769 extern const cpp_token *cpp_get_token_with_location (cpp_reader *, 770 source_location *); 771 extern const unsigned char *cpp_macro_definition (cpp_reader *, 772 cpp_hashnode *); 773 extern void _cpp_backup_tokens (cpp_reader *, unsigned int); 774 extern const cpp_token *cpp_peek_token (cpp_reader *, int); 775 776 /* Evaluate a CPP_*CHAR* token. */ 777 extern cppchar_t cpp_interpret_charconst (cpp_reader *, const cpp_token *, 778 unsigned int *, int *); 779 /* Evaluate a vector of CPP_*STRING* tokens. */ 780 extern bool cpp_interpret_string (cpp_reader *, 781 const cpp_string *, size_t, 782 cpp_string *, enum cpp_ttype); 783 extern bool cpp_interpret_string_notranslate (cpp_reader *, 784 const cpp_string *, size_t, 785 cpp_string *, enum cpp_ttype); 786 787 /* Convert a host character constant to the execution character set. */ 788 extern cppchar_t cpp_host_to_exec_charset (cpp_reader *, cppchar_t); 789 790 /* Used to register macros and assertions, perhaps from the command line. 791 The text is the same as the command line argument. */ 792 extern void cpp_define (cpp_reader *, const char *); 793 extern void cpp_define_formatted (cpp_reader *pfile, 794 const char *fmt, ...) ATTRIBUTE_PRINTF_2; 795 extern void cpp_assert (cpp_reader *, const char *); 796 extern void cpp_undef (cpp_reader *, const char *); 797 extern void cpp_unassert (cpp_reader *, const char *); 798 799 /* Undefine all macros and assertions. */ 800 extern void cpp_undef_all (cpp_reader *); 801 802 extern cpp_buffer *cpp_push_buffer (cpp_reader *, const unsigned char *, 803 size_t, int); 804 extern int cpp_defined (cpp_reader *, const unsigned char *, int); 805 806 /* A preprocessing number. Code assumes that any unused high bits of 807 the double integer are set to zero. */ 808 typedef unsigned HOST_WIDE_INT cpp_num_part; 809 typedef struct cpp_num cpp_num; 810 struct cpp_num 811 { 812 cpp_num_part high; 813 cpp_num_part low; 814 bool unsignedp; /* True if value should be treated as unsigned. */ 815 bool overflow; /* True if the most recent calculation overflowed. */ 816 }; 817 818 /* cpplib provides two interfaces for interpretation of preprocessing 819 numbers. 820 821 cpp_classify_number categorizes numeric constants according to 822 their field (integer, floating point, or invalid), radix (decimal, 823 octal, hexadecimal), and type suffixes. */ 824 825 #define CPP_N_CATEGORY 0x000F 826 #define CPP_N_INVALID 0x0000 827 #define CPP_N_INTEGER 0x0001 828 #define CPP_N_FLOATING 0x0002 829 830 #define CPP_N_WIDTH 0x00F0 831 #define CPP_N_SMALL 0x0010 /* int, float, shrot _Fract/Accum */ 832 #define CPP_N_MEDIUM 0x0020 /* long, double, long _Fract/_Accum. */ 833 #define CPP_N_LARGE 0x0040 /* long long, long double, 834 long long _Fract/Accum. */ 835 836 #define CPP_N_WIDTH_MD 0xF0000 /* machine defined. */ 837 #define CPP_N_MD_W 0x10000 838 #define CPP_N_MD_Q 0x20000 839 840 #define CPP_N_RADIX 0x0F00 841 #define CPP_N_DECIMAL 0x0100 842 #define CPP_N_HEX 0x0200 843 #define CPP_N_OCTAL 0x0400 844 #define CPP_N_BINARY 0x0800 845 846 #define CPP_N_UNSIGNED 0x1000 /* Properties. */ 847 #define CPP_N_IMAGINARY 0x2000 848 #define CPP_N_DFLOAT 0x4000 849 #define CPP_N_DEFAULT 0x8000 850 851 #define CPP_N_FRACT 0x100000 /* Fract types. */ 852 #define CPP_N_ACCUM 0x200000 /* Accum types. */ 853 854 #define CPP_N_USERDEF 0x1000000 /* C++0x user-defined literal. */ 855 856 /* Classify a CPP_NUMBER token. The return value is a combination of 857 the flags from the above sets. */ 858 extern unsigned cpp_classify_number (cpp_reader *, const cpp_token *, 859 const char **, source_location); 860 861 /* Return the classification flags for a float suffix. */ 862 extern unsigned int cpp_interpret_float_suffix (cpp_reader *, const char *, 863 size_t); 864 865 /* Return the classification flags for an int suffix. */ 866 extern unsigned int cpp_interpret_int_suffix (cpp_reader *, const char *, 867 size_t); 868 869 /* Evaluate a token classified as category CPP_N_INTEGER. */ 870 extern cpp_num cpp_interpret_integer (cpp_reader *, const cpp_token *, 871 unsigned int); 872 873 /* Sign extend a number, with PRECISION significant bits and all 874 others assumed clear, to fill out a cpp_num structure. */ 875 cpp_num cpp_num_sign_extend (cpp_num, size_t); 876 877 /* Diagnostic levels. To get a diagnostic without associating a 878 position in the translation unit with it, use cpp_error_with_line 879 with a line number of zero. */ 880 881 enum { 882 /* Warning, an error with -Werror. */ 883 CPP_DL_WARNING = 0, 884 /* Same as CPP_DL_WARNING, except it is not suppressed in system headers. */ 885 CPP_DL_WARNING_SYSHDR, 886 /* Warning, an error with -pedantic-errors or -Werror. */ 887 CPP_DL_PEDWARN, 888 /* An error. */ 889 CPP_DL_ERROR, 890 /* An internal consistency check failed. Prints "internal error: ", 891 otherwise the same as CPP_DL_ERROR. */ 892 CPP_DL_ICE, 893 /* An informative note following a warning. */ 894 CPP_DL_NOTE, 895 /* A fatal error. */ 896 CPP_DL_FATAL 897 }; 898 899 /* Warning reason codes. Use a reason code of zero for unclassified warnings 900 and errors that are not warnings. */ 901 enum { 902 CPP_W_NONE = 0, 903 CPP_W_DEPRECATED, 904 CPP_W_COMMENTS, 905 CPP_W_MISSING_INCLUDE_DIRS, 906 CPP_W_TRIGRAPHS, 907 CPP_W_MULTICHAR, 908 CPP_W_TRADITIONAL, 909 CPP_W_LONG_LONG, 910 CPP_W_ENDIF_LABELS, 911 CPP_W_NUM_SIGN_CHANGE, 912 CPP_W_VARIADIC_MACROS, 913 CPP_W_BUILTIN_MACRO_REDEFINED, 914 CPP_W_DOLLARS, 915 CPP_W_UNDEF, 916 CPP_W_UNUSED_MACROS, 917 CPP_W_CXX_OPERATOR_NAMES, 918 CPP_W_NORMALIZE, 919 CPP_W_INVALID_PCH, 920 CPP_W_WARNING_DIRECTIVE, 921 CPP_W_LITERAL_SUFFIX 922 }; 923 924 /* Output a diagnostic of some kind. */ 925 extern bool cpp_error (cpp_reader *, int, const char *msgid, ...) 926 ATTRIBUTE_PRINTF_3; 927 extern bool cpp_warning (cpp_reader *, int, const char *msgid, ...) 928 ATTRIBUTE_PRINTF_3; 929 extern bool cpp_pedwarning (cpp_reader *, int, const char *msgid, ...) 930 ATTRIBUTE_PRINTF_3; 931 extern bool cpp_warning_syshdr (cpp_reader *, int, const char *msgid, ...) 932 ATTRIBUTE_PRINTF_3; 933 934 /* Output a diagnostic with "MSGID: " preceding the 935 error string of errno. No location is printed. */ 936 extern bool cpp_errno (cpp_reader *, int, const char *msgid); 937 938 /* Same as cpp_error, except additionally specifies a position as a 939 (translation unit) physical line and physical column. If the line is 940 zero, then no location is printed. */ 941 extern bool cpp_error_with_line (cpp_reader *, int, source_location, 942 unsigned, const char *msgid, ...) 943 ATTRIBUTE_PRINTF_5; 944 extern bool cpp_warning_with_line (cpp_reader *, int, source_location, 945 unsigned, const char *msgid, ...) 946 ATTRIBUTE_PRINTF_5; 947 extern bool cpp_pedwarning_with_line (cpp_reader *, int, source_location, 948 unsigned, const char *msgid, ...) 949 ATTRIBUTE_PRINTF_5; 950 extern bool cpp_warning_with_line_syshdr (cpp_reader *, int, source_location, 951 unsigned, const char *msgid, ...) 952 ATTRIBUTE_PRINTF_5; 953 954 /* In lex.c */ 955 extern int cpp_ideq (const cpp_token *, const char *); 956 extern void cpp_output_line (cpp_reader *, FILE *); 957 extern unsigned char *cpp_output_line_to_string (cpp_reader *, 958 const unsigned char *); 959 extern void cpp_output_token (const cpp_token *, FILE *); 960 extern const char *cpp_type2name (enum cpp_ttype, unsigned char flags); 961 /* Returns the value of an escape sequence, truncated to the correct 962 target precision. PSTR points to the input pointer, which is just 963 after the backslash. LIMIT is how