1 /* Definitions for CPP library.
2 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002
3 Free Software Foundation, Inc.
4 Written by Per Bothner, 1994-95.
6 This program is free software; you can redistribute it and/or modify it
7 under the terms of the GNU General Public License as published by the
8 Free Software Foundation; either version 2, or (at your option) any
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
20 In other words, you are welcome to use, share and improve this program.
21 You are forbidden to forbid anyone else to use, share and improve
22 what you give them. Help stamp out software-hoarding! */
26 #include <sys/types.h>
27 #include "hashtable.h"
34 /* For complex reasons, cpp_reader is also typedefed in c-pragma.h. */
35 #ifndef GCC_C_PRAGMA_H
36 typedef struct cpp_reader cpp_reader;
38 typedef struct cpp_buffer cpp_buffer;
39 typedef struct cpp_options cpp_options;
40 typedef struct cpp_token cpp_token;
41 typedef struct cpp_string cpp_string;
42 typedef struct cpp_hashnode cpp_hashnode;
43 typedef struct cpp_macro cpp_macro;
44 typedef struct cpp_callbacks cpp_callbacks;
47 struct file_name_map_list;
49 /* The first two groups, apart from '=', can appear in preprocessor
50 expressions. This allows a lookup table to be implemented in
53 The first group, to CPP_LAST_EQ, can be immediately followed by an
54 '='. The lexer needs operators ending in '=', like ">>=", to be in
55 the same order as their counterparts without the '=', like ">>". */
57 /* Positions in the table. */
58 #define CPP_LAST_EQ CPP_MAX
59 #define CPP_FIRST_DIGRAPH CPP_HASH
60 #define CPP_LAST_PUNCTUATOR CPP_DOT_STAR
65 OP(CPP_GREATER, ">") /* compare */ \
67 OP(CPP_PLUS, "+") /* math */ \
72 OP(CPP_AND, "&") /* bit ops */ \
75 OP(CPP_RSHIFT, ">>") \
76 OP(CPP_LSHIFT, "<<") \
77 OP(CPP_MIN, "<?") /* extension */ \
81 OP(CPP_AND_AND, "&&") /* logical */ \
85 OP(CPP_COMMA, ",") /* grouping */ \
86 OP(CPP_OPEN_PAREN, "(") \
87 OP(CPP_CLOSE_PAREN, ")") \
88 OP(CPP_EQ_EQ, "==") /* compare */ \
89 OP(CPP_NOT_EQ, "!=") \
90 OP(CPP_GREATER_EQ, ">=") \
91 OP(CPP_LESS_EQ, "<=") \
93 OP(CPP_PLUS_EQ, "+=") /* math */ \
94 OP(CPP_MINUS_EQ, "-=") \
95 OP(CPP_MULT_EQ, "*=") \
96 OP(CPP_DIV_EQ, "/=") \
97 OP(CPP_MOD_EQ, "%=") \
98 OP(CPP_AND_EQ, "&=") /* bit ops */ \
100 OP(CPP_XOR_EQ, "^=") \
101 OP(CPP_RSHIFT_EQ, ">>=") \
102 OP(CPP_LSHIFT_EQ, "<<=") \
103 OP(CPP_MIN_EQ, "<?=") /* extension */ \
104 OP(CPP_MAX_EQ, ">?=") \
105 /* Digraphs together, beginning with CPP_FIRST_DIGRAPH. */ \
106 OP(CPP_HASH, "#") /* digraphs */ \
107 OP(CPP_PASTE, "##") \
108 OP(CPP_OPEN_SQUARE, "[") \
109 OP(CPP_CLOSE_SQUARE, "]") \
110 OP(CPP_OPEN_BRACE, "{") \
111 OP(CPP_CLOSE_BRACE, "}") \
112 /* The remainder of the punctuation. Order is not significant. */ \
113 OP(CPP_SEMICOLON, ";") /* structure */ \
114 OP(CPP_ELLIPSIS, "...") \
115 OP(CPP_PLUS_PLUS, "++") /* increment */ \
