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
;
39 /* The first two groups, apart from '=', can appear in preprocessor
40 expressions. This allows a lookup table to be implemented in
43 The first group, to CPP_LAST_EQ, can be immediately followed by an
44 '='. The lexer needs operators ending in '=', like ">>=", to be in
45 the same order as their counterparts without the '=', like ">>". */
47 /* Positions in the table. */
48 #define CPP_LAST_EQ CPP_LSHIFT
49 #define CPP_FIRST_DIGRAPH CPP_HASH
54 T(CPP_GREATER, ">") /* compare */ \
56 T(CPP_PLUS, "+") /* math */ \
61 T(CPP_AND, "&") /* bit ops */ \
68 T(CPP_AND_AND, "&&") /* logical */ \
72 T(CPP_COMMA, ",") /* grouping */ \
73 T(CPP_OPEN_PAREN, "(") \
74 T(CPP_CLOSE_PAREN, ")") \
75 T(CPP_EQ_EQ, "==") /* compare */ \
77 T(CPP_GREATER_EQ, ">=") \
78 T(CPP_LESS_EQ, "<=") \
80 T(CPP_PLUS_EQ, "+=") /* math */ \
81 T(CPP_MINUS_EQ, "-=") \
82 T(CPP_MULT_EQ, "*=") \
85 T(CPP_AND_EQ, "&=") /* bit ops */ \
88 T(CPP_RSHIFT_EQ, ">>=") \
89 T(CPP_LSHIFT_EQ, "<<=") \
90 /* Digraphs together, beginning with CPP_FIRST_DIGRAPH. */ \
91 T(CPP_HASH, "#") /* digraphs */ \
93 T(CPP_OPEN_SQUARE, "[") \
94 T(CPP_CLOSE_SQUARE, "]") \
95 T(CPP_OPEN_BRACE, "{") \
96 T(CPP_CLOSE_BRACE, "}") \
97 /* The remainder of the punctuation. Order is not significant. */ \
98 T(CPP_SEMICOLON, ";") /* structure */ \
99 T(CPP_ELLIPSIS, "...") \
100 T(CPP_BACKSLASH, "\\") \
101 T(CPP_PLUS_PLUS, "++") /* increment */ \
102 T(CPP_MINUS_MINUS, "--") \
103 T(CPP_DEREF, "->") /* accessors */ \
106 T(CPP_DEREF_STAR, "->*") \
107 T(CPP_DOT_STAR, ".*") \
108 T(CPP_MIN, "<?") /* extension */ \
110 C(CPP_OTHER, 0) /* stray punctuation */ \
112 I(CPP_NAME, 0) /* word */ \
113 I(CPP_INT, 0) /* 23 */ \
114 I(CPP_FLOAT, 0) /* 3.14159 */ \
115 I(CPP_NUMBER, 0) /* 34_be+ta */ \
116 S(CPP_CHAR, 0) /* 'char' */ \
117 S(CPP_WCHAR, 0) /* L'char' */ \
118 S(CPP_STRING, 0) /* "string" */ \
119 S(CPP_WSTRING, 0) /* L"string" */ \
121 I(CPP_COMMENT, 0) /* Only if output comments. */ \
122 N(CPP_MACRO_ARG, 0) /* Macro argument. */ \
123 N(CPP_SUBLIST, 0) /* Sublist. */ \
124 T(CPP_VSPACE, "\n") /* End of line. */ \
125 N(CPP_EOF, 0) /* End of file. */ \
126 N(CPP_HEADER_NAME, 0) /* <stdio.h> in #include */ \
127 N(CPP_ASSERTION, 0) /* (...) in #assert */ \
129 /* Obsolete - will be removed when no code uses them still. */ \
130 N(CPP_HSPACE, 0) /* Horizontal white space. */ \
131 N(CPP_DIRECTIVE, 0) /* #define and the like */ \
132 N(CPP_MACRO, 0) /* Like a NAME, but expanded. */
150 /* Payload of a NAME, NUMBER, FLOAT, STRING, or COMMENT token. */
154 const unsigned char *text
;
157 /* Accessor macros for token lists - all expect you have a
158 list and an index. */
160 #define TOK_TYPE(l_, i_) ((l_)->tokens[i_].type)
161 #define TOK_FLAGS(l_, i_) ((l_)->tokens[i_].flags)
162 #define TOK_AUX(l_, i_) ((l_)->tokens[i_].aux)
163 #define TOK_COL(l_, i_) ((l_)->tokens[i_].col)
164 #define TOK_INT(l_, i_) ((l_)->tokens[i_].val.integer)
165 #define TOK_NAME(l_, i_) ((l_)->tokens[i_].val.name.text)
166 #define TOK_LEN(l_, i_) ((l_)->tokens[i_].val.name.len)
168 #define TOK_PREV_WHITE(l_, i_) (TOK_FLAGS(l_, i_) & PREV_WHITESPACE)
170 /* Flags for the cpp_token structure. */
171 #define PREV_WHITESPACE 1 /* If whitespace before this token. */
172 #define DIGRAPH 2 /* If it was a digraph. */
173 #define UNSIGNED_INT 4 /* If int preprocessing token unsigned. */
175 /* A preprocessing token. This has been carefully packed and should
176 occupy 16 bytes on both 32- and 64-bit hosts. */
179 unsigned short col
; /* starting column of this token */
180 ENUM_BITFIELD(cpp_ttype
) type
: CHAR_BIT
; /* node type */
181 unsigned char flags
; /* flags - see above */
182 unsigned int aux
; /* CPP_OTHER character. Hash of a
183 NAME, or something - see uses
187 struct cpp_name name
; /* a string */
188 HOST_WIDEST_INT integer
; /* an integer */
192 /* Directive flags. */
193 #define SYNTAX_INCLUDE (1 << 8)
195 typedef int (*directive_handler
) PARAMS ((cpp_reader
*));
199 cpp_token
*tokens
; /* actual tokens as an array */
200 unsigned int tokens_used
; /* tokens used */
201 unsigned int tokens_cap
; /* tokens allocated */
203 unsigned char *namebuf
; /* names buffer */
204 unsigned int name_used
; /* _bytes_ used */
205 unsigned int name_cap
; /* _bytes_ allocated */
207 unsigned int line
; /* starting line number */
209 /* Comment copying. */
210 cpp_token
*comments
; /* comment tokens. */
211 unsigned int comments_used
; /* comment tokens used. */
212 unsigned int comments_cap
; /* comment token capacity. */
214 /* Only used if tokens[0].type == CPP_DIRECTIVE. This is the
215 handler to call after lexing the rest of this line. The flags
216 indicate whether the rest of the line gets special treatment
217 during lexing (#include, #if, #assert, #unassert). */
218 directive_handler dir_handler
;
219 unsigned short dir_flags
;
224 const unsigned char *cur
; /* current position */
225 const unsigned char *rlimit
; /* end of valid data */
226 const unsigned char *buf
; /* entire buffer */
227 const unsigned char *line_base
; /* start of current line */
228 const unsigned char *mark
; /* Saved position for lengthy backtrack. */
230 struct cpp_buffer
*prev
;
232 /* Filename specified with #line command. */
233 const char *nominal_fname
;
234 /* Actual directory of this file, used only for "" includes */
235 struct file_name_list
*actual_dir
;
237 /* Pointer into the include hash table. Used for include_next and
238 to record control macros. */
241 /* If the buffer is the expansion of a macro, this points to the
242 macro's hash table entry. */
243 struct hashnode
*macro
;
245 /* Value of if_stack at start of this file.
246 Used to prohibit unmatched #endif (etc) in an include file. */
247 struct if_stack
*if_stack
;
249 /* Line number at line_base (above). */
252 /* True if this is a header file included using <FILENAME>. */
253 char system_header_p
;
255 /* True if buffer contains escape sequences.