much text we have. WIDE is true 964 if the escape sequence is part of a wide character constant or 965 string literal. Handles all relevant diagnostics. */ 966 extern cppchar_t cpp_parse_escape (cpp_reader *, const unsigned char ** pstr, 967 const unsigned char *limit, int wide); 968 969 /* Structure used to hold a comment block at a given location in the 970 source code. */ 971 972 typedef struct 973 { 974 /* Text of the comment including the terminators. */ 975 char *comment; 976 977 /* source location for the given comment. */ 978 source_location sloc; 979 } cpp_comment; 980 981 /* Structure holding all comments for a given cpp_reader. */ 982 983 typedef struct 984 { 985 /* table of comment entries. */ 986 cpp_comment *entries; 987 988 /* number of actual entries entered in the table. */ 989 int count; 990 991 /* number of entries allocated currently. */ 992 int allocated; 993 } cpp_comment_table; 994 995 /* Returns the table of comments encountered by the preprocessor. This 996 table is only populated when pfile->state.save_comments is true. */ 997 extern cpp_comment_table *cpp_get_comments (cpp_reader *); 998 999 /* In hash.c */ 1000 1001 /* Lookup an identifier in the hashtable. Puts the identifier in the 1002 table if it is not already there. */ 1003 extern cpp_hashnode *cpp_lookup (cpp_reader *, const unsigned char *, 1004 unsigned int); 1005 1006 typedef int (*cpp_cb) (cpp_reader *, cpp_hashnode *, void *); 1007 extern void cpp_forall_identifiers (cpp_reader *, cpp_cb, void *); 1008 1009 /* In macro.c */ 1010 extern void cpp_scan_nooutput (cpp_reader *); 1011 extern int cpp_sys_macro_p (cpp_reader *); 1012 extern unsigned char *cpp_quote_string (unsigned char *, const unsigned char *, 1013 unsigned int); 1014 1015 /* In files.c */ 1016 extern bool cpp_included (cpp_reader *, const char *); 1017 extern bool cpp_included_before (cpp_reader *, const char *, source_location); 1018 extern void cpp_make_system_header (cpp_reader *, int, int); 1019 extern bool cpp_push_include (cpp_reader *, const char *); 1020 extern bool cpp_push_default_include (cpp_reader *, const char *); 1021 extern void cpp_change_file (cpp_reader *, enum lc_reason, const char *); 1022 extern const char *cpp_get_path (struct _cpp_file *); 1023 extern cpp_dir *cpp_get_dir (struct _cpp_file *); 1024 extern cpp_buffer *cpp_get_buffer (cpp_reader *); 1025 extern struct _cpp_file *cpp_get_file (cpp_buffer *); 1026 extern cpp_buffer *cpp_get_prev (cpp_buffer *); 1027 extern void cpp_clear_file_cache (cpp_reader *); 1028 1029 /* In pch.c */ 1030 struct save_macro_data; 1031 extern int cpp_save_state (cpp_reader *, FILE *); 1032 extern int cpp_write_pch_deps (cpp_reader *, FILE *); 1033 extern int cpp_write_pch_state (cpp_reader *, FILE *); 1034 extern int cpp_valid_state (cpp_reader *, const char *, int); 1035 extern void cpp_prepare_state (cpp_reader *, struct save_macro_data **); 1036 extern int cpp_read_state (cpp_reader *, const char *, FILE *, 1037 struct save_macro_data *); 1038 1039 /* In lex.c */ 1040 extern void cpp_force_token_locations (cpp_reader *, source_location *); 1041 extern void cpp_stop_forcing_token_locations (cpp_reader *); 1042 1043 /* In expr.c */ 1044 extern enum cpp_ttype cpp_userdef_string_remove_type 1045 (enum cpp_ttype type); 1046 extern enum cpp_ttype cpp_userdef_string_add_type 1047 (enum cpp_ttype type); 1048 extern enum cpp_ttype cpp_userdef_char_remove_type 1049 (enum cpp_ttype type); 1050 extern enum cpp_ttype cpp_userdef_char_add_type 1051 (enum cpp_ttype type); 1052 extern bool cpp_userdef_string_p 1053 (enum cpp_ttype type); 1054 extern bool cpp_userdef_char_p 1055 (enum cpp_ttype type); 1056 extern const char * cpp_get_userdef_suffix 1057 (const cpp_token *); 1058 1059 #endif /* ! LIBCPP_CPPLIB_H */ 1060