116 OP(CPP_MINUS_MINUS, "--") \
117 OP(CPP_DEREF, "->") /* accessors */ \
119 OP(CPP_SCOPE, "::") \
120 OP(CPP_DEREF_STAR, "->*") \
121 OP(CPP_DOT_STAR, ".*") \
122 OP(CPP_ATSIGN, "@") /* used in Objective C */ \
124 TK(CPP_NAME, SPELL_IDENT) /* word */ \
125 TK(CPP_NUMBER, SPELL_NUMBER) /* 34_be+ta */ \
127 TK(CPP_CHAR, SPELL_STRING) /* 'char' */ \
128 TK(CPP_WCHAR, SPELL_STRING) /* L'char' */ \
129 TK(CPP_OTHER, SPELL_CHAR) /* stray punctuation */ \
131 TK(CPP_STRING, SPELL_STRING) /* "string" */ \
132 TK(CPP_WSTRING, SPELL_STRING) /* L"string" */ \
133 TK(CPP_HEADER_NAME, SPELL_STRING) /* <stdio.h> in #include */ \
135 TK(CPP_COMMENT, SPELL_NUMBER) /* Only if output comments. */ \
136 /* SPELL_NUMBER happens to DTRT. */ \
137 TK(CPP_MACRO_ARG, SPELL_NONE) /* Macro argument. */ \
138 TK(CPP_PADDING, SPELL_NONE) /* Whitespace for cpp0. */ \
139 TK(CPP_EOF, SPELL_NONE) /* End of line or file. */ \
141 /* SDCC _asm specific */ \
142 TK(CPP_ASM, SPELL_STRING) /* _asm ... _endasm ; */
154 /* C language kind, used when calling cpp_reader_init. */
155 enum c_lang {CLK_GNUC89 = 0, CLK_GNUC99, CLK_STDC89, CLK_STDC94, CLK_STDC99,
156 CLK_GNUCXX, CLK_CXX98, CLK_OBJC, CLK_OBJCXX, CLK_ASM};
158 /* Payload of a NUMBER, STRING, CHAR or COMMENT token. */
162 const unsigned char *text;
165 /* Flags for the cpp_token structure. */
166 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */
167 #define DIGRAPH (1 << 1) /* If it was a digraph. */
168 #define STRINGIFY_ARG (1 << 2) /* If macro argument to be stringified. */
169 #define PASTE_LEFT (1 << 3) /* If on LHS of a ## operator. */
170 #define NAMED_OP (1 << 4) /* C++ named operators. */
171 #define NO_EXPAND (1 << 5) /* Do not macro-expand this token. */
172 #define BOL (1 << 6) /* Token at beginning of line. */
174 /* A preprocessing token. This has been carefully packed and should
175 occupy 16 bytes on 32-bit hosts and 24 bytes on 64-bit hosts. */
178 unsigned int line; /* Logical line of first char of token. */
179 unsigned short col; /* Column of first char of token. */
180 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
181 unsigned char flags; /* flags - see above */
185 cpp_hashnode *node; /* An identifier. */
186 const cpp_token *source; /* Inherit padding from this token. */
187 struct cpp_string str; /* A string, or number. */
188 unsigned int arg_no; /* Argument no. for a CPP_MACRO_ARG. */
189 unsigned char c; /* Character represented by CPP_OTHER. */
193 /* A standalone character. We may want to make it unsigned for the
194 same reason we use unsigned char - to avoid signedness issues. */
195 typedef int cppchar_t;
197 /* Values for opts.dump_macros.
198 dump_only means inhibit output of the preprocessed text
199 and instead output the definitions of all user-defined
200 macros in a form suitable for use as input to cpp.
201 dump_names means pass #define and the macro name through to output.
