1 /* Definitions for CPP library.
2 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001
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"
33 /* For complex reasons, cpp_reader is also typedefed in c-pragma.h. */
34 #ifndef GCC_C_PRAGMA_H
35 typedef struct cpp_reader cpp_reader;
37 typedef struct cpp_buffer cpp_buffer;
38 typedef struct cpp_options cpp_options;
39 typedef struct cpp_token cpp_token;
40 typedef struct cpp_string cpp_string;
41 typedef struct cpp_hashnode cpp_hashnode;
42 typedef struct cpp_macro cpp_macro;
43 typedef struct cpp_lexer_pos cpp_lexer_pos;
44 typedef struct cpp_lookahead cpp_lookahead;
45 typedef struct cpp_callbacks cpp_callbacks;
48 struct file_name_map_list;
51 /* The first two groups, apart from '=', can appear in preprocessor
52 expressions. This allows a lookup table to be implemented in
55 The first group, to CPP_LAST_EQ, can be immediately followed by an
56 '='. The lexer needs operators ending in '=', like ">>=", to be in
57 the same order as their counterparts without the '=', like ">>". */
59 /* Positions in the table. */
60 #define CPP_LAST_EQ CPP_MAX
61 #define CPP_FIRST_DIGRAPH CPP_HASH
62 #define CPP_LAST_PUNCTUATOR CPP_DOT_STAR
67 OP(CPP_GREATER, ">") /* compare */ \
69 OP(CPP_PLUS, "+") /* math */ \
74 OP(CPP_AND, "&") /* bit ops */ \
77 OP(CPP_RSHIFT, ">>") \
78 OP(CPP_LSHIFT, "<<") \
79 OP(CPP_MIN, "<?") /* extension */ \
83 OP(CPP_AND_AND, "&&") /* logical */ \
87 OP(CPP_COMMA, ",") /* grouping */ \
88 OP(CPP_OPEN_PAREN, "(") \
89 OP(CPP_CLOSE_PAREN, ")") \
90 OP(CPP_EQ_EQ, "==") /* compare */ \
91 OP(CPP_NOT_EQ, "!=") \
92 OP(CPP_GREATER_EQ, ">=") \
93 OP(CPP_LESS_EQ, "<=") \
95 OP(CPP_PLUS_EQ, "+=") /* math */ \
96 OP(CPP_MINUS_EQ, "-=") \
97 OP(CPP_MULT_EQ, "*=") \
98 OP(CPP_DIV_EQ, "/=") \
99 OP(CPP_MOD_EQ, "%=") \
100 OP(CPP_AND_EQ, "&=") /* bit ops */ \
101 OP(CPP_OR_EQ, "|=") \
102 OP(CPP_XOR_EQ, "^=") \
103 OP(CPP_RSHIFT_EQ, ">>=") \
104 OP(CPP_LSHIFT_EQ, "<<=") \
105 OP(CPP_MIN_EQ, "<?=") /* extension */ \
106 OP(CPP_MAX_EQ, ">?=") \
107 /* Digraphs together, beginning with CPP_FIRST_DIGRAPH. */ \
108 OP(CPP_HASH, "#") /* digraphs */ \
109 OP(CPP_PASTE, "##") \
110 OP(CPP_OPEN_SQUARE, "[") \
111 OP(CPP_CLOSE_SQUARE, "]") \
112 OP(CPP_OPEN_BRACE, "{") \
113 OP(CPP_CLOSE_BRACE, "}") \
114 /* The remainder of the punctuation. Order is not significant. */ \
115 OP(CPP_SEMICOLON, ";") /* structure */ \
116 OP(CPP_ELLIPSIS, "...") \
117 OP(CPP_PLUS_PLUS, "++") /* increment */ \
118 OP(CPP_MINUS_MINUS, "--") \
119 OP(CPP_DEREF, "->") /* accessors */ \
121 OP(CPP_SCOPE, "::") \
122 OP(CPP_DEREF_STAR, "->*") \
123 OP(CPP_DOT_STAR, ".*") \
124 OP(CPP_ATSIGN, "@") /* used in Objective C */ \
126 TK(CPP_NAME, SPELL_IDENT) /* word */ \
127 TK(CPP_NUMBER, SPELL_STRING) /* 34_be+ta */ \
129 TK(CPP_CHAR, SPELL_STRING) /* 'char' */ \
130 TK(CPP_WCHAR, SPELL_STRING) /* L'char' */ \
131 TK(CPP_OTHER, SPELL_CHAR) /* stray punctuation */ \
133 TK(CPP_STRING, SPELL_STRING) /* "string" */ \
134 TK(CPP_WSTRING, SPELL_STRING) /* L"string" */ \
135 TK(CPP_HEADER_NAME, SPELL_STRING) /* <stdio.h> in #include */ \
137 TK(CPP_COMMENT, SPELL_STRING) /* Only if output comments. */ \
138 TK(CPP_MACRO_ARG, SPELL_NONE) /* Macro argument. */ \
139 OP(CPP_EOF, "EOL") /* End of line or file. */
151 /* C language kind, used when calling cpp_reader_init. */
152 enum c_lang {CLK_GNUC89 = 0, CLK_GNUC99, CLK_STDC89, CLK_STDC94, CLK_STDC99,
153 CLK_GNUCXX, CLK_CXX98, CLK_OBJC, CLK_OBJCXX, CLK_ASM};
155 /* Payload of a NUMBER, FLOAT, STRING, or COMMENT token. */
159 const unsigned char *text;
162 /* Flags for the cpp_token structure. */
163 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */
164 #define DIGRAPH (1 << 1) /* If it was a digraph. */
165 #define STRINGIFY_ARG (1 << 2) /* If macro argument to be stringified. */
166 #define PASTE_LEFT (1 << 3) /* If on LHS of a ## operator. */
167 #define NAMED_OP (1 << 4) /* C++ named operators. */
168 #define NO_EXPAND (1 << 5) /* Do not macro-expand this token. */
169 #define AVOID_LPASTE (1 << 6) /* Check left for accidental pastes. */
171 /* A preprocessing token. This has been carefully packed and should
