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"
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_STRING) /* 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_STRING) /* Only if output comments. */ \
136 TK(CPP_MACRO_ARG, SPELL_NONE) /* Macro argument. */ \
137 OP(CPP_PADDING, "") /* Whitespace for cpp0. */ \
138 OP(CPP_EOF, "EOL") /* End of line or file. */
150 /* C language kind, used when calling cpp_reader_init. */
151 enum c_lang {CLK_GNUC89 = 0, CLK_GNUC99, CLK_STDC89, CLK_STDC94, CLK_STDC99,
152 CLK_GNUCXX, CLK_CXX98, CLK_OBJC, CLK_OBJCXX, CLK_ASM};
154 /* Payload of a NUMBER, FLOAT, STRING, or COMMENT token. */
158 const unsigned char *text;
161 /* Flags for the cpp_token structure. */
162 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */
163 #define DIGRAPH (1 << 1) /* If it was a digraph. */
164 #define STRINGIFY_ARG (1 << 2) /* If macro argument to be stringified. */
165 #define PASTE_LEFT (1 << 3) /* If on LHS of a ## operator. */
166 #define NAMED_OP (1 << 4) /* C++ named operators. */
167 #define NO_EXPAND (1 << 5) /* Do not macro-expand this token. */
168 #define BOL (1 << 6) /* Token at beginning of line. */
170 /* A preprocessing token. This has been carefully packed and should
171 occupy 16 bytes on 32-bit hosts and 24 bytes on 64-bit hosts. */
174 unsigned int line; /* Logical line of first char of token. */
175 unsigned short col; /* Column of first char of token. */
176 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
177 unsigned char flags; /* flags - see above */
181 cpp_hashnode *node; /* An identifier. */
182 const cpp_token *source; /* Inherit padding from this token. */
183 struct cpp_string str; /* A string, or number. */
184 unsigned int arg_no; /* Argument no. for a CPP_MACRO_ARG. */
185 unsigned char c; /* Character represented by CPP_OTHER. */
189 /* A standalone character. We may want to make it unsigned for the
190 same reason we use unsigned char - to avoid signedness issues. */
191 typedef int cppchar_t;
193 /* Values for opts.dump_macros.
194 dump_only means inhibit output of the preprocessed text
195 and instead output the definitions of all user-defined
196 macros in a form suitable for use as input to cpp.
197 dump_names means pass #define and the macro name through to output.
198 dump_definitions means pass the whole definition (plus #define) through
200 enum { dump_none = 0, dump_only, dump_names, dump_definitions };
202 /* This structure is nested inside struct cpp_reader, and
203 carries all the options visible to the command line. */
206 /* Name of input and output files. */
207 const char *in_fname;
208 const char *out_fname;
210 /* Characters between tab stops. */
211 unsigned int tabstop;
213 /* Pending options - -D, -U, -A, -I, -ixxx. */
214 struct cpp_pending *pending;
216 /* File name which deps are being written to. This is 0 if deps are
217 being written to stdout. */
218 const char *deps_file;
220 /* Search paths for include files. */
221 struct search_path *quote_include; /* "" */
222 struct search_path *bracket_include; /* <> */
224 /* Map between header names and file names, used only on DOS where
225 file names are limited in length. */
226 struct file_name_map_list *map_list;
228 /* Directory prefix that should replace `/usr/lib/gcc-lib/TARGET/VERSION'
229 in the standard include file directories. */
230 const char *include_prefix;
231 unsigned int include_prefix_len;
233 /* -fleading_underscore sets this to "_". */
234 const char *user_label_prefix;
236 /* The language we're preprocessing. */
239 /* Non-0 means -v, so print the full set of include dirs. */
240 unsigned char verbose;
242 /* Nonzero means use extra default include directories for C++. */
243 unsigned char cplusplus;
245 /* Nonzero means handle cplusplus style comments */
246 unsigned char cplusplus_comments;
248 /* Nonzero means handle #import, for objective C. */
251 /* Nonzero means don't copy comments into the output file. */
252 unsigned char discard_comments;
254 /* Nonzero means process the ISO trigraph sequences. */
255 unsigned char trigraphs;
257 /* Nonzero means process the ISO digraph sequences. */
258 unsigned char digraphs;
260 /* Nonzero means to allow hexadecimal floats and LL suffixes. */
261 unsigned char extended_numbers;
263 /* Nonzero means print the names of included files rather than the
264 preprocessed output. 1 means just the #include "...", 2 means
265 #include <...> as well. */
266 unsigned char print_deps;
268 /* Nonzero if phony targets are created for each header. */
269 unsigned char deps_phony_targets;
271 /* Nonzero if missing .h files in -M output are assumed to be
272 generated files and not errors. */
273 unsigned char print_deps_missing_files;
275 /* If true, fopen (deps_file, "a") else fopen (deps_file, "w"). */
276 unsigned char print_deps_append;
278 /* Nonzero means print names of header files (-H). */
279 unsigned char print_include_names;
281 /* Nonzero means cpp_pedwarn causes a hard error. */
282 unsigned char pedantic_errors;
284 /* Nonzero means don't print warning messages. */
285 unsigned char inhibit_warnings;
287 /* Nonzero means don't suppress warnings from system headers. */
288 unsigned char warn_system_headers;
290 /* Nonzero means don't print error messages. Has no option to
291 select it, but can be set by a user of cpplib (e.g. fix-header). */
292 unsigned char inhibit_errors;
294 /* Nonzero means warn if slash-star appears in a comment. */
295 unsigned char warn_comments;
297 /* Nonzero means warn if there are any trigraphs. */
298 unsigned char warn_trigraphs;
300 /* Nonzero means warn if #import is used. */
301 unsigned char warn_import;
303 /* Nonzero means warn about various incompatibilities with
305 unsigned char warn_traditional;
307 /* Nonzero means turn warnings into errors. */
308 unsigned char warnings_are_errors;
310 /* Nonzero causes output not to be done, but directives such as
311 #define that have side effects are still obeyed. */
312 unsigned char no_output;
314 /* Nonzero means we should look for header.gcc files that remap file
318 /* Nonzero means don't output line number information. */
319 unsigned char no_line_commands;
321 /* Nonzero means -I- has been seen, so don't look for #include "foo"
322 the source-file directory. */
323 unsigned char ignore_srcdir;
325 /* Zero means dollar signs are punctuation. */
326 unsigned char dollars_in_ident;
328 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
329 unsigned char warn_undef;
331 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
334 /* Nonzero means give all the error messages the ANSI standard requires. */
335 unsigned char pedantic;
337 /* Nonzero means we're looking at already preprocessed code, so don't
338 bother trying to do macro expansion and whatnot. */
339 unsigned char preprocessed;
341 /* Nonzero disables all the standard directories for headers. */
342 unsigned char no_standard_includes;
344 /* Nonzero disables the C++-specific standard directories for headers. */
345 unsigned char no_standard_cplusplus_includes;
347 /* Nonzero means dump macros in some fashion - see above. */
348 unsigned char dump_macros;
350 /* Nonzero means pass #include lines through to the output. */
351 unsigned char dump_includes;
353 /* Print column number in error messages. */
354 unsigned char show_column;
356 /* Nonzero means handle C++ alternate operator names. */
357 unsigned char operator_names;
359 /* True if --help, --version or --target-help appeared in the
360 options. Stand-alone CPP should then bail out after option
361 parsing; drivers might want to continue printing help. */
362 unsigned char help_only;
368 /* Called when a new line of preprocessed output is started. */
369 void (*line_change) PARAMS ((cpp_reader *, const cpp_token *, int));
370 void (*file_change) PARAMS ((cpp_reader *, const struct line_map *));
371 void (*include) PARAMS ((cpp_reader *, unsigned int,
372 const unsigned char *, const cpp_token *));
373 void (*define) PARAMS ((cpp_reader *, unsigned int, cpp_hashnode *));
374 void (*undef) PARAMS ((cpp_reader *, unsigned int, cpp_hashnode *));
375 void (*ident) PARAMS ((cpp_reader *, unsigned int, const cpp_string *));
376 void (*def_pragma) PARAMS ((cpp_reader *, unsigned int));
379 #define CPP_FATAL_LIMIT 1000
380 /* True if we have seen a "fatal" error. */
381 #define CPP_FATAL_ERRORS(PFILE) (cpp_errors (PFILE) >= CPP_FATAL_LIMIT)
383 /* Name under which this program was invoked. */
384 extern const char *progname;
386 /* The structure of a node in the hash table. The hash table has
387 entries for all identifiers: either macros defined by #define
388 commands (type NT_MACRO), assertions created with #assert
389 (NT_ASSERTION), or neither of the above (NT_VOID). Builtin macros
390 like __LINE__ are flagged NODE_BUILTIN. Poisioned identifiers are
391 flagged NODE_POISONED. NODE_OPERATOR (C++ only) indicates an
392 identifier that behaves like an operator such as "xor".
393 NODE_DIAGNOSTIC is for speed in lex_token: it indicates a
394 diagnostic may be required for this node. Currently this only
395 applies to __VA_ARGS__ and poisoned identifiers. */
397 /* Hash node flags. */
398 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */
399 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */
400 #define NODE_BUILTIN (1 << 2) /* Builtin macro. */
401 #define NODE_DIAGNOSTIC (1 << 3) /* Possible diagnostic when lexed. */
402 #define NODE_WARN (1 << 4) /* Warn if redefined or undefined. */
403 #define NODE_DISABLED (1 << 5) /* A disabled macro. */
405 /* Different flavors of hash node. */
408 NT_VOID = 0, /* No definition yet. */
409 NT_MACRO, /* A macro of some form. */
410 NT_ASSERTION /* Predicate for #assert. */
413 /* Different flavors of builtin macro. _Pragma is an operator, but we
414 handle it with the builtin code for efficiency reasons. */
417 BT_SPECLINE = 0, /* `__LINE__' */
418 BT_DATE, /* `__DATE__' */
419 BT_FILE, /* `__FILE__' */
420 BT_BASE_FILE, /* `__BASE_FILE__' */
421 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
422 BT_TIME, /* `__TIME__' */
423 BT_STDC, /* `__STDC__' */
424 BT_PRAGMA /* `_Pragma' operator */
427 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE))
428 #define HT_NODE(NODE) ((ht_identifier *) (NODE))
429 #define NODE_LEN(NODE) HT_LEN (&(NODE)->ident)
430 #define NODE_NAME(NODE) HT_STR (&(NODE)->ident)
432 /* The common part of an identifier node shared amongst all 3 C front
433 ends. Also used to store CPP identifiers, which are a superset of
434 identifiers in the grammatical sense. */
437 struct ht_identifier ident;
438 unsigned short arg_index; /* Macro argument index. */
439 unsigned char directive_index; /* Index into directive table. */
440 unsigned char rid_code; /* Rid code - for front ends. */
441 ENUM_BITFIELD(node_type) type : 8; /* CPP node type. */
442 unsigned char flags; /* CPP flags. */
446 cpp_macro *macro; /* If a macro. */
447 struct answer *answers; /* Answers to an assertion. */
448 enum cpp_ttype operator; /* Code for a named operator. */
449 enum builtin_type builtin; /* Code for a builtin macro. */
453 /* Call this first to get a handle to pass to other functions. If you
454 want cpplib to manage its own hashtable, pass in a NULL pointer.
455 Or you can pass in an initialised hash table that cpplib will use;
456 this technique is used by the C front ends. */
457 extern cpp_reader *cpp_create_reader PARAMS ((struct ht *,
460 /* Call this to release the handle. Any use of the handle after this
461 function returns is invalid. Returns cpp_errors (pfile). */
462 extern int cpp_destroy PARAMS ((cpp_reader *));
464 /* Call these to get pointers to the options and callback structures
465 for a given reader. These pointers are good until you call
466 cpp_finish on that reader. You can either edit the callbacks
467 through the pointer returned from cpp_get_callbacks, or set them
468 with cpp_set_callbacks. */
469 extern cpp_options *cpp_get_options PARAMS ((cpp_reader *));
470 extern const struct line_maps *cpp_get_line_maps PARAMS ((cpp_reader *));
471 extern cpp_callbacks *cpp_get_callbacks PARAMS ((cpp_reader *));
472 extern void cpp_set_callbacks PARAMS ((cpp_reader *, cpp_callbacks *));
474 /* Now call cpp_handle_option[s] to handle 1[or more] switches. The
475 return value is the number of arguments used. If
476 cpp_handle_options returns without using all arguments, it couldn't
477 understand the next switch. When there are no switches left, you
478 must call cpp_post_options before calling cpp_start_read. Only
479 after cpp_post_options are the contents of the cpp_options
480 structure reliable. */
481 extern int cpp_handle_options PARAMS ((cpp_reader *, int, char **));
482 extern int cpp_handle_option PARAMS ((cpp_reader *, int, char **));
483 extern void cpp_post_options PARAMS ((cpp_reader *));
486 extern unsigned int cpp_errors PARAMS ((cpp_reader *));
488 extern unsigned int cpp_token_len PARAMS ((const cpp_token *));
489 extern unsigned char *cpp_token_as_text PARAMS ((cpp_reader *,
491 extern unsigned char *cpp_spell_token PARAMS ((cpp_reader *, const cpp_token *,
493 extern void cpp_register_pragma PARAMS ((cpp_reader *,
494 const char *, const char *,
495 void (*) PARAMS ((cpp_reader *))));
496 extern void cpp_register_pragma_space PARAMS ((cpp_reader *, const char *));
498 extern int cpp_start_read PARAMS ((cpp_reader *, const char *));
499 extern void cpp_finish PARAMS ((cpp_reader *));
500 extern int cpp_avoid_paste PARAMS ((cpp_reader *, const cpp_token *,
502 extern const cpp_token *cpp_get_token PARAMS ((cpp_reader *));
503 extern const unsigned char *cpp_macro_definition PARAMS ((cpp_reader *,
504 const cpp_hashnode *));
505 extern void _cpp_backup_tokens PARAMS ((cpp_reader *, unsigned int));
507 /* Evaluate a CPP_CHAR or CPP_WCHAR token. */
509 cpp_interpret_charconst PARAMS ((cpp_reader *, const cpp_token *,
510 int, int, unsigned int *));
512 extern void cpp_define PARAMS ((cpp_reader *, const char *));
513 extern void cpp_assert PARAMS ((cpp_reader *, const char *));
514 extern void cpp_undef PARAMS ((cpp_reader *, const char *));
515 extern void cpp_unassert PARAMS ((cpp_reader *, const char *));
517 extern cpp_buffer *cpp_push_buffer PARAMS ((cpp_reader *,
518 const unsigned char *, size_t,
520 extern int cpp_defined PARAMS ((cpp_reader *, const unsigned char *, int));
522 /* N.B. The error-message-printer prototypes have not been nicely
523 formatted because exgettext needs to see 'msgid' on the same line
524 as the name of the function in order to work properly. Only the
525 string argument gets a name in an effort to keep the lines from
526 getting ridiculously oversized. */
528 extern void cpp_ice PARAMS ((cpp_reader *, const char *msgid, ...))
530 extern void cpp_fatal PARAMS ((cpp_reader *, const char *msgid, ...))
532 extern void cpp_error PARAMS ((cpp_reader *, const char *msgid, ...))
534 extern void cpp_warning PARAMS ((cpp_reader *, const char *msgid, ...))
536 extern void cpp_pedwarn PARAMS ((cpp_reader *, const char *msgid, ...))
538 extern void cpp_notice PARAMS ((cpp_reader *, const char *msgid, ...))
540 extern void cpp_error_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
542 extern void cpp_warning_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
544 extern void cpp_pedwarn_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
546 extern void cpp_error_from_errno PARAMS ((cpp_reader *, const char *));
547 extern void cpp_notice_from_errno PARAMS ((cpp_reader *, const char *));
550 extern int cpp_ideq PARAMS ((const cpp_token *,
552 extern void cpp_output_line PARAMS ((cpp_reader *, FILE *));
553 extern void cpp_output_token PARAMS ((const cpp_token *, FILE *));
554 extern const char *cpp_type2name PARAMS ((enum cpp_ttype));
555 extern unsigned int cpp_parse_escape PARAMS ((cpp_reader *,
556 const unsigned char **,
557 const unsigned char *,
558 unsigned HOST_WIDE_INT, int));
562 /* Lookup an identifier in the hashtable. Puts the identifier in the
563 table if it is not already there. */
564 extern cpp_hashnode *cpp_lookup PARAMS ((cpp_reader *,
565 const unsigned char *,
568 typedef int (*cpp_cb) PARAMS ((cpp_reader *, cpp_hashnode *, void *));
569 extern void cpp_forall_identifiers PARAMS ((cpp_reader *,
573 extern void cpp_scan_nooutput PARAMS ((cpp_reader *));
574 extern int cpp_sys_macro_p PARAMS ((cpp_reader *));
577 extern int cpp_included PARAMS ((cpp_reader *, const char *));
578 extern void cpp_make_system_header PARAMS ((cpp_reader *, int, int));
584 #endif /* ! GCC_CPPLIB_H */