202 dump_definitions means pass the whole definition (plus #define) through
204 enum { dump_none = 0, dump_only, dump_names, dump_definitions };
206 /* This structure is nested inside struct cpp_reader, and
207 carries all the options visible to the command line. */
210 /* Name of input and output files. */
211 const char *in_fname;
212 const char *out_fname;
214 /* Characters between tab stops. */
215 unsigned int tabstop;
217 /* Pending options - -D, -U, -A, -I, -ixxx. */
218 struct cpp_pending *pending;
220 /* File name which deps are being written to. This is 0 if deps are
221 being written to stdout. */
222 const char *deps_file;
224 /* Search paths for include files. */
225 struct search_path *quote_include; /* "" */
226 struct search_path *bracket_include; /* <> */
228 /* Map between header names and file names, used only on DOS where
229 file names are limited in length. */
230 struct file_name_map_list *map_list;
232 /* Directory prefix that should replace `/usr/lib/gcc-lib/TARGET/VERSION'
233 in the standard include file directories. */
234 const char *include_prefix;
235 unsigned int include_prefix_len;
237 /* -fleading_underscore sets this to "_". */
238 const char *user_label_prefix;
240 /* The language we're preprocessing. */
243 /* Non-0 means -v, so print the full set of include dirs. */
244 unsigned char verbose;
246 /* Nonzero means chars are signed. */
247 unsigned char signed_char;
249 /* Nonzero means use extra default include directories for C++. */
250 unsigned char cplusplus;
252 /* Nonzero means handle cplusplus style comments */
253 unsigned char cplusplus_comments;
255 /* Nonzero means handle #import, for objective C. */
258 /* Nonzero means don't copy comments into the output file. */
259 unsigned char discard_comments;
261 /* Nonzero means process the ISO trigraph sequences. */
262 unsigned char trigraphs;
264 /* Nonzero means process the ISO digraph sequences. */
265 unsigned char digraphs;
267 /* Nonzero means to allow hexadecimal floats and LL suffixes. */
268 unsigned char extended_numbers;
270 /* Nonzero means print the names of included files rather than the
271 preprocessed output. 1 means just the #include "...", 2 means
272 #include <...> as well. */
273 unsigned char print_deps;
275 /* Nonzero if phony targets are created for each header. */
276 unsigned char deps_phony_targets;
278 /* Nonzero if missing .h files in -M output are assumed to be
279 generated files and not errors. */
280 unsigned char print_deps_missing_files;
282 /* If true, fopen (deps_file, "a") else fopen (deps_file, "w"). */
283 unsigned char print_deps_append;
285 /* If true, no dependency is generated on the main file. */
286 unsigned char deps_ignore_main_file;
288 /* Nonzero means print names of header files (-H). */
289 unsigned char print_include_names;
291 /* Nonzero means cpp_pedwarn causes a hard error. */
292 unsigned char pedantic_errors;
294 /* Nonzero means don't print warning messages. */
295 unsigned char inhibit_warnings;
297 /* Nonzero means don't suppress warnings from system headers. */
298 unsigned char warn_system_headers;
300 /* Nonzero means don't print error messages. Has no option to
301 select it, but can be set by a user of cpplib (e.g. fix-header). */
302 unsigned char inhibit_errors;
304 /* Nonzero means warn if slash-star appears in a comment. */
305 unsigned char warn_comments;
307 /* Nonzero means warn if there are any trigraphs. */
308 unsigned char warn_trigraphs;
310 /* Nonzero means warn if #import is used. */
311 unsigned char warn_import;
313 /* Nonzero means warn about various incompatibilities with
