1 /* Definitions for CPP library.
2 Copyright (C) 1995, 96-99, 2000 Free Software Foundation, Inc.
3 Written by Per Bothner, 1994-95.
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 2, or (at your option) any
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.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software
17 Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
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 __GCC_CPPLIB__
23 #define __GCC_CPPLIB__
25 #include <sys/types.h>
31 typedef struct cpp_reader cpp_reader;
32 typedef struct cpp_buffer cpp_buffer;
33 typedef struct cpp_options cpp_options;
34 typedef struct cpp_printer cpp_printer;
35 typedef struct cpp_token cpp_token;
36 typedef struct cpp_toklist cpp_toklist;
37 typedef struct cpp_name cpp_name;
38 typedef struct cpp_hashnode cpp_hashnode;
40 /* The first two groups, apart from '=', can appear in preprocessor
41 expressions. This allows a lookup table to be implemented in
44 The first group, to CPP_LAST_EQ, can be immediately followed by an
45 '='. The lexer needs operators ending in '=', like ">>=", to be in
46 the same order as their counterparts without the '=', like ">>". */
48 /* Positions in the table. */
49 #define CPP_LAST_EQ CPP_LSHIFT
50 #define CPP_FIRST_DIGRAPH CPP_HASH
55 T(CPP_GREATER, ">") /* compare */ \
57 T(CPP_PLUS, "+") /* math */ \
62 T(CPP_AND, "&") /* bit ops */ \
69 T(CPP_AND_AND, "&&") /* logical */ \
73 T(CPP_COMMA, ",") /* grouping */ \
74 T(CPP_OPEN_PAREN, "(") \
75 T(CPP_CLOSE_PAREN, ")") \
76 T(CPP_EQ_EQ, "==") /* compare */ \
78 T(CPP_GREATER_EQ, ">=") \
79 T(CPP_LESS_EQ, "<=") \
81 T(CPP_PLUS_EQ, "+=") /* math */ \
82 T(CPP_MINUS_EQ, "-=") \
83 T(CPP_MULT_EQ, "*=") \
86 T(CPP_AND_EQ, "&=") /* bit ops */ \
89 T(CPP_RSHIFT_EQ, ">>=") \
90 T(CPP_LSHIFT_EQ, "<<=") \
91 /* Digraphs together, beginning with CPP_FIRST_DIGRAPH. */ \
92 T(CPP_HASH, "#") /* digraphs */ \
94 T(CPP_OPEN_SQUARE, "[") \
95 T(CPP_CLOSE_SQUARE, "]") \
96 T(CPP_OPEN_BRACE, "{") \
97 T(CPP_CLOSE_BRACE, "}") \
98 /* The remainder of the punctuation. Order is not significant. */ \
99 T(CPP_SEMICOLON, ";") /* structure */ \
100 T(CPP_ELLIPSIS, "...") \
101 T(CPP_BACKSLASH, "\\") \
102 T(CPP_PLUS_PLUS, "++") /* increment */ \
103 T(CPP_MINUS_MINUS, "--") \
104 T(CPP_DEREF, "->") /* accessors */ \
107 T(CPP_DEREF_STAR, "->*") \
108 T(CPP_DOT_STAR, ".*") \
109 T(CPP_MIN, "<?") /* extension */ \
111 T(CPP_PLACEMARKER, "") /* Placemarker token. */ \
112 C(CPP_OTHER, 0) /* stray punctuation */ \
114 I(CPP_NAME, 0) /* word */ \
115 I(CPP_INT, 0) /* 23 */ \
116 I(CPP_FLOAT, 0) /* 3.14159 */ \
117 I(CPP_NUMBER, 0) /* 34_be+ta */ \
118 S(CPP_CHAR, 0) /* 'char' */ \
119 S(CPP_WCHAR, 0) /* L'char' */ \
120 S(CPP_STRING, 0) /* "string" */ \
121 S(CPP_WSTRING, 0) /* L"string" */ \
123 I(CPP_COMMENT, 0) /* Only if output comments. */ \
124 N(CPP_MACRO_ARG, 0) /* Macro argument. */ \
125 N(CPP_EOF, 0) /* End of file. */ \
126 I(CPP_HEADER_NAME, 0) /* <stdio.h> in #include */ \
128 /* Obsolete - will be removed when no code uses them still. */ \
129 T(CPP_VSPACE, "\n") /* End of line. */
147 /* Payload of a NAME, NUMBER, FLOAT, STRING, or COMMENT token. */
151 const unsigned char *text;
154 /* Flags for the cpp_token structure. */
155 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */
156 #define BOL (1 << 1) /* Beginning of logical line. */
157 #define DIGRAPH (1 << 2) /* If it was a digraph. */
158 #define STRINGIFY_ARG (1 << 3) /* If macro argument to be stringified. */
159 #define PASTE_LEFT (1 << 4) /* If on LHS of a ## operator. */
160 #define PASTED (1 << 5) /* The result of a ## operator. */
161 #define GNU_VARARGS (1 << 6) /* GNU ## kludge. */
163 /* A preprocessing token. This has been carefully packed and should
164 occupy 16 bytes on 32-bit hosts and 24 bytes on 64-bit hosts. */
167 unsigned int line; /* starting line number of this token */
168 unsigned short col; /* starting column of this token */
169 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
170 unsigned char flags; /* flags - see above */
174 HOST_WIDEST_INT integer; /* an integer */
175 struct cpp_name name; /* a string */
176 unsigned int aux; /* argument no. for a CPP_MACRO_ARG, or
177 character represented by CPP_OTHER. */
181 /* cpp_toklist flags. */
182 #define LIST_OFFSET (1 << 0)
183 #define VAR_ARGS (1 << 1)
184 #define BEG_OF_FILE (1 << 2)
186 struct directive; /* These are deliberately incomplete. */
193 cpp_token *tokens; /* actual tokens as an array */
194 unsigned int tokens_used; /* tokens used */
195 unsigned int tokens_cap; /* tokens allocated */
197 unsigned char *namebuf; /* names buffer */
198 unsigned int name_used; /* _bytes_ used */
199 unsigned int name_cap; /* _bytes_ allocated */
201 /* If the list represents a directive, this points to it. */
202 const struct directive *directive;
204 const char *file; /* in file name */
205 unsigned int line; /* starting line number */
207 unsigned short params_len; /* length of macro parameter names. */
209 short int paramc; /* no. of macro params (-1 = obj-like). */
211 /* Per-list flags, see above */
212 unsigned short flags;
217 const unsigned char *cur; /* current position */
218 const unsigned char *rlimit; /* end of valid data */
219 const unsigned char *buf; /* entire buffer */
220 const unsigned char *line_base; /* start of current line */
222 struct cpp_buffer *prev;
224 /* Filename specified with #line command. */
225 const char *nominal_fname;
227 /* Actual directory of this file, used only for "" includes */
228 struct file_name_list *actual_dir;
230 /* Pointer into the include table. Used for include_next and
231 to record control macros. */
232 struct include_file *inc;
234 /* Value of if_stack at start of this file.
235 Used to prohibit unmatched #endif (etc) in an include file. */
236 struct if_stack *if_stack;
238 /* Line number at line_base (above). */
241 /* True if we have already warned about C++ comments in this file.
242 The warning happens only for C89 extended mode with -pedantic on,
243 or for -Wtraditional, and only once per file (otherwise it would
244 be far too noisy). */
245 char warned_cplusplus_comments;
247 /* True if this buffer's data is mmapped. */
251 struct file_name_map_list;
254 /* Maximum nesting of cpp_buffers. We use a static limit, partly for
255 efficiency, and partly to limit runaway recursion. */
256 #define CPP_STACK_MAX 200
258 /* Values for opts.dump_macros.
259 dump_only means inhibit output of the preprocessed text
260 and instead output the definitions of all user-defined
261 macros in a form suitable for use as input to cpp.
262 dump_names means pass #define and the macro name through to output.