256 Currently there are two kinds:
257 "\r-" means following identifier should not be macro-expanded.
258 "\r " means a token-separator. This turns into " " in final output
259 if not stringizing and needed to separate tokens; otherwise nothing.
260 Any other two-character sequence beginning with \r is an error.
262 If this is NOT set, then \r is a one-character escape meaning backslash
263 newline. This is guaranteed not to occur in the middle of a token.
264 The two interpretations of \r do not conflict, because the two-character
265 escapes are used only in macro buffers, and backslash-newline is removed
266 from macro expansion text in collect_expansion and/or macarg. */
269 /* True if we have already warned about C++ comments in this file.
270 The warning happens only for C89 extended mode with -pedantic on,
271 or for -Wtraditional, and only once per file (otherwise it would
272 be far too noisy). */
273 char warned_cplusplus_comments
;
275 /* In a file buffer, true if this buffer's data is mmapped
276 (currently never the case). In a macro buffer, true if this
277 buffer's data must be freed. */
281 struct file_name_map_list
;
284 /* Maximum nesting of cpp_buffers. We use a static limit, partly for
285 efficiency, and partly to limit runaway recursion. */
286 #define CPP_STACK_MAX 200
288 /* Values for opts.dump_macros.
289 dump_only means inhibit output of the preprocessed text
290 and instead output the definitions of all user-defined
291 macros in a form suitable for use as input to cpp.
292 dump_names means pass #define and the macro name through to output.
293 dump_definitions means pass the whole definition (plus #define) through
295 enum { dump_none
= 0, dump_only
, dump_names
, dump_definitions
};
297 /* This structure is nested inside struct cpp_reader, and
298 carries all the options visible to the command line. */
301 /* Name of input and output files. */
302 const char *in_fname
;
303 const char *out_fname
;
305 /* Pending options - -D, -U, -A, -I, -ixxx. */
306 struct cpp_pending
*pending
;
308 /* File name which deps are being written to. This is 0 if deps are
309 being written to stdout. */
310 const char *deps_file
;
312 /* Target-name to write with the dependency information. */
315 /* Search paths for include files. */
316 struct file_name_list
*quote_include
; /* First dir to search for "file" */
317 struct file_name_list
*bracket_include
;/* First dir to search for <file> */
319 /* Map between header names and file names, used only on DOS where
320 file names are limited in length. */
321 struct file_name_map_list
*map_list
;
323 /* Directory prefix that should replace `/usr/lib/gcc-lib/TARGET/VERSION'
324 in the standard include file directories. */
325 const char *include_prefix
;
326 unsigned int include_prefix_len
;
328 /* Non-0 means -v, so print the full set of include dirs. */
329 unsigned char verbose
;
331 /* Nonzero means use extra default include directories for C++. */
332 unsigned char cplusplus
;
334 /* Nonzero means handle cplusplus style comments */
335 unsigned char cplusplus_comments
;
337 /* Nonzero means handle #import, for objective C. */
340 /* Nonzero means this is an assembly file, so ignore unrecognized
341 directives and the "# 33" form of #line, both of which are
342 probably comments. Also, permit unbalanced ' strings (again,
343 likely to be in comments). */
344 unsigned char lang_asm
;
346 /* Nonzero means this is Fortran, and we don't know where the
347 comments are, so permit unbalanced ' strings. Unlike lang_asm,
348 this does not ignore unrecognized directives. */
349 unsigned char lang_fortran
;
351 /* Nonzero means handle CHILL comment syntax and output CHILL string
352 delimiters for __DATE__ etc. */
355 /* Nonzero means don't copy comments into the output file. */
356 unsigned char discard_comments
;
358 /* Nonzero means process the ANSI trigraph sequences. */
359 unsigned char trigraphs
;
361 /* Nonzero means print the names of included files rather than the
362 preprocessed output. 1 means just the #include "...", 2 means
363 #include <...> as well. */
364 unsigned char print_deps
;
366 /* Nonzero if missing .h files in -M output are assumed to be
367 generated files and not errors. */
368 unsigned char print_deps_missing_files
;
370 /* If true, fopen (deps_file, "a") else fopen (deps_file, "w"). */
371 unsigned char print_deps_append
;
373 /* Nonzero means print names of header files (-H). */
374 unsigned char print_include_names
;
376 /* Nonzero means cpp_pedwarn causes a hard error. */
377 unsigned char pedantic_errors
;
379 /* Nonzero means don't print warning messages. */
380 unsigned char inhibit_warnings
;
382 /* Nonzero means don't print error messages. Has no option to
383 select it, but can be set by a user of cpplib (e.g. fix-header). */
384 unsigned char inhibit_errors
;
386 /* Nonzero means warn if slash-star appears in a comment. */
387 unsigned char warn_comments
;
389 /* Nonzero means warn if there are any trigraphs. */
390 unsigned char warn_trigraphs
;
392 /* Nonzero means warn if #import is used. */
393 unsigned char warn_import
;
395 /* Nonzero means warn if a macro argument is (or would be)
396 stringified with -traditional, and warn about directives
397 with the # indented from the beginning of the line. */
398 unsigned char warn_traditional
;
400 /* Nonzero means turn warnings into errors. */
401 unsigned char warnings_are_errors
;
403 /* Nonzero causes output not to be done, but directives such as
404 #define that have side effects are still obeyed. */
405 unsigned char no_output
;
407 /* Nonzero means we should look for header.gcc files that remap file
411 /* Nonzero means don't output line number information. */
412 unsigned char no_line_commands
;
414 /* Nonzero means -I- has been seen, so don't look for #include "foo"
415 the source-file directory. */
416 unsigned char ignore_srcdir
;
418 /* Zero means dollar signs are punctuation. */
419 unsigned char dollars_in_ident
;
421 /* Nonzero means try to imitate old fashioned non-ANSI preprocessor. */
422 unsigned char traditional
;
424 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
425 unsigned char warn_undef
;
427 /* Nonzero for the 1989 C Standard, including corrigenda and amendments. */
430 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
433 /* Nonzero means give all the error messages the ANSI standard requires. */
434 unsigned char pedantic
;
436 /* Nonzero means we're looking at already preprocessed code, so don't
437 bother trying to do macro expansion and whatnot. */
438 unsigned char preprocessed
;
440 /* Nonzero disables all the standard directories for headers. */
441 unsigned char no_standard_includes
;
443 /* Nonzero disables the C++-specific standard directories for headers. */
444 unsigned char no_standard_cplusplus_includes
;
446 /* Nonzero means dump macros in some fashion - see above. */
447 unsigned char dump_macros
;
449 /* Nonzero means pass all #define and #undef directives which we
450 actually process through to the output stream. This feature is
451 used primarily to allow cc1 to record the #defines and #undefs
452 for the sake of debuggers which understand about preprocessor
453 macros, but it may also be useful with -E to figure out how
454 symbols are defined, and where they are defined. */
455 unsigned char debug_output
;
457 /* Nonzero means pass #include lines through to the output. */
458 unsigned char dump_includes
;
460 /* Print column number in error messages. */
461 unsigned char show_column
;
465 /* A cpp_reader encapsulates the "state" of a pre-processor run.
466 Applying cpp_get_token repeatedly yields a stream of pre-processor
467 tokens. Usually, there is only one cpp_reader object active. */
471 /* Top of buffer stack. */
474 /* Token list used by get_directive_token. */
475 cpp_toklist directbuf
;
477 /* A buffer used for both for cpp_get_token's output, and also internally. */
478 unsigned char *token_buffer
;
479 /* Allocated size of token_buffer. CPP_RESERVE allocates space. */
480 unsigned int token_buffer_size
;
481 /* End of the written part of token_buffer. */
482 unsigned char *limit
;
484 /* Error counter for exit code */
487 /* Line where a newline was first seen in a string constant. */
488 unsigned int multiline_string_line
;
490 /* Current depth in #include directives that use <...>. */
491 unsigned int system_include_depth
;
493 /* Current depth of buffer stack. */
494 unsigned int buffer_stack_depth
;
496 /* Hash table of macros and assertions. See cpphash.c */
497 struct htab
*hashtab
;
499 /* Hash table of other included files. See cppfiles.c */
500 struct htab
*all_include_files
;
502 /* Chain of `actual directory' file_name_list entries,
504 struct file_name_list
*actual_dirs
;
506 /* Current maximum length of directory names in the search path
507 for include files. (Altered as we get more of them.) */
508 unsigned int max_include_len
;
510 struct if_stack
*if_stack
;
511 const unsigned char *potential_control_macro
;
513 /* Buffer of -M output. */
516 /* A buffer used only by read_and_prescan (in cppfiles.c), which is
517 allocated once per cpp_reader object to keep it off the stack. */
518 unsigned char *input_buffer
;
519 size_t input_buffer_len
;
521 /* User visible options. */
522 struct cpp_options opts
;
524 /* Nonzero means we have printed (while error reporting) a list of
525 containing files that matches the current status. */
526 unsigned char input_stack_listing_current
;
528 /* If non-zero, macros are not expanded. */
529 unsigned char no_macro_expand
;
531 /* If non-zero, directives cause a hard error. Used when parsing
533 unsigned char no_directives
;
535 /* We're printed a warning recommending against using #import. */
536 unsigned char import_warning
;
538 /* If true, characters between '<' and '>' are a single (string) token. */
539 unsigned char parsing_include_directive
;
541 /* If true, # introduces an assertion (see do_assert) */
542 unsigned char parsing_if_directive
;
544 /* If true, # and ## are the STRINGIZE and TOKPASTE operators */
545 unsigned char parsing_define_directive
;
547 /* True if escape sequences (as described for has_escapes in
548 parse_buffer) should be emitted. */
549 unsigned char output_escapes
;
551 /* 0: Have seen non-white-space on this line.
552 1: Only seen white space so far on this line.
553 2: Only seen white space so far in this file. */
554 unsigned char only_seen_white
;
556 /* True after cpp_start_read completes. Used to inhibit some
557 warnings while parsing the command line. */
558 unsigned char done_initializing
;
561 /* struct cpp_printer encapsulates state used to convert the stream of
562 tokens coming from cpp_get_token back into a text file. Not
563 everyone wants to do that, hence we separate the function. */
567 FILE *outf
; /* stream to write to */
568 const char *last_fname
; /* previous file name */
569 unsigned int last_bsd
; /* did we just push? */
570 unsigned int lineno
; /* line currently being written */
571 unsigned int written
; /* low water mark in token buffer */
574 #define CPP_FATAL_LIMIT 1000
575 /* True if we have seen a "fatal" error. */
576 #define CPP_FATAL_ERRORS(READER) ((READER)->errors >= CPP_FATAL_LIMIT)
578 /* Macros for manipulating the token_buffer. */
580 /* Number of characters currently in PFILE's output buffer. */
581 #define CPP_WRITTEN(PFILE) ((size_t)((PFILE)->limit - (PFILE)->token_buffer))
582 #define CPP_PWRITTEN(PFILE) ((PFILE)->limit)
583 #define CPP_ADJUST_WRITTEN(PFILE,DELTA) ((PFILE)->limit += (DELTA))
584 #define CPP_SET_WRITTEN(PFILE,N) ((PFILE)->limit = (PFILE)->token_buffer + (N))
586 #define CPP_OPTION(PFILE, OPTION) ((PFILE)->opts.OPTION)
587 #define CPP_BUFFER(PFILE) ((PFILE)->buffer)
588 #define CPP_BUF_LINE(BUF) ((BUF)->lineno)
589 #define CPP_BUF_COL(BUF) ((BUF)->cur - (BUF)->line_base)
591 /* Name under which this program was invoked. */
592 extern const char *progname
;
594 extern void _cpp_lex_file
PARAMS((cpp_reader
*));
595 extern int cpp_handle_options
PARAMS ((cpp_reader
*, int, char **));
596 extern enum cpp_ttype cpp_get_token
PARAMS ((cpp_reader
*));
597 extern enum cpp_ttype cpp_get_non_space_token
PARAMS ((cpp_reader
*));
599 extern void cpp_reader_init
PARAMS ((cpp_reader
*));
600 extern cpp_printer
*cpp_printer_init
PARAMS ((cpp_reader
*, cpp_printer
*));
601 extern int cpp_start_read
PARAMS ((cpp_reader
*, cpp_printer
*, const char *));
602 extern void cpp_output_tokens
PARAMS ((cpp_reader
*, cpp_printer
*));
603 extern void cpp_output_list
PARAMS ((cpp_reader
*, cpp_printer
*,
604 const cpp_toklist
*));
605 extern void cpp_finish
PARAMS ((cpp_reader
*, cpp_printer
*));
606 extern void cpp_cleanup
PARAMS ((cpp_reader
*));
608 extern cpp_buffer
*cpp_file_buffer
PARAMS((cpp_reader
*));
609 extern void cpp_define
PARAMS ((cpp_reader
*, const char *));
610 extern void cpp_assert
PARAMS ((cpp_reader
*, const char *));
611 extern void cpp_undef
PARAMS ((cpp_reader
*, const char *));
612 extern void cpp_unassert
PARAMS ((cpp_reader
*, const char *));
614 extern void cpp_free_token_list
PARAMS ((cpp_toklist
*));
616 /* N.B. The error-message-printer prototypes have not been nicely
617 formatted because exgettext needs to see 'msgid' on the same line
618 as the name of the function in order to work properly. Only the
619 string argument gets a name in an effort to keep the lines from
620 getting ridiculously oversized. */
622 extern void cpp_ice
PARAMS ((cpp_reader
*, const char *msgid
, ...))
624 extern void cpp_fatal
PARAMS ((cpp_reader
*, const char *msgid
, ...))
626 extern void cpp_error
PARAMS ((cpp_reader
*, const char *msgid
, ...))
628 extern void cpp_warning
PARAMS ((cpp_reader
*, const char *msgid
, ...))
630 extern void cpp_pedwarn
PARAMS ((cpp_reader
*, const char *msgid
, ...))
632 extern void cpp_notice
PARAMS ((cpp_reader
*, const char *msgid
, ...))
634 extern void cpp_error_with_line
PARAMS ((cpp_reader
*, int, int, const char *msgid
, ...))
636 extern void cpp_warning_with_line
PARAMS ((cpp_reader
*, int, int, const char *msgid
, ...))
638 extern void cpp_pedwarn_with_line
PARAMS ((cpp_reader
*, int, int, const char *msgid
, ...))
640 extern void cpp_pedwarn_with_file_and_line
PARAMS ((cpp_reader
*, const char *, int, int, const char *msgid
, ...))
642 extern void cpp_error_from_errno
PARAMS ((cpp_reader
*, const char *));
643 extern void cpp_notice_from_errno
PARAMS ((cpp_reader
*, const char *));
646 extern cpp_buffer
*cpp_push_buffer
PARAMS ((cpp_reader
*,
647 const unsigned char *, long));
648 extern cpp_buffer
*cpp_pop_buffer
PARAMS ((cpp_reader
*));
649 extern void cpp_scan_buffer
PARAMS ((cpp_reader
*, cpp_printer
*));
650 extern void cpp_scan_buffer_nooutput
PARAMS ((cpp_reader
*));
651 extern int cpp_idcmp
PARAMS ((const unsigned char *,
652 size_t, const char *));
655 extern int cpp_defined
PARAMS ((cpp_reader
*,
656 const unsigned char *, int));
659 extern int cpp_included
PARAMS ((cpp_reader
*, const char *));
660 extern int cpp_read_file
PARAMS ((cpp_reader
*, const char *));
665 #endif /* __GCC_CPPLIB__ */