315 unsigned char warn_traditional;
317 /* Nonzero means turn warnings into errors. */
318 unsigned char warnings_are_errors;
320 /* Nonzero causes output not to be done, but directives such as
321 #define that have side effects are still obeyed. */
322 unsigned char no_output;
324 /* Nonzero means we should look for header.gcc files that remap file
328 /* Nonzero means don't output line number information. */
329 unsigned char no_line_commands;
331 /* Nonzero means -I- has been seen, so don't look for #include "foo"
332 the source-file directory. */
333 unsigned char ignore_srcdir;
335 /* Zero means dollar signs are punctuation. */
336 unsigned char dollars_in_ident;
338 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
339 unsigned char warn_undef;
341 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
344 /* Nonzero if conforming to some particular standard. */
347 /* Nonzero means give all the error messages the ANSI standard requires. */
348 unsigned char pedantic;
350 /* Nonzero means we're looking at already preprocessed code, so don't
351 bother trying to do macro expansion and whatnot. */
352 unsigned char preprocessed;
354 /* Nonzero disables all the standard directories for headers. */
355 unsigned char no_standard_includes;
357 /* Nonzero disables the C++-specific standard directories for headers. */
358 unsigned char no_standard_cplusplus_includes;
360 /* Nonzero means dump macros in some fashion - see above. */
361 unsigned char dump_macros;
363 /* Nonzero means pass #include lines through to the output. */
364 unsigned char dump_includes;
366 /* Print column number in error messages. */
367 unsigned char show_column;
369 /* Nonzero means handle C++ alternate operator names. */
370 unsigned char operator_names;
372 /* True if --help, --version or --target-help appeared in the
373 options. Stand-alone CPP should then bail out after option
374 parsing; drivers might want to continue printing help. */
375 unsigned char help_only;
377 /* SDCC abuse by Kevin: allow naked '#' characters in expanded macros
378 * (see _cpp_create_definition in cppmacro.c)
380 unsigned char allow_naked_hash;
382 /* SDCC _asm specific
383 switch _asm block preprocessing on / off */
384 unsigned char preproc_asm;
387 object file exetnsion */
394 /* Called when a new line of preprocessed output is started. */
395 void (*line_change) PARAMS ((cpp_reader *, const cpp_token *, int));
396 void (*file_change) PARAMS ((cpp_reader *, const struct line_map *));
397 void (*include) PARAMS ((cpp_reader *, unsigned int,
398 const unsigned char *, const cpp_token *));
399 void (*define) PARAMS ((cpp_reader *, unsigned int, cpp_hashnode *));
400 void (*undef) PARAMS ((cpp_reader *, unsigned int, cpp_hashnode *));
401 void (*ident) PARAMS ((cpp_reader *, unsigned int, const cpp_string *));
402 void (*def_pragma) PARAMS ((cpp_reader *, unsigned int));
405 #define CPP_FATAL_LIMIT 1000
406 /* True if we have seen a "fatal" error. */
407 #define CPP_FATAL_ERRORS(PFILE) (cpp_errors (PFILE) >= CPP_FATAL_LIMIT)
409 /* Name under which this program was invoked. */
410 extern const char *progname;
412 /* The structure of a node in the hash table. The hash table has
413 entries for all identifiers: either macros defined by #define
414 commands (type NT_MACRO), assertions created with #assert
415 (NT_ASSERTION), or neither of the above (NT_VOID). Builtin macros
416 like __LINE__ are flagged NODE_BUILTIN. Poisioned identifiers are
417 flagged NODE_POISONED. NODE_OPERATOR (C++ only) indicates an
418 identifier that behaves like an operator such as "xor".
419 NODE_DIAGNOSTIC is for speed in lex_token: it indicates a
420 diagnostic may be required for this node. Currently this only
421 applies to __VA_ARGS__ and poisoned identifiers. */
423 /* Hash node flags. */
424 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */
425 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */
426 #define NODE_BUILTIN (1 << 2) /* Builtin macro. */
427 #define NODE_DIAGNOSTIC (1 << 3) /* Possible diagnostic when lexed. */
428 #define NODE_WARN (1 << 4) /* Warn if redefined or undefined. */
429 #define NODE_DISABLED (1 << 5) /* A disabled macro. */
431 /* Different flavors of hash node. */
434 NT_VOID = 0, /* No definition yet. */
435 NT_MACRO, /* A macro of some form. */
436 NT_ASSERTION /* Predicate for #assert. */
439 /* Different flavors of builtin macro. _Pragma is an operator, but we
440 handle it with the builtin code for efficiency reasons. */
443 BT_SPECLINE = 0, /* `__LINE__' */
444 BT_DATE, /* `__DATE__' */
445 BT_FILE, /* `__FILE__' */
446 BT_BASE_FILE, /* `__BASE_FILE__' */
447 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
448 BT_TIME, /* `__TIME__' */
449 BT_STDC, /* `__STDC__' */
450 BT_PRAGMA /* `_Pragma' operator */
453 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE))
454 #define HT_NODE(NODE) ((ht_identifier *) (NODE))
455 #define NODE_LEN(NODE) HT_LEN (&(NODE)->ident)
456 #define NODE_NAME(NODE) HT_STR (&(NODE)->ident)
458 /* The common part of an identifier node shared amongst all 3 C front
459 ends. Also used to store CPP identifiers, which are a superset of
460 identifiers in the grammatical sense. */
463 struct ht_identifier ident;
464 unsigned short arg_index; /* Macro argument index. */
465 unsigned char directive_index; /* Index into directive table. */
466 unsigned char rid_code; /* Rid code - for front ends. */
467 ENUM_BITFIELD(node_type) type : 8; /* CPP node type. */
468 unsigned char flags; /* CPP flags. */
472 cpp_macro *macro; /* If a macro. */
473 struct answer *answers; /* Answers to an assertion. */
474 enum cpp_ttype operator; /* Code for a named operator. */
475 enum builtin_type builtin; /* Code for a builtin macro. */
479 /* Call this first to get a handle to pass to other functions. */
480 extern cpp_reader *cpp_create_reader PARAMS ((enum c_lang));
482 /* Call these to get pointers to the options and callback structures
483 for a given reader. These pointers are good until you call
484 cpp_finish on that reader. You can either edit the callbacks
485 through the pointer returned from cpp_get_callbacks, or set them
486 with cpp_set_callbacks. */
487 extern cpp_options *cpp_get_options PARAMS ((cpp_reader *));
488 extern const struct line_maps *cpp_get_line_maps PARAMS ((cpp_reader *));
489 extern cpp_callbacks *cpp_get_callbacks PARAMS ((cpp_reader *));
490 extern void cpp_set_callbacks PARAMS ((cpp_reader *, cpp_callbacks *));
492 /* Now call cpp_handle_option[s] to handle 1[or more] switches. The
493 return value is the number of arguments used. If
494 cpp_handle_options returns without using all arguments, it couldn't
495 understand the next switch. When there are no switches left, you
496 must call cpp_post_options before calling cpp_read_main_file. Only
497 after cpp_post_options are the contents of the cpp_options
498 structure reliable. Options processing is not completed until you
499 call cpp_finish_options. */
500 extern int cpp_handle_options PARAMS ((cpp_reader *, int, char **));
501 extern int cpp_handle_option PARAMS ((cpp_reader *, int, char **, int));
502 extern void cpp_post_options PARAMS ((cpp_reader *));
504 /* This function reads the file, but does not start preprocessing. It
505 returns the name of the original file; this is the same as the
506 input file, except for preprocessed input. This will generate at
507 least one file change callback, and possibly a line change callback
508 too. If there was an error opening the file, it returns NULL.
510 If you want cpplib to manage its own hashtable, pass in a NULL
511 pointer. Otherise you should pass in an initialised hash table
512 that cpplib will share; this technique is used by the C front
514 extern const char *cpp_read_main_file PARAMS ((cpp_reader *, const char *,
517 /* Deferred handling of command line options that can generate debug
518 callbacks, such as -D and -imacros. Call this after
519 cpp_read_main_file. The front ends need this separation so they
520 can initialize debug output with the original file name, returned
521 from cpp_read_main_file, before they get debug callbacks. */
522 extern void cpp_finish_options PARAMS ((cpp_reader *));
524 /* Call this to release the handle at the end of preprocessing. Any
525 use of the handle after this function returns is invalid. Returns
526 cpp_errors (pfile). */
527 extern int cpp_destroy PARAMS ((cpp_reader *));
530 extern unsigned int cpp_errors PARAMS ((cpp_reader *));
532 extern unsigned int cpp_token_len PARAMS ((const cpp_token *));
533 extern unsigned char *cpp_token_as_text PARAMS ((cpp_reader *,
535 extern unsigned char *cpp_spell_token PARAMS ((cpp_reader *, const cpp_token *,
537 extern void cpp_register_pragma PARAMS ((cpp_reader *,
538 const char *, const char *,
539 void (*) PARAMS ((cpp_reader *))));
541 extern void cpp_finish PARAMS ((cpp_reader *));
542 extern int cpp_avoid_paste PARAMS ((cpp_reader *, const cpp_token *,
544 extern const cpp_token *cpp_get_token PARAMS ((cpp_reader *));
545 extern const unsigned char *cpp_macro_definition PARAMS ((cpp_reader *,
546 const cpp_hashnode *));
547 extern void _cpp_backup_tokens PARAMS ((cpp_reader *, unsigned int));
549 /* Evaluate a CPP_CHAR or CPP_WCHAR token. */
551 cpp_interpret_charconst PARAMS ((cpp_reader *, const cpp_token *,
552 int, int, unsigned int *));
554 extern void cpp_define PARAMS ((cpp_reader *, const char *));
555 extern void cpp_assert PARAMS ((cpp_reader *, const char *));
556 extern void cpp_undef PARAMS ((cpp_reader *, const char *));
557 extern void cpp_unassert PARAMS ((cpp_reader *, const char *));
559 extern cpp_buffer *cpp_push_buffer PARAMS ((cpp_reader *,
560 const unsigned char *, size_t,
562 extern int cpp_defined PARAMS ((cpp_reader *, const unsigned char *, int));
564 /* N.B. The error-message-printer prototypes have not been nicely
565 formatted because exgettext needs to see 'msgid' on the same line
566 as the name of the function in order to work properly. Only the
567 string argument gets a name in an effort to keep the lines from
568 getting ridiculously oversized. */
570 extern void cpp_ice PARAMS ((cpp_reader *, const char *msgid, ...))
572 extern void cpp_fatal PARAMS ((cpp_reader *, const char *msgid, ...))
574 extern void cpp_error PARAMS ((cpp_reader *, const char *msgid, ...))
576 extern void cpp_warning PARAMS ((cpp_reader *, const char *msgid, ...))
578 extern void cpp_pedwarn PARAMS ((cpp_reader *, const char *msgid, ...))
580 extern void cpp_notice PARAMS ((cpp_reader *, const char *msgid, ...))
582 extern void cpp_error_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
584 extern void cpp_warning_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
586 extern void cpp_pedwarn_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
588 extern void cpp_error_from_errno PARAMS ((cpp_reader *, const char *));
589 extern void cpp_notice_from_errno PARAMS ((cpp_reader *, const char *));
592 extern int cpp_ideq PARAMS ((const cpp_token *,
594 extern void cpp_output_line PARAMS ((cpp_reader *, FILE *));
595 extern void cpp_output_token PARAMS ((const cpp_token *, FILE *));
596 extern const char *cpp_type2name PARAMS ((enum cpp_ttype));
597 extern unsigned int cpp_parse_escape PARAMS ((cpp_reader *,
598 const unsigned char **,
599 const unsigned char *,
600 unsigned HOST_WIDE_INT, int));
604 /* Lookup an identifier in the hashtable. Puts the identifier in the
605 table if it is not already there. */
606 extern cpp_hashnode *cpp_lookup PARAMS ((cpp_reader *,
607 const unsigned char *,
610 typedef int (*cpp_cb) PARAMS ((cpp_reader *, cpp_hashnode *, void *));
611 extern void cpp_forall_identifiers PARAMS ((cpp_reader *,
615 extern void cpp_scan_nooutput PARAMS ((cpp_reader *));
616 extern int cpp_sys_macro_p PARAMS ((cpp_reader *));
617 extern unsigned char *cpp_quote_string PARAMS ((unsigned char *,
618 const unsigned char *,
622 extern int cpp_included PARAMS ((cpp_reader *, const char *));
623 extern void cpp_make_system_header PARAMS ((cpp_reader *, int, int));
629 #endif /* ! GCC_CPPLIB_H */