263 dump_definitions means pass the whole definition (plus #define) through
265 enum { dump_none = 0, dump_only, dump_names, dump_definitions };
267 /* This structure is nested inside struct cpp_reader, and
268 carries all the options visible to the command line. */
271 /* Name of input and output files. */
272 const char *in_fname;
273 const char *out_fname;
275 /* Characters between tab stops. */
276 unsigned int tabstop;
278 /* Pending options - -D, -U, -A, -I, -ixxx. */
279 struct cpp_pending *pending;
281 /* File name which deps are being written to. This is 0 if deps are
282 being written to stdout. */
283 const char *deps_file;
285 /* Target-name to write with the dependency information. */
288 /* Search paths for include files. */
289 struct file_name_list *quote_include; /* First dir to search for "file" */
290 struct file_name_list *bracket_include;/* First dir to search for <file> */
292 /* Map between header names and file names, used only on DOS where
293 file names are limited in length. */
294 struct file_name_map_list *map_list;
296 /* Directory prefix that should replace `/usr/lib/gcc-lib/TARGET/VERSION'
297 in the standard include file directories. */
298 const char *include_prefix;
299 unsigned int include_prefix_len;
301 /* Non-0 means -v, so print the full set of include dirs. */
302 unsigned char verbose;
304 /* Nonzero means use extra default include directories for C++. */
305 unsigned char cplusplus;
307 /* Nonzero means handle cplusplus style comments */
308 unsigned char cplusplus_comments;
310 /* Nonzero means handle #import, for objective C. */
313 /* Nonzero means this is an assembly file, so ignore unrecognized
314 directives and the "# 33" form of #line, both of which are
315 probably comments. Also, permit unbalanced ' strings (again,
316 likely to be in comments). */
317 unsigned char lang_asm;
319 /* Nonzero means this is Fortran, and we don't know where the
320 comments are, so permit unbalanced ' strings. Unlike lang_asm,
321 this does not ignore unrecognized directives. */
322 unsigned char lang_fortran;
324 /* Nonzero means handle CHILL comment syntax and output CHILL string
325 delimiters for __DATE__ etc. */
328 /* Nonzero means don't copy comments into the output file. */
329 unsigned char discard_comments;
331 /* Nonzero means process the ISO trigraph sequences. */
332 unsigned char trigraphs;
334 /* Nonzero means print the names of included files rather than the
335 preprocessed output. 1 means just the #include "...", 2 means
336 #include <...> as well. */
337 unsigned char print_deps;
339 /* Nonzero if missing .h files in -M output are assumed to be
340 generated files and not errors. */
341 unsigned char print_deps_missing_files;
343 /* If true, fopen (deps_file, "a") else fopen (deps_file, "w"). */
344 unsigned char print_deps_append;
346 /* Nonzero means print names of header files (-H). */
347 unsigned char print_include_names;
349 /* Nonzero means cpp_pedwarn causes a hard error. */
350 unsigned char pedantic_errors;
352 /* Nonzero means don't print warning messages. */
353 unsigned char inhibit_warnings;
355 /* Nonzero means don't print error messages. Has no option to
356 select it, but can be set by a user of cpplib (e.g. fix-header). */
357 unsigned char inhibit_errors;
359 /* Nonzero means warn if slash-star appears in a comment. */
360 unsigned char warn_comments;
362 /* Nonzero means warn if there are any trigraphs. */
363 unsigned char warn_trigraphs;
365 /* Nonzero means warn if #import is used. */
366 unsigned char warn_import;
368 /* Nonzero means warn if a macro argument is (or would be)
369 stringified with -traditional, and warn about directives
370 with the # indented from the beginning of the line. */
371 unsigned char warn_traditional;
373 /* Nonzero means warn if ## is applied to two tokens that cannot be
375 unsigned char warn_paste;
377 /* Nonzero means turn warnings into errors. */
378 unsigned char warnings_are_errors;
380 /* Nonzero causes output not to be done, but directives such as
381 #define that have side effects are still obeyed. */
382 unsigned char no_output;
384 /* Nonzero means we should look for header.gcc files that remap file
388 /* Nonzero means don't output line number information. */
389 unsigned char no_line_commands;
391 /* Nonzero means -I- has been seen, so don't look for #include "foo"
392 the source-file directory. */
393 unsigned char ignore_srcdir;
395 /* Zero means dollar signs are punctuation. */
396 unsigned char dollars_in_ident;
398 /* Nonzero means try to imitate old fashioned non-ISO preprocessor. */
399 unsigned char traditional;
401 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
402 unsigned char warn_undef;
404 /* Nonzero for the 1989 C Standard, including corrigenda and amendments. */
407 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
410 /* Nonzero means give all the error messages the ANSI standard requires. */
411 unsigned char pedantic;
413 /* Nonzero means we're looking at already preprocessed code, so don't
414 bother trying to do macro expansion and whatnot. */
415 unsigned char preprocessed;
417 /* Nonzero disables all the standard directories for headers. */
418 unsigned char no_standard_includes;
420 /* Nonzero disables the C++-specific standard directories for headers. */
421 unsigned char no_standard_cplusplus_includes;
423 /* Nonzero means dump macros in some fashion - see above. */
424 unsigned char dump_macros;
426 /* Nonzero means pass all #define and #undef directives which we
427 actually process through to the output stream. This feature is
428 used primarily to allow cc1 to record the #defines and #undefs
429 for the sake of debuggers which understand about preprocessor
430 macros, but it may also be useful with -E to figure out how
431 symbols are defined, and where they are defined. */
432 unsigned char debug_output;
434 /* Nonzero means pass #include lines through to the output. */
435 unsigned char dump_includes;
437 /* Print column number in error messages. */
438 unsigned char show_column;
441 /* A cpp_reader encapsulates the "state" of a pre-processor run.
442 Applying cpp_get_token repeatedly yields a stream of pre-processor
443 tokens. Usually, there is only one cpp_reader object active. */
447 /* HACK FIXME. Maybe make into cpp_printer printer later. */
448 cpp_printer *printer;
450 /* Top of buffer stack. */
453 /* A buffer used for both for cpp_get_token's output, and also internally. */
454 unsigned char *token_buffer;
455 /* Allocated size of token_buffer. CPP_RESERVE allocates space. */
456 unsigned int token_buffer_size;
457 /* End of the written part of token_buffer. */
458 unsigned char *limit;
460 /* Error counter for exit code */
463 /* Line and column where a newline was first seen in a string constant. */
464 unsigned int multiline_string_line;
465 unsigned int multiline_string_column;
467 /* Current depth in #include directives that use <...>. */
468 unsigned int system_include_depth;
470 /* Current depth of buffer stack. */
471 unsigned int buffer_stack_depth;
473 /* Current depth in #include directives. */
474 unsigned int include_depth;
476 /* Hash table of macros and assertions. See cpphash.c */
477 struct htab *hashtab;
479 /* Tree of other included files. See cppfiles.c */
480 struct splay_tree_s *all_include_files;
482 /* Chain of `actual directory' file_name_list entries,
484 struct file_name_list *actual_dirs;
486 /* Current maximum length of directory names in the search path
487 for include files. (Altered as we get more of them.) */
488 unsigned int max_include_len;
490 /* Potential controlling macro for the current buffer. This is only
491 live between the #endif and the end of file, and there can only
492 be one at a time, so it is per-reader not per-buffer. */
493 const cpp_hashnode *potential_control_macro;
495 /* Token column position adjustment owing to tabs in whitespace. */
496 unsigned int col_adjust;
498 /* Token list used to store logical lines with new lexer. */
499 cpp_toklist token_list;
501 /* Temporary token store. */
502 cpp_token **temp_tokens;
503 unsigned int temp_cap;
504 unsigned int temp_alloced;
505 unsigned int temp_used;
507 /* Date and time tokens. Calculated together if either is requested. */
511 /* The # of a the current directive. It may not be first in line if
512 we append, and finding it is tedious. */
513 const cpp_token *first_directive_token;
515 /* Context stack. Used for macro expansion and for determining
516 which macros are disabled. */
517 unsigned int context_cap;
518 unsigned int cur_context;
519 unsigned int no_expand_level;
520 unsigned int paste_level;
521 struct cpp_context *contexts;
523 /* Current arguments when scanning arguments. Used for pointer
525 struct macro_args *args;
527 /* Buffer of -M output. */
530 /* Obstack holding all macro hash nodes. This never shrinks.