172 occupy 12 bytes on 32-bit hosts and 16 bytes on 64-bit hosts. */
175 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
176 unsigned char flags; /* flags - see above */
180 cpp_hashnode *node; /* An identifier. */
181 struct cpp_string str; /* A string, or number. */
182 unsigned int arg_no; /* Argument no. for a CPP_MACRO_ARG. */
183 unsigned char c; /* Character represented by CPP_OTHER. */
187 /* The position of a token in the current file. */
191 unsigned int output_line;
195 typedef struct cpp_token_with_pos cpp_token_with_pos;
196 struct cpp_token_with_pos
202 /* Token lookahead. */
205 struct cpp_lookahead *next;
206 cpp_token_with_pos *tokens;
208 unsigned int cur, count, cap;
211 /* A standalone character. We may want to make it unsigned for the
212 same reason we use unsigned char - to avoid signedness issues. */
213 typedef int cppchar_t;
215 /* Values for opts.dump_macros.
216 dump_only means inhibit output of the preprocessed text
217 and instead output the definitions of all user-defined
218 macros in a form suitable for use as input to cpp.
219 dump_names means pass #define and the macro name through to output.
220 dump_definitions means pass the whole definition (plus #define) through
222 enum { dump_none = 0, dump_only, dump_names, dump_definitions };
224 /* This structure is nested inside struct cpp_reader, and
225 carries all the options visible to the command line. */
228 /* Name of input and output files. */
229 const char *in_fname;
230 const char *out_fname;
232 /* Characters between tab stops. */
233 unsigned int tabstop;
235 /* Pending options - -D, -U, -A, -I, -ixxx. */
236 struct cpp_pending *pending;
238 /* File name which deps are being written to. This is 0 if deps are
239 being written to stdout. */
240 const char *deps_file;
242 /* Search paths for include files. */
243 struct search_path *quote_include; /* "" */
244 struct search_path *bracket_include; /* <> */
246 /* Map between header names and file names, used only on DOS where
247 file names are limited in length. */
248 struct file_name_map_list *map_list;
250 /* Directory prefix that should replace `/usr/lib/gcc-lib/TARGET/VERSION'
251 in the standard include file directories. */
252 const char *include_prefix;
253 unsigned int include_prefix_len;
255 /* -fleading_underscore sets this to "_". */
256 const char *user_label_prefix;
258 /* The language we're preprocessing. */
261 /* Non-0 means -v, so print the full set of include dirs. */
262 unsigned char verbose;
264 /* Nonzero means use extra default include directories for C++. */
265 unsigned char cplusplus;
267 /* Nonzero means handle cplusplus style comments */
268 unsigned char cplusplus_comments;
270 /* Nonzero means handle #import, for objective C. */
273 /* Nonzero means don't copy comments into the output file. */
274 unsigned char discard_comments;
276 /* Nonzero means process the ISO trigraph sequences. */
277 unsigned char trigraphs;
279 /* Nonzero means process the ISO digraph sequences. */
280 unsigned char digraphs;
282 /* Nonzero means to allow hexadecimal floats and LL suffixes. */
283 unsigned char extended_numbers;
285 /* Nonzero means print the names of included files rather than the
286 preprocessed output. 1 means just the #include "...", 2 means
287 #include <...> as well. */
288 unsigned char print_deps;
290 /* Nonzero if phony targets are created for each header. */
291 unsigned char deps_phony_targets;
293 /* Nonzero if missing .h files in -M output are assumed to be
294 generated files and not errors. */
295 unsigned char print_deps_missing_files;
297 /* If true, fopen (deps_file, "a") else fopen (deps_file, "w"). */
298 unsigned char print_deps_append;
300 /* Nonzero means print names of header files (-H). */
301 unsigned char print_include_names;
303 /* Nonzero means cpp_pedwarn causes a hard error. */
304 unsigned char pedantic_errors;
306 /* Nonzero means don't print warning messages. */
307 unsigned char inhibit_warnings;
309 /* Nonzero means don't suppress warnings from system headers. */
310 unsigned char warn_system_headers;
312 /* Nonzero means don't print error messages. Has no option to
313 select it, but can be set by a user of cpplib (e.g. fix-header). */
314 unsigned char inhibit_errors;
316 /* Nonzero means warn if slash-star appears in a comment. */
317 unsigned char warn_comments;
319 /* Nonzero means warn if there are any trigraphs. */
320 unsigned char warn_trigraphs;
322 /* Nonzero means warn if #import is used. */
323 unsigned char warn_import;
325 /* Nonzero means warn about various incompatibilities with
327 unsigned char warn_traditional;
329 /* Nonzero means turn warnings into errors. */
330 unsigned char warnings_are_errors;
332 /* Nonzero causes output not to be done, but directives such as
333 #define that have side effects are still obeyed. */
334 unsigned char no_output;
336 /* Nonzero means we should look for header.gcc files that remap file
340 /* Nonzero means don't output line number information. */
341 unsigned char no_line_commands;
343 /* Nonzero means -I- has been seen, so don't look for #include "foo"
344 the source-file directory. */
345 unsigned char ignore_srcdir;
347 /* Zero means dollar signs are punctuation. */
348 unsigned char dollars_in_ident;
350 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
351 unsigned char warn_undef;
353 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
356 /* Nonzero means give all the error messages the ANSI standard requires. */
357 unsigned char pedantic;
359 /* Nonzero means we're looking at already preprocessed code, so don't
360 bother trying to do macro expansion and whatnot. */
361 unsigned char preprocessed;
363 /* Nonzero disables all the standard directories for headers. */
364 unsigned char no_standard_includes;
366 /* Nonzero disables the C++-specific standard directories for headers. */
367 unsigned char no_standard_cplusplus_includes;
369 /* Nonzero means dump macros in some fashion - see above. */
370 unsigned char dump_macros;
372 /* Nonzero means pass #include lines through to the output. */
373 unsigned char dump_includes;
375 /* Print column number in error messages. */
376 unsigned char show_column;
378 /* Treat C++ alternate operator names special. */
379 unsigned char operator_names;
381 /* True if --help, --version or --target-help appeared in the
382 options. Stand-alone CPP should then bail out after option
383 parsing; drivers might want to continue printing help. */
384 unsigned char help_only;
387 /* This structure is passed to the call back when changing file. */
388 enum cpp_fc_reason {FC_ENTER = 0, FC_LEAVE, FC_RENAME};
392 const char *filename;
396 typedef struct cpp_file_change cpp_file_change;
397 struct cpp_file_change
399 struct cpp_file_loc from; /* Line of #include or #line. */
400 struct cpp_file_loc to; /* Line after #include or #line, or start. */
401 enum cpp_fc_reason reason; /* Reason for change. */
402 unsigned char sysp; /* Nonzero if system header. */
403 unsigned char externc; /* Nonzero if wrapper needed. */
409 void (*file_change) PARAMS ((cpp_reader *, const cpp_file_change *));
410 void (*include) PARAMS ((cpp_reader *, const unsigned char *,
412 void (*define) PARAMS ((cpp_reader *, cpp_hashnode *));
413 void (*undef) PARAMS ((cpp_reader *, cpp_hashnode *));
414 void (*ident) PARAMS ((cpp_reader *, const cpp_string *));
415 void (*def_pragma) PARAMS ((cpp_reader *));
418 #define CPP_FATAL_LIMIT 1000
419 /* True if we have seen a "fatal" error. */
420 #define CPP_FATAL_ERRORS(PFILE) (cpp_errors (PFILE) >= CPP_FATAL_LIMIT)
422 /* Name under which this program was invoked. */
423 extern const char *progname;
425 /* Where does this buffer come from? A faked include, a source file,
426 a builtin macro, a command-line option, or a _Pragma operator. */
427 enum cpp_buffer_type {BUF_FAKE, BUF_FILE, BUF_BUILTIN,
428 BUF_CL_OPTION, BUF_PRAGMA};
430 /* The structure of a node in the hash table. The hash table has
431 entries for all identifiers: either macros defined by #define
432 commands (type NT_MACRO), assertions created with #assert
433 (NT_ASSERTION), or neither of the above (NT_VOID). Builtin macros
434 like __LINE__ are flagged NODE_BUILTIN. Poisioned identifiers are
435 flagged NODE_POISONED. NODE_OPERATOR (C++ only) indicates an
436 identifier that behaves like an operator such as "xor".
437 NODE_DIAGNOSTIC is for speed in lex_token: it indicates a
438 diagnostic may be required for this node. Currently this only
439 applies to __VA_ARGS__ and poisoned identifiers. */
441 /* Hash node flags. */
442 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */
443 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */
444 #define NODE_BUILTIN (1 << 2) /* Builtin macro. */
445 #define NODE_DIAGNOSTIC (1 << 3) /* Possible diagnostic when lexed. */
446 #define NODE_WARN (1 << 4) /* Warn if redefined or undefined. */
448 /* Different flavors of hash node. */
451 NT_VOID = 0, /* No definition yet. */
452 NT_MACRO, /* A macro of some form. */
453 NT_ASSERTION /* Predicate for #assert. */
456 /* Different flavors of builtin macro. */
459 BT_SPECLINE = 0, /* `__LINE__' */
460 BT_DATE, /* `__DATE__' */
461 BT_FILE, /* `__FILE__' */
462 BT_BASE_FILE, /* `__BASE_FILE__' */
463 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
464 BT_TIME, /* `__TIME__' */
465 BT_STDC /* `__STDC__' */
468 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE))
469 #define HT_NODE(NODE) ((ht_identifier *) (NODE))
470 #define NODE_LEN(NODE) HT_LEN (&(NODE)->ident)
471 #define NODE_NAME(NODE) HT_STR (&(NODE)->ident)
473 /* The common part of an identifier node shared amongst all 3 C front
474 ends. Also used to store CPP identifiers, which are a superset of
475 identifiers in the grammatical sense. */
478 struct ht_identifier ident;
479 unsigned short arg_index; /* Macro argument index. */
480 unsigned char directive_index; /* Index into directive table. */
481 unsigned char rid_code; /* Rid code - for front ends. */
482 ENUM_BITFIELD(node_type) type : 8; /* CPP node type. */
483 unsigned char flags; /* CPP flags. */
487 cpp_macro *macro; /* If a macro. */
488 struct answer *answers; /* Answers to an assertion. */
489 enum cpp_ttype operator; /* Code for a named operator. */
490 enum builtin_type builtin; /* Code for a builtin macro. */
494 /* Call this first to get a handle to pass to other functions. If you
495 want cpplib to manage its own hashtable, pass in a NULL pointer.
496 Or you can pass in an initialised hash table that cpplib will use;
497 this technique is used by the C front ends. */
498 extern cpp_reader *cpp_create_reader PARAMS ((struct ht *,
501 /* Call this to release the handle. Any use of the handle after this
502 function returns is invalid. Returns cpp_errors (pfile). */
503 extern int cpp_destroy PARAMS ((cpp_reader *));
505 /* Call these to get pointers to the options and callback structures
506 for a given reader. These pointers are good until you call
507 cpp_finish on that reader. You can either edit the callbacks
508 through the pointer returned from cpp_get_callbacks, or set them
509 with cpp_set_callbacks. */
510 extern cpp_options *cpp_get_options PARAMS ((cpp_reader *));
511 extern cpp_callbacks *cpp_get_callbacks PARAMS ((cpp_reader *));
512 extern void cpp_set_callbacks PARAMS ((cpp_reader *, cpp_callbacks *));
514 /* Now call cpp_handle_option[s] to handle 1[or more] switches. The
515 return value is the number of arguments used. If
516 cpp_handle_options returns without using all arguments, it couldn't
517 understand the next switch. When there are no switches left, you
518 must call cpp_post_options before calling cpp_start_read. Only
519 after cpp_post_options are the contents of the cpp_options
520 structure reliable. */
521 extern int cpp_handle_options PARAMS ((cpp_reader *, int, char **));
522 extern int cpp_handle_option PARAMS ((cpp_reader *, int, char **));
523 extern void cpp_post_options PARAMS ((cpp_reader *));
526 extern unsigned int cpp_errors PARAMS ((cpp_reader *));
528 extern unsigned int cpp_token_len PARAMS ((const cpp_token *));
529 extern unsigned char *cpp_token_as_text PARAMS ((cpp_reader *,
531 extern unsigned char *cpp_spell_token PARAMS ((cpp_reader *, const cpp_token *,
533 extern void cpp_register_pragma PARAMS ((cpp_reader *,
534 const char *, const char *,
535 void (*) PARAMS ((cpp_reader *))));
536 extern void cpp_register_pragma_space PARAMS ((cpp_reader *, const char *));
538 extern int cpp_start_read PARAMS ((cpp_reader *, const char *));
539 extern void cpp_finish PARAMS ((cpp_reader *));
540 extern int cpp_avoid_paste PARAMS ((cpp_reader *, const cpp_token *,
542 extern enum cpp_ttype cpp_can_paste PARAMS ((cpp_reader *, const cpp_token *,
543 const cpp_token *, int *));
544 extern void cpp_get_token PARAMS ((cpp_reader *, cpp_token *));
545 extern const cpp_lexer_pos *cpp_get_line PARAMS ((cpp_reader *));
546 extern const unsigned char *cpp_macro_definition PARAMS ((cpp_reader *,
547 const cpp_hashnode *));
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,
561 enum cpp_buffer_type,
563 extern cpp_buffer *cpp_pop_buffer PARAMS ((cpp_reader *));
564 extern int cpp_defined PARAMS ((cpp_reader *, const unsigned char *, int));
566 /* N.B. The error-message-printer prototypes have not been nicely
567 formatted because exgettext needs to see 'msgid' on the same line
568 as the name of the function in order to work properly. Only the
569 string argument gets a name in an effort to keep the lines from
570 getting ridiculously oversized. */
572 extern void cpp_ice PARAMS ((cpp_reader *, const char *msgid, ...))
574 extern void cpp_fatal PARAMS ((cpp_reader *, const char *msgid, ...))
576 extern void cpp_error PARAMS ((cpp_reader *, const char *msgid, ...))
578 extern void cpp_warning PARAMS ((cpp_reader *, const char *msgid, ...))
580 extern void cpp_pedwarn PARAMS ((cpp_reader *, const char *msgid, ...))
582 extern void cpp_notice PARAMS ((cpp_reader *, const char *msgid, ...))
584 extern void cpp_error_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
586 extern void cpp_warning_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
588 extern void cpp_pedwarn_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
590 extern void cpp_pedwarn_with_file_and_line PARAMS ((cpp_reader *, const char *, int, int, const char *msgid, ...))
592 extern void cpp_error_from_errno PARAMS ((cpp_reader *, const char *));
593 extern void cpp_notice_from_errno PARAMS ((cpp_reader *, const char *));
596 extern int cpp_ideq PARAMS ((const cpp_token *,
598 extern void cpp_output_line PARAMS ((cpp_reader *, FILE *));
599 extern void cpp_output_token PARAMS ((const cpp_token *, FILE *));
600 extern const char *cpp_type2name PARAMS ((enum cpp_ttype));
601 extern unsigned int cpp_parse_escape PARAMS ((cpp_reader *,
602 const unsigned char **,
603 const unsigned char *,
604 unsigned HOST_WIDE_INT, int));
608 /* Lookup an identifier in the hashtable. Puts the identifier in the
609 table if it is not already there. */
610 extern cpp_hashnode *cpp_lookup PARAMS ((cpp_reader *,
611 const unsigned char *,
614 typedef int (*cpp_cb) PARAMS ((cpp_reader *, cpp_hashnode *, void *));
615 extern void cpp_forall_identifiers PARAMS ((cpp_reader *,
619 extern void cpp_scan_buffer_nooutput PARAMS ((cpp_reader *, int));
620 extern void cpp_start_lookahead PARAMS ((cpp_reader *));
621 extern void cpp_stop_lookahead PARAMS ((cpp_reader *, int));
622 extern int cpp_sys_macro_p PARAMS ((cpp_reader *));
625 extern int cpp_included PARAMS ((cpp_reader *, const char *));
626 extern void cpp_make_system_header PARAMS ((cpp_reader *, int, int));
632 #endif /* ! GCC_CPPLIB_H */