1/* Part of CPP library.
2   Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002
3   Free Software Foundation, Inc.
4
5This program is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10This program is distributed in the hope that it will be useful,
11but WITHOUT ANY WARRANTY; without even the implied warranty of
12MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13GNU General Public License for more details.
14
15You should have received a copy of the GNU General Public License
16along with this program; if not, write to the Free Software
17Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
18
19/* This header defines all the internal data structures and functions
20   that need to be visible across files.  It's called cpphash.h for
21   historical reasons.  */
22
23#ifndef GCC_CPPHASH_H
24#define GCC_CPPHASH_H
25
26#include "hashtable.h"
27
28struct directive;		/* Deliberately incomplete.  */
29struct pending_option;
30struct op;
31
32#ifndef HAVE_UCHAR
33typedef unsigned char uchar;
34#endif
35#define U (const uchar *)  /* Intended use: U"string" */
36
37#define BITS_PER_CPPCHAR_T (CHAR_BIT * sizeof (cppchar_t))
38
39/* Test if a sign is valid within a preprocessing number.  */
40#define VALID_SIGN(c, prevc) \
41  (((c) == '+' || (c) == '-') && \
42   ((prevc) == 'e' || (prevc) == 'E' \
43    || (((prevc) == 'p' || (prevc) == 'P') \
44        && CPP_OPTION (pfile, extended_numbers))))
45
46#define CPP_OPTION(PFILE, OPTION) ((PFILE)->opts.OPTION)
47#define CPP_BUFFER(PFILE) ((PFILE)->buffer)
48#define CPP_BUF_COLUMN(BUF, CUR) ((CUR) - (BUF)->line_base + (BUF)->col_adjust)
49#define CPP_BUF_COL(BUF) CPP_BUF_COLUMN(BUF, (BUF)->cur)
50
51/* Maximum nesting of cpp_buffers.  We use a static limit, partly for
52   efficiency, and partly to limit runaway recursion.  */
53#define CPP_STACK_MAX 200
54
55/* Host alignment handling.  */
56struct dummy
57{
58  char c;
59  union
60  {
61    double d;
62    int *p;
63  } u;
64};
65
66#define DEFAULT_ALIGNMENT offsetof (struct dummy, u)
67#define CPP_ALIGN2(size, align) (((size) + ((align) - 1)) & ~((align) - 1))
68#define CPP_ALIGN(size) CPP_ALIGN2 (size, DEFAULT_ALIGNMENT)
69
70/* Each macro definition is recorded in a cpp_macro structure.
71   Variadic macros cannot occur with traditional cpp.  */
72struct cpp_macro
73{
74  /* Parameters, if any.  */
75  cpp_hashnode **params;
76
77  /* Replacement tokens (ISO) or replacement text (traditional).  See
78     comment at top of cpptrad.c for how traditional function-like
79     macros are encoded.  */
80  union
81  {
82    cpp_token *tokens;
83    const uchar *text;
84  } exp;
85
86  /* Definition line number.  */
87  unsigned int line;
88
89  /* Number of tokens in expansion, or bytes for traditional macros.  */
90  unsigned int count;
91
92  /* Number of parameters.  */
93  unsigned short paramc;
94
95  /* If a function-like macro.  */
96  unsigned int fun_like : 1;
97
98  /* If a variadic macro.  */
99  unsigned int variadic : 1;
100
101  /* If macro defined in system header.  */
102  unsigned int syshdr   : 1;
103
104  /* Nonzero if it has been expanded or had its existence tested.  */
105  unsigned int used     : 1;
106};
107
108#define _cpp_mark_macro_used(NODE) do {					\
109  if ((NODE)->type == NT_MACRO && !((NODE)->flags & NODE_BUILTIN))	\