532 struct obstack *hash_ob;
534 /* Obstack holding buffer and conditional structures. This is a
535 real stack. See cpplib.c */
536 struct obstack *buffer_ob;
538 /* User visible options. */
539 struct cpp_options opts;
541 /* Nonzero means we have printed (while error reporting) a list of
542 containing files that matches the current status. */
543 unsigned char input_stack_listing_current;
545 /* If non-zero, macros are not expanded. */
546 unsigned char no_macro_expand;
548 /* We're printed a warning recommending against using #import. */
549 unsigned char import_warning;
551 /* True after cpp_start_read completes. Used to inhibit some
552 warnings while parsing the command line. */
553 unsigned char done_initializing;
555 /* True if we are skipping a failed conditional group. */
556 unsigned char skipping;
558 /* Do we need to save paramter spellings. */
559 unsigned char save_parameter_spellings;
561 /* If we're in lex_line. */
562 unsigned char in_lex_line;
564 /* True if output_line_command needs to output a newline. */
565 unsigned char need_newline;
568 /* struct cpp_printer encapsulates state used to convert the stream of
569 tokens coming from cpp_get_token back into a text file. Not
570 everyone wants to do that, hence we separate the function. */
574 FILE *outf; /* stream to write to */
575 const char *last_fname; /* previous file name */
576 unsigned int last_id; /* did we just push? */
577 unsigned int lineno; /* line currently being written */
578 unsigned int written; /* low water mark in token buffer */
581 #define CPP_FATAL_LIMIT 1000
582 /* True if we have seen a "fatal" error. */
583 #define CPP_FATAL_ERRORS(READER) ((READER)->errors >= CPP_FATAL_LIMIT)
585 /* Macros for manipulating the token_buffer. */
587 /* Number of characters currently in PFILE's output buffer. */
588 #define CPP_WRITTEN(PFILE) ((size_t)((PFILE)->limit - (PFILE)->token_buffer))
589 #define CPP_PWRITTEN(PFILE) ((PFILE)->limit)
590 #define CPP_ADJUST_WRITTEN(PFILE,DELTA) ((PFILE)->limit += (DELTA))
591 #define CPP_SET_WRITTEN(PFILE,N) ((PFILE)->limit = (PFILE)->token_buffer + (N))
593 #define CPP_OPTION(PFILE, OPTION) ((PFILE)->opts.OPTION)
594 #define CPP_BUFFER(PFILE) ((PFILE)->buffer)
595 #define CPP_BUF_LINE(BUF) ((BUF)->lineno)
596 #define CPP_BUF_COLUMN(BUF, CUR) ((CUR) - (BUF)->line_base + pfile->col_adjust)
597 #define CPP_BUF_COL(BUF) CPP_BUF_COLUMN(BUF, (BUF)->cur)
599 /* Name under which this program was invoked. */
600 extern const char *progname;
602 /* The structure of a node in the hash table. The hash table
603 has entries for all tokens defined by #define commands (type T_MACRO),
604 plus some special tokens like __LINE__ (these each have their own
605 type, and the appropriate code is run when that type of node is seen.
606 It does not contain control words like "#define", which are recognized
607 by a separate piece of code. */
609 /* different flavors of hash nodes */
612 T_VOID = 0, /* no definition yet */
613 T_SPECLINE, /* `__LINE__' */
614 T_DATE, /* `__DATE__' */
615 T_FILE, /* `__FILE__' */
616 T_BASE_FILE, /* `__BASE_FILE__' */
617 T_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
618 T_TIME, /* `__TIME__' */
619 T_STDC, /* `__STDC__' */
620 T_POISON, /* poisoned identifier */
621 T_MACRO, /* a macro, either object-like or function-like */
622 T_ASSERTION /* predicate for #assert */
625 /* There is a slot in the hashnode for use by front ends when integrated
626 with cpplib. It holds a tree (see tree.h) but we mustn't drag that
627 header into every user of cpplib.h. cpplib does not do anything with
628 this slot except clear it when a new node is created. */
633 unsigned int hash; /* cached hash value */
634 unsigned short length; /* length of name */
635 ENUM_BITFIELD(node_type) type : 8; /* node type */
639 const cpp_toklist *expansion; /* a macro's replacement list. */
640 struct answer *answers; /* answers to an assertion. */
643 union tree_node *fe_value; /* front end value */
645 const unsigned char name[1]; /* name[length] */
648 extern int cpp_handle_options PARAMS ((cpp_reader *, int, char **));
649 extern void cpp_reader_init PARAMS ((cpp_reader *));
650 extern cpp_printer *cpp_printer_init PARAMS ((cpp_reader *, cpp_printer *));
651 extern int cpp_start_read PARAMS ((cpp_reader *, cpp_printer *, const char *));
652 extern void cpp_output_tokens PARAMS ((cpp_reader *, cpp_printer *,
654 extern void cpp_finish PARAMS ((cpp_reader *, cpp_printer *));
655 extern void cpp_cleanup PARAMS ((cpp_reader *));
657 extern const cpp_token *cpp_get_token PARAMS ((cpp_reader *));
659 extern void cpp_define PARAMS ((cpp_reader *, const char *));
660 extern void cpp_assert PARAMS ((cpp_reader *, const char *));
661 extern void cpp_undef PARAMS ((cpp_reader *, const char *));
662 extern void cpp_unassert PARAMS ((cpp_reader *, const char *));
664 extern void cpp_free_token_list PARAMS ((cpp_toklist *));
666 /* N.B. The error-message-printer prototypes have not been nicely
667 formatted because exgettext needs to see 'msgid' on the same line
668 as the name of the function in order to work properly. Only the
669 string argument gets a name in an effort to keep the lines from
670 getting ridiculously oversized. */
672 extern void cpp_ice PARAMS ((cpp_reader *, const char *msgid, ...))
674 extern void cpp_fatal PARAMS ((cpp_reader *, const char *msgid, ...))
676 extern void cpp_error PARAMS ((cpp_reader *, const char *msgid, ...))
678 extern void cpp_warning PARAMS ((cpp_reader *, const char *msgid, ...))
680 extern void cpp_pedwarn PARAMS ((cpp_reader *, const char *msgid, ...))
682 extern void cpp_notice PARAMS ((cpp_reader *, const char *msgid, ...))
684 extern void cpp_error_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
686 extern void cpp_warning_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
688 extern void cpp_pedwarn_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
690 extern void cpp_pedwarn_with_file_and_line PARAMS ((cpp_reader *, const char *, int, int, const char *msgid, ...))
692 extern void cpp_error_from_errno PARAMS ((cpp_reader *, const char *));
693 extern void cpp_notice_from_errno PARAMS ((cpp_reader *, const char *));
696 extern cpp_buffer *cpp_push_buffer PARAMS ((cpp_reader *,
697 const unsigned char *, long));
698 extern cpp_buffer *cpp_pop_buffer PARAMS ((cpp_reader *));
699 extern void cpp_scan_buffer PARAMS ((cpp_reader *, cpp_printer *));
700 extern void cpp_scan_buffer_nooutput PARAMS ((cpp_reader *));
701 extern int cpp_idcmp PARAMS ((const unsigned char *,
702 size_t, const char *));
705 extern int cpp_defined PARAMS ((cpp_reader *,
706 const unsigned char *, int));
707 extern cpp_hashnode *cpp_lookup PARAMS ((cpp_reader *,
708 const unsigned char *, int));
711 extern int cpp_included PARAMS ((cpp_reader *, const char *));
712 extern int cpp_read_file PARAMS ((cpp_reader *, const char *));
713 extern void cpp_make_system_header PARAMS ((cpp_reader *,
719 #endif /* __GCC_CPPLIB__ */