?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Document the problems with MS-Windows antivirus problems.
[emacs.git] / lib-src / ebrowse.c
blobb274a6410442c493a2ebfdc81fc21f05afc181a4
1 /* ebrowse.c --- parsing files for the ebrowse C++ browser
2
3 Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 2000, 2001
5 Free Software Foundation Inc.
6
7 Author: Gerd Moellmann <gerd@gnu.org>
8 Maintainer: FSF
9
10 This file is part of GNU Emacs.
11
12 GNU Emacs is free software; you can redistribute it and/or modify
13 it under the terms of the GNU General Public License as published by
14 the Free Software Foundation; either version 2, or (at your option)
15 any later version.
16
17 GNU Emacs is distributed in the hope that it will be useful,
18 but WITHOUT ANY WARRANTY; without even the implied warranty of
19 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 GNU General Public License for more details.
21
22 You should have received a copy of the GNU General Public License
23 along with GNU Emacs; see the file COPYING. If not, write to
24 the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
25
26 #ifdef HAVE_CONFIG_H
27 #include <config.h>
28 #endif
29
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <ctype.h>
34 #include <assert.h>
35 #include "getopt.h"
36
37 /* Conditionalize function prototypes. */
38
39 #ifdef PROTOTYPES /* From config.h. */
40 #define P_(x) x
41 #else
42 #define P_(x) ()
43 #endif
44
45 /* Value is non-zero if strings X and Y compare equal. */
46
47 #define streq(X, Y) (*(X) == *(Y) && strcmp ((X) + 1, (Y) + 1) == 0)
48
49 /* The ubiquitous `max' and `min' macros. */
50
51 #ifndef max
52 #define max(X, Y) ((X) > (Y) ? (X) : (Y))
53 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
54 #endif
55
56 /* Files are read in chunks of this number of bytes. */
57
58 #define READ_CHUNK_SIZE (100 * 1024)
59
60 /* The character used as a separator in path lists (like $PATH). */
61
62 #if defined(__MSDOS__)
63 #define PATH_LIST_SEPARATOR ';'
64 #define FILENAME_EQ(X,Y) (strcasecmp(X,Y) == 0)
65 #else
66 #if defined(WINDOWSNT)
67 #define PATH_LIST_SEPARATOR ';'
68 #define FILENAME_EQ(X,Y) (stricmp(X,Y) == 0)
69 #else
70 #define PATH_LIST_SEPARATOR ':'
71 #define FILENAME_EQ(X,Y) (streq(X,Y))
72 #endif
73 #endif
74 /* The default output file name. */
75
76 #define DEFAULT_OUTFILE "BROWSE"
77
78 /* A version string written to the output file. Change this whenever
79 the structure of the output file changes. */
80
81 #define EBROWSE_FILE_VERSION "ebrowse 5.0"
82
83 /* The output file consists of a tree of Lisp objects, with major
84 nodes built out of Lisp structures. These are the heads of the
85 Lisp structs with symbols identifying their type. */
86
87 #define TREE_HEADER_STRUCT "[ebrowse-hs "
88 #define TREE_STRUCT "[ebrowse-ts "
89 #define MEMBER_STRUCT "[ebrowse-ms "
90 #define BROWSE_STRUCT "[ebrowse-bs "
91 #define CLASS_STRUCT "[ebrowse-cs "
92
93 /* The name of the symbol table entry for global functions, variables,
94 defines etc. This name also appears in the browser display. */
95
96 #define GLOBALS_NAME "*Globals*"
97
98 /* Token definitions. */
99
100 enum token
101 {
102 YYEOF = 0, /* end of file */
103 CSTRING = 256, /* string constant */
104 CCHAR, /* character constant */
105 CINT, /* integral constant */
106 CFLOAT, /* real constant */
107
108 ELLIPSIS, /* ... */
109 LSHIFTASGN, /* <<= */
110 RSHIFTASGN, /* >>= */
111 ARROWSTAR, /* ->* */
112 IDENT, /* identifier */
113 DIVASGN, /* /= */
114 INC, /* ++ */
115 ADDASGN, /* += */
116 DEC, /* -- */
117 ARROW, /* -> */
118 SUBASGN, /* -= */
119 MULASGN, /* *= */
120 MODASGN, /* %= */
121 LOR, /* || */
122 ORASGN, /* |= */
123 LAND, /* && */
124 ANDASGN, /* &= */
125 XORASGN, /* ^= */
126 POINTSTAR, /* .* */
127 DCOLON, /* :: */
128 EQ, /* == */
129 NE, /* != */
130 LE, /* <= */
131 LSHIFT, /* << */
132 GE, /* >= */
133 RSHIFT, /* >> */
134
135 /* Keywords. The undef's are there because these
136 three symbols are very likely to be defined somewhere. */
137 #undef BOOL
138 #undef TRUE
139 #undef FALSE
140
141 ASM, /* asm */
142 AUTO, /* auto */
143 BREAK, /* break */
144 CASE, /* case */
145 CATCH, /* catch */
146 CHAR, /* char */
147 CLASS, /* class */
148 CONST, /* const */
149 CONTINUE, /* continue */
150 DEFAULT, /* default */
151 DELETE, /* delete */
152 DO, /* do */
153 DOUBLE, /* double */
154 ELSE, /* else */
155 ENUM, /* enum */
156 EXTERN, /* extern */
157 FLOAT, /* float */
158 FOR, /* for */
159 FRIEND, /* friend */
160 GOTO, /* goto */
161 IF, /* if */
162 T_INLINE, /* inline */
163 INT, /* int */
164 LONG, /* long */
165 NEW, /* new */
166 OPERATOR, /* operator */
167 PRIVATE, /* private */
168 PROTECTED, /* protected */
169 PUBLIC, /* public */
170 REGISTER, /* register */
171 RETURN, /* return */
172 SHORT, /* short */
173 SIGNED, /* signed */
174 SIZEOF, /* sizeof */
175 STATIC, /* static */
176 STRUCT, /* struct */
177 SWITCH, /* switch */
178 TEMPLATE, /* template */
179 THIS, /* this */
180 THROW, /* throw */
181 TRY, /* try */
182 TYPEDEF, /* typedef */
183 UNION, /* union */
184 UNSIGNED, /* unsigned */
185 VIRTUAL, /* virtual */
186 VOID, /* void */
187 VOLATILE, /* volatile */
188 WHILE, /* while */
189 MUTABLE, /* mutable */
190 BOOL, /* bool */
191 TRUE, /* true */
192 FALSE, /* false */
193 SIGNATURE, /* signature (GNU extension) */
194 NAMESPACE, /* namespace */
195 EXPLICIT, /* explicit */
196 TYPENAME, /* typename */
197 CONST_CAST, /* const_cast */
198 DYNAMIC_CAST, /* dynamic_cast */
199 REINTERPRET_CAST, /* reinterpret_cast */
200 STATIC_CAST, /* static_cast */
201 TYPEID, /* typeid */
202 USING, /* using */
203 WCHAR /* wchar_t */
204 };
205
206 /* Storage classes, in a wider sense. */
207
208 enum sc
209 {
210 SC_UNKNOWN,
211 SC_MEMBER, /* Is an instance member. */
212 SC_STATIC, /* Is static member. */
213 SC_FRIEND, /* Is friend function. */
214 SC_TYPE /* Is a type definition. */
215 };
216
217 /* Member visibility. */
218
219 enum visibility
220 {
221 V_PUBLIC,
222 V_PROTECTED,
223 V_PRIVATE
224 };
225
226 /* Member flags. */
227
228 #define F_VIRTUAL 1 /* Is virtual function. */
229 #define F_INLINE 2 /* Is inline function. */
230 #define F_CONST 4 /* Is const. */
231 #define F_PURE 8 /* Is pure virtual function. */
232 #define F_MUTABLE 16 /* Is mutable. */
233 #define F_TEMPLATE 32 /* Is a template. */
234 #define F_EXPLICIT 64 /* Is explicit constructor. */
235 #define F_THROW 128 /* Has a throw specification. */
236 #define F_EXTERNC 256 /* Is declared extern "C". */
237 #define F_DEFINE 512 /* Is a #define. */
238
239 /* Two macros to set and test a bit in an int. */
240
241 #define SET_FLAG(F, FLAG) ((F) |= (FLAG))
242 #define HAS_FLAG(F, FLAG) (((F) & (FLAG)) != 0)
243
244 /* Structure describing a class member. */
245
246 struct member
247 {
248 struct member *next; /* Next in list of members. */
249 struct member *anext; /* Collision chain in member_table. */
250 struct member **list; /* Pointer to list in class. */
251 unsigned param_hash; /* Hash value for parameter types. */
252 int vis; /* Visibility (public, ...). */
253 int flags; /* See F_* above. */
254 char *regexp; /* Matching regular expression. */
255 char *filename; /* Don't free this shared string. */
256 int pos; /* Buffer position of occurrence. */
257 char *def_regexp; /* Regular expression matching definition. */
258 char *def_filename; /* File name of definition. */
259 int def_pos; /* Buffer position of definition. */
260 char name[1]; /* Member name. */
261 };
262
263 /* Structures of this type are used to connect class structures with
264 their super and subclasses. */
265
266 struct link
267 {
268 struct sym *sym; /* The super or subclass. */
269 struct link *next; /* Next in list or NULL. */
270 };
271
272 /* Structure used to record namespace aliases. */
273
274 struct alias
275 {
276 struct alias *next; /* Next in list. */
277 char name[1]; /* Alias name. */
278 };
279
280 /* The structure used to describe a class in the symbol table,
281 or a namespace in all_namespaces. */
282
283 struct sym
284 {
285 int flags; /* Is class a template class?. */
286 unsigned char visited; /* Used to find circles. */
287 struct sym *next; /* Hash collision list. */
288 struct link *subs; /* List of subclasses. */
289 struct link *supers; /* List of superclasses. */
290 struct member *vars; /* List of instance variables. */
291 struct member *fns; /* List of instance functions. */
292 struct member *static_vars; /* List of static variables. */
293 struct member *static_fns; /* List of static functions. */
294 struct member *friends; /* List of friend functions. */
295 struct member *types; /* List of local types. */
296 char *regexp; /* Matching regular expression. */
297 int pos; /* Buffer position. */
298 char *filename; /* File in which it can be found. */
299 char *sfilename; /* File in which members can be found. */
300 struct sym *namesp; /* Namespace in which defined. . */
301 struct alias *namesp_aliases; /* List of aliases for namespaces. */
302 char name[1]; /* Name of the class. */
303 };
304
305 /* Experimental: Print info for `--position-info'. We print
306 '(CLASS-NAME SCOPE MEMBER-NAME). */
307
308 #define P_DEFN 1
309 #define P_DECL 2
310
311 int info_where;
312 struct sym *info_cls = NULL;
313 struct member *info_member = NULL;
314
315 /* Experimental. For option `--position-info', the buffer position we
316 are interested in. When this position is reached, print out
317 information about what we know about that point. */
318
319 int info_position = -1;
320
321 /* Command line options structure for getopt_long. */
322
323 struct option options[] =
324 {
325 {"append", no_argument, NULL, 'a'},
326 {"files", required_argument, NULL, 'f'},
327 {"help", no_argument, NULL, -2},
328 {"min-regexp-length", required_argument, NULL, 'm'},
329 {"max-regexp-length", required_argument, NULL, 'M'},
330 {"no-nested-classes", no_argument, NULL, 'n'},
331 {"no-regexps", no_argument, NULL, 'x'},
332 {"no-structs-or-unions", no_argument, NULL, 's'},
333 {"output-file", required_argument, NULL, 'o'},
334 {"position-info", required_argument, NULL, 'p'},
335 {"search-path", required_argument, NULL, 'I'},
336 {"verbose", no_argument, NULL, 'v'},
337 {"version", no_argument, NULL, -3},
338 {"very-verbose", no_argument, NULL, 'V'},
339 {NULL, 0, NULL, 0}
340 };
341
342 /* Semantic values of tokens. Set by yylex.. */
343
344 unsigned yyival; /* Set for token CINT. */
345 char *yytext; /* Set for token IDENT. */
346 char *yytext_end;
347
348 /* Output file. */
349
350 FILE *yyout;
351
352 /* Current line number. */
353
354 int yyline;
355
356 /* The name of the current input file. */
357
358 char *filename;
359
360 /* Three character class vectors, and macros to test membership
361 of characters. */
362
363 char is_ident[255];
364 char is_digit[255];
365 char is_white[255];
366
367 #define IDENTP(C) is_ident[(unsigned char) (C)]
368 #define DIGITP(C) is_digit[(unsigned char) (C)]
369 #define WHITEP(C) is_white[(unsigned char) (C)]
370
371 /* Command line flags. */
372
373 int f_append;
374 int f_verbose;
375 int f_very_verbose;
376 int f_structs = 1;
377 int f_regexps = 1;
378 int f_nested_classes = 1;
379
380 /* Maximum and minimum lengths of regular expressions matching a
381 member, class etc., for writing them to the output file. These are
382 overridable from the command line. */
383
384 int min_regexp = 5;
385 int max_regexp = 50;
386
387 /* Input buffer. */
388
389 char *inbuffer;
390 char *in;
391 int inbuffer_size;