110    (NODE)->value.macro->used = 1; } while (0)
111
112/* A generic memory buffer, and operations on it.  */
113typedef struct _cpp_buff _cpp_buff;
114struct _cpp_buff
115{
116  struct _cpp_buff *next;
117  unsigned char *base, *cur, *limit;
118};
119
120extern _cpp_buff *_cpp_get_buff PARAMS ((cpp_reader *, size_t));
121extern void _cpp_release_buff PARAMS ((cpp_reader *, _cpp_buff *));
122extern void _cpp_extend_buff PARAMS ((cpp_reader *, _cpp_buff **, size_t));
123extern _cpp_buff *_cpp_append_extend_buff PARAMS ((cpp_reader *, _cpp_buff *,
124						   size_t));
125extern void _cpp_free_buff PARAMS ((_cpp_buff *));
126extern unsigned char *_cpp_aligned_alloc PARAMS ((cpp_reader *, size_t));
127extern unsigned char *_cpp_unaligned_alloc PARAMS ((cpp_reader *, size_t));
128
129#define BUFF_ROOM(BUFF) (size_t) ((BUFF)->limit - (BUFF)->cur)
130#define BUFF_FRONT(BUFF) ((BUFF)->cur)
131#define BUFF_LIMIT(BUFF) ((BUFF)->limit)
132
133/* List of directories to look for include files in.  */
134struct search_path
135{
136  struct search_path *next;
137
138  /* NOTE: NAME may not be null terminated for the case of the current
139     file's directory!  */
140  const char *name;
141  unsigned int len;
142  /* We use these to tell if the directory mentioned here is a duplicate
143     of an earlier directory on the search path.  */
144  ino_t ino;
145  dev_t dev;
146  /* Nonzero if it is a system include directory.  */
147  int sysp;
148  /* Mapping of file names for this directory.  Only used on MS-DOS
149     and related platforms.  */
150  struct file_name_map *name_map;
151};
152
153/* #include types.  */
154enum include_type {IT_INCLUDE, IT_INCLUDE_NEXT, IT_IMPORT, IT_CMDLINE};
155
156union utoken
157{
158  const cpp_token *token;
159  const cpp_token **ptoken;
160};
161
162/* A "run" of tokens; part of a chain of runs.  */
163typedef struct tokenrun tokenrun;
164struct tokenrun
165{
166  tokenrun *next, *prev;
167  cpp_token *base, *limit;
168};
169
170/* Accessor macros for struct cpp_context.  */
171#define FIRST(c) ((c)->u.iso.first)
172#define LAST(c) ((c)->u.iso.last)
173#define CUR(c) ((c)->u.trad.cur)
174#define RLIMIT(c) ((c)->u.trad.rlimit)
175
176typedef struct cpp_context cpp_context;
177struct cpp_context
178{
179  /* Doubly-linked list.  */
180  cpp_context *next, *prev;
181
182  union
183  {
184    /* For ISO macro expansion.  Contexts other than the base context
185       are contiguous tokens.  e.g. macro expansions, expanded
186       argument tokens.  */
187    struct
188    {
189      union utoken first;
190      union utoken last;
191    } iso;
192
193    /* For traditional macro expansion.  */
194    struct
195    {
196      const uchar *cur;
197      const uchar *rlimit;
198    } trad;
199  } u;
200
201  /* If non-NULL, a buffer used for storage related to this context.
202     When the context is popped, the buffer is released.  */
203  _cpp_buff *buff;
204
205  /* For a macro context, the macro node, otherwise NULL.  */
206  cpp_hashnode *macro;
207
208  /* True if utoken element is token, else ptoken.  */
209  bool direct_p;
210};
211
212struct lexer_state
213{
214  /* Nonzero if first token on line is CPP_HASH.  */
215  unsigned char in_directive;
216
217  /* Nonzero if in a directive that will handle padding tokens itself.
218     #include needs this to avoid problems with computed include and
219     spacing between tokens.  */
220  unsigned char directive_wants_padding;
221
222  /* True if we are skipping a failed conditional group.  */
223  unsigned char skipping;
224
225  /* Nonzero if in a directive that takes angle-bracketed headers.  */
226  unsigned char angled_headers;
227
228  /* Nonzero if in a #if or #elif directive.  */
229  unsigned char in_expression;
230
231  /* Nonzero to save comments.  Turned off if discard_comments, and in
232     all directives apart from #define.  */
233  unsigned char save_comments;
234
235  /* Nonzero if we're mid-comment.  */
236  unsigned char lexing_comment;
237
238  /* Nonzero if lexing __VA_ARGS__ is valid.  */
239  unsigned char va_args_ok;
240
241  /* Nonzero if lexing poisoned identifiers is valid.  */
242  unsigned char poisoned_ok;
243
244  /* Nonzero to prevent macro expansion.  */
245  unsigned char prevent_expansion;
246
247  /* Nonzero when parsing arguments to a function-like macro.  */
248  unsigned char parsing_args;
249
250  /* Nonzero to skip evaluating part of an expression.  */
251  unsigned int skip_eval;
252};
253
254/* Special nodes - identifiers with predefined significance.  */
255struct spec_nodes
256{
257  cpp_hashnode *n_defined;		/* defined operator */
258  cpp_hashnode *n_true;			/* C++ keyword true */
259  cpp_hashnode *n_false;		/* C++ keyword false */
260  cpp_hashnode *n__VA_ARGS__;		/* C99 vararg macros */
261};
262
263/* Encapsulates state used to convert a stream of tokens into a text
264   file.  */
265struct printer
266{
267  FILE *outf;			/* Stream to write to.  */
268  const struct line_map *map;	/* Logical to physical line mappings.  */
269  const cpp_token *prev;	/* Previous token.  */
270  const cpp_token *source;	/* Source token for spacing.  */
271  unsigned int line;		/* Line currently being written.  */
272  unsigned char printed;	/* Nonzero if something output at line.  */
273};
274
275/* Represents the contents of a file cpplib has read in.  */
276struct cpp_buffer
277{
278  const unsigned char *cur;	 /* current position */
279  const unsigned char *backup_to; /* if peeked character is not wanted */
280  const unsigned char *rlimit; /* end of valid data */
281  const unsigned char *line_base; /* start of current line */
282
283  struct cpp_buffer *prev;
284
285  const unsigned char *buf;	 /* Entire character buffer.  */
286
287  /* Pointer into the include table; non-NULL if this is a file
288     buffer.  Used for include_next and to record control macros.  */
289  struct include_file *inc;
290
291  /* Value of if_stack at start of this file.
292     Used to prohibit unmatched #endif (etc) in an include file.  */
293  struct if_stack *if_stack;
294
295  /* Token column position adjustment owing to tabs in whitespace.  */
296  unsigned int col_adjust;
297
298  /* Contains PREV_WHITE and/or AVOID_LPASTE.  */
299  unsigned char saved_flags;
300
301  /* Because of the way the lexer works, -Wtrigraphs can sometimes
302     warn twice for the same trigraph.  This helps prevent that.  */
303  const unsigned char *last_Wtrigraphs;
304
305  /* True if we have already warned about C++ comments in this file.
306     The warning happens only for C89 extended mode with -pedantic on,
307     or for -Wtraditional, and only once per file (otherwise it would
308     be far too noisy).  */
309  unsigned char warned_cplusplus_comments;
310
311  /* True if we don't process trigraphs and escaped newlines.  True
312     for preprocessed input, command line directives, and _Pragma
313     buffers.  */
314  unsigned char from_stage3;
315
316  /* Nonzero means that the directory to start searching for ""
317     include files has been calculated and stored in "dir" below.  */
318  unsigned char search_cached;
319
320  /* At EOF, a buffer is automatically popped.  If RETURN_AT_EOF is
321     true, a CPP_EOF token is then returned.  Otherwise, the next
322     token from the enclosing buffer is returned.  */
323  bool return_at_eof;
324
325  /* The directory of the this buffer's file.  Its NAME member is not
326     allocated, so we don't need to worry about freeing it.  */
327  struct search_path dir;
328
329  /* Used for buffer overlays by cpptrad.c.  */
330  const uchar *saved_cur, *saved_rlimit;
331};
332
333/* A cpp_reader encapsulates the "state" of a pre-processor run.
334   Applying cpp_get_token repeatedly yields a stream of pre-processor
335   tokens.  Usually, there is only one cpp_reader object active.  */
336struct cpp_reader
337{
338  /* Top of buffer stack.  */
339  cpp_buffer *buffer;
340
341  /* Overlaid buffer (can be different after processing #include).  */
342  cpp_buffer *overlaid_buffer;
343
344  /* Lexer state.  */
345  struct lexer_state state;
346
347  /* Source line tracking.  */
348  struct line_maps line_maps;
349  const struct line_map *map;
350  unsigned int line;
351
352  /* The line of the '#' of the current directive.  */
353  unsigned int directive_line;
354
355  /* Memory buffers.  */
356  _cpp_buff *a_buff;		/* Aligned permanent storage.  */
357  _cpp_buff *u_buff;		/* Unaligned permanent storage.  */
358  _cpp_buff *free_buffs;	/* Free buffer chain.  */
359
360  /* Context stack.  */
361  struct cpp_context base_context;
362  struct cpp_context *context;
363
364  /* If in_directive, the directive if known.  */
365  const struct directive *directive;
366
367  /* The next -include-d file; NULL if they all are done.  If it
368     points to NULL, the last one is in progress, and
369     _cpp_maybe_push_include_file has yet to restore the line map.  */
370  struct pending_option **next_include_file;
371
372  /* Multiple inlcude optimisation.  */
373  const cpp_hashnode *mi_cmacro;
374  const cpp_hashnode *mi_ind_cmacro;
375  bool mi_valid;
376
377  /* Lexing.  */
378  cpp_token *cur_token;
379  tokenrun base_run, *cur_run;
380  unsigned int lookaheads;
381
382  /* Nonzero prevents the lexer from re-using the token runs.  */
383  unsigned int keep_tokens;
384
385  /* Error counter for exit code.  */
386  unsigned int errors;
387
388  /* Buffer to hold macro definition string.  */
389  unsigned char *macro_buffer;
390  unsigned int macro_buffer_len;
391
392  /* Tree of other included files.  See cppfiles.c.  */
393  struct splay_tree_s *all_include_files;
394
395  /* Current maximum length of directory names in the search path
396     for include files.  (Altered as we get more of them.)  */
397  unsigned int max_include_len;
398
399  /* Macros on or after this line are warned about if unused.  */
400  unsigned int first_unused_line;
401
402  /* Date and time text.  Calculated together if either is requested.  */
403  const uchar *date;
404  const uchar *time;
405
406  /* EOF token, and a token forcing paste avoidance.  */
407  cpp_token avoid_paste;
408  cpp_token eof;
409
410  /* Opaque handle to the dependencies of mkdeps.c.  */
411  struct deps *deps;
412
413  /* Obstack holding all macro hash nodes.  This never shrinks.
414     See cpphash.c */
415  struct obstack hash_ob;
416
417  /* Obstack holding buffer and conditional structures.  This is a
418     real stack.  See cpplib.c.  */
419  struct obstack buffer_ob;
420
421  /* Pragma table - dynamic, because a library user can add to the
422     list of recognized pragmas.  */
423  struct pragma_entry *pragmas;
424
425  /* Call backs.  */
426  struct cpp_callbacks cb;
427
428  /* Identifier hash table.  */
429  struct ht *hash_table;
430
431  /* Expression parser stack.  */
432  struct op *op_stack, *op_limit;
433
434  /* User visible options.  */
435  struct cpp_options opts;
436
437  /* Special nodes - identifiers with predefined significance to the
438     preprocessor.  */
439  struct spec_nodes spec_nodes;
440
441  /* Used when doing preprocessed output.  */
442  struct printer print;
443
444  /* Whether cpplib owns the hashtable.  */
445  unsigned char our_hashtable;
446
447  /* Traditional preprocessing output buffer (a logical line).  */
448  struct
449  {
450    uchar *base;
451    uchar *limit;
452    uchar *cur;
453    unsigned int first_line;
454  } out;
455
456  /* Used to save the original line number during traditional
457     preprocessing.  */
458  unsigned int saved_line;
459};
460
461/* Character classes.  Based on the more primitive macros in safe-ctype.h.
462   If the definition of `numchar' looks odd to you, please look up the
463   definition of a pp-number in the C standard [section 6.4.8 of C99].
464
465   In the unlikely event that characters other than \r and \n enter
466   the set is_vspace, the macro handle_newline() in cpplex.c must be
467   updated.  */
468#define _dollar_ok(x)	((x) == '$' && CPP_OPTION (pfile, dollars_in_ident))
469
470#define is_idchar(x)	(ISIDNUM(x) || _dollar_ok(x))
471#define is_numchar(x)	ISIDNUM(x)
472#define is_idstart(x)	(ISIDST(x) || _dollar_ok(x))
473#define is_numstart(x)	ISDIGIT(x)
474#define is_hspace(x)	ISBLANK(x)
475#define is_vspace(x)	IS_VSPACE(x)
476#define is_nvspace(x)	IS_NVSPACE(x)
477#define is_space(x)	IS_SPACE_OR_NUL(x)
478
479/* This table is constant if it can be initialized at compile time,
480   which is the case if cpp was compiled with GCC >=2.7, or another
481   compiler that supports C99.  */
482#if HAVE_DESIGNATED_INITIALIZERS
483extern const unsigned char _cpp_trigraph_map[UCHAR_MAX + 1];
484#else
485extern unsigned char _cpp_trigraph_map[UCHAR_MAX + 1];
486#endif
487
488/* Macros.  */
489
490#define CPP_IN_SYSTEM_HEADER(PFILE) ((PFILE)->map && (PFILE)->map->sysp)
491#define CPP_PEDANTIC(PF) CPP_OPTION (PF, pedantic)
492#define CPP_WTRADITIONAL(PF) CPP_OPTION (PF, warn_traditional)
493
494/* In cpperror.c  */
495extern int _cpp_begin_message PARAMS ((cpp_reader *, int,
496				       unsigned int, unsigned int));
497
498/* In cppmacro.c */
499extern void _cpp_free_definition	PARAMS ((cpp_hashnode *));
500extern bool _cpp_create_definition	PARAMS ((cpp_reader *, cpp_hashnode *));
501extern void _cpp_pop_context		PARAMS ((cpp_reader *));
502extern void _cpp_push_text_context	PARAMS ((cpp_reader *, cpp_hashnode *,
503						 const uchar *, size_t));
504extern bool _cpp_save_parameter		PARAMS ((cpp_reader *, cpp_macro *,
505						 cpp_hashnode *));
506extern bool _cpp_arguments_ok		PARAMS ((cpp_reader *, cpp_macro *,
507						 const cpp_hashnode *,
508						 unsigned int));
509extern const uchar *_cpp_builtin_macro_text PARAMS ((cpp_reader *,
510						     cpp_hashnode *));
511int _cpp_warn_if_unused_macro		PARAMS ((cpp_reader *, cpp_hashnode *,
512						 void *));
513/* In cpphash.c */
514extern void _cpp_init_hashtable		PARAMS ((cpp_reader *, hash_table *));
515extern void _cpp_destroy_hashtable	PARAMS ((cpp_reader *));
516
517/* In cppfiles.c */
518extern void _cpp_fake_include		PARAMS ((cpp_reader *, const char *));
519extern void _cpp_never_reread		PARAMS ((struct include_file *));
520extern char *_cpp_simplify_pathname	PARAMS ((char *));
521extern bool _cpp_read_file		PARAMS ((cpp_reader *, const char *));
522extern bool _cpp_execute_include	PARAMS ((cpp_reader *,
523						 const cpp_token *,
524						 enum include_type));
525extern int _cpp_compare_file_date       PARAMS ((cpp_reader *,
526						 const cpp_token *));
527extern void _cpp_report_missing_guards	PARAMS ((cpp_reader *));
528extern void _cpp_init_includes		PARAMS ((cpp_reader *));
529extern void _cpp_cleanup_includes	PARAMS ((cpp_reader *));
530extern void _cpp_pop_file_buffer	PARAMS ((cpp_reader *,
531						 struct include_file *));
532
533/* In cppexp.c */
534extern bool _cpp_parse_expr		PARAMS ((cpp_reader *));
535extern struct op *_cpp_expand_op_stack	PARAMS ((cpp_reader *));
536
537/* In cpplex.c */
538extern cpp_token *_cpp_temp_token	PARAMS ((cpp_reader *));
539extern const cpp_token *_cpp_lex_token	PARAMS ((cpp_reader *));
540extern cpp_token *_cpp_lex_direct	PARAMS ((cpp_reader *));
541extern int _cpp_equiv_tokens		PARAMS ((const cpp_token *,
542						 const cpp_token *));
543extern void _cpp_init_tokenrun		PARAMS ((tokenrun *, unsigned int));
544
545/* In cppinit.c.  */
546extern void _cpp_maybe_push_include_file PARAMS ((cpp_reader *));
547
548/* In cpplib.c */
549extern int _cpp_test_assertion PARAMS ((cpp_reader *, unsigned int *));
550extern int _cpp_handle_directive PARAMS ((cpp_reader *, int));
551extern void _cpp_define_builtin	PARAMS ((cpp_reader *, const char *));
552extern void _cpp_do__Pragma	PARAMS ((cpp_reader *));
553extern void _cpp_init_directives PARAMS ((cpp_reader *));
554extern void _cpp_init_internal_pragmas PARAMS ((cpp_reader *));
555extern void _cpp_do_file_change PARAMS ((cpp_reader *, enum lc_reason,
556					 const char *,
557					 unsigned int, unsigned int));
558extern void _cpp_pop_buffer PARAMS ((cpp_reader *));
559
560/* In cpptrad.c.  */
561extern bool _cpp_read_logical_line_trad PARAMS ((cpp_reader *));
562extern void _cpp_overlay_buffer PARAMS ((cpp_reader *pfile, const uchar *,
563					 size_t));
564extern void _cpp_remove_overlay PARAMS ((cpp_reader *));
565extern bool _cpp_create_trad_definition PARAMS ((cpp_reader *, cpp_macro *));
566extern bool _cpp_expansions_different_trad PARAMS ((const cpp_macro *,
567						    const cpp_macro *));
568extern uchar *_cpp_copy_replacement_text PARAMS ((const cpp_macro *, uchar *));
569extern size_t _cpp_replacement_text_len PARAMS ((const cpp_macro *));
570
571/* Utility routines and macros.  */
572#define DSC(str) (const uchar *)str, sizeof str - 1
573#define xnew(T)		(T *) xmalloc (sizeof(T))
574#define xcnew(T)	(T *) xcalloc (1, sizeof(T))
575#define xnewvec(T, N)	(T *) xmalloc (sizeof(T) * (N))
576#define xcnewvec(T, N)	(T *) xcalloc (N, sizeof(T))
577#define xobnew(O, T)	(T *) obstack_alloc (O, sizeof(T))
578
579/* These are inline functions instead of macros so we can get type
580   checking.  */
581static inline int ustrcmp	PARAMS ((const uchar *, const uchar *));
582static inline int ustrncmp	PARAMS ((const uchar *, const uchar *,
583					 size_t));
584static inline size_t ustrlen	PARAMS ((const uchar *));
585static inline uchar *uxstrdup	PARAMS ((const uchar *));
586static inline uchar *ustrchr	PARAMS ((const uchar *, int));
587static inline int ufputs	PARAMS ((const uchar *, FILE *));
588
589static inline int
590ustrcmp (s1, s2)
591     const uchar *s1, *s2;
592{
593  return strcmp ((const char *)s1, (const char *)s2);
594}
595
596static inline int
597ustrncmp (s1, s2, n)
598     const uchar *s1, *s2;
599     size_t n;
600{
601  return strncmp ((const char *)s1, (const char *)s2, n);
602}
603
604static inline size_t
605ustrlen (s1)
606     const uchar *s1;
607{
608  return strlen ((const char *)s1);
609}
610
611static inline uchar *
612uxstrdup (s1)
613     const uchar *s1;
614{
615  return (uchar *) xstrdup ((const char *)s1);
616}
617
618static inline uchar *
619ustrchr (s1, c)
620     const uchar *s1;
621     int c;
622{
623  return (uchar *) strchr ((const char *)s1, c);
624}
625
626static inline int
627ufputs (s, f)
628     const uchar *s;
629     FILE *f;
630{
631  return fputs ((const char *)s, f);
632}
633
634#endif /* ! GCC_CPPHASH_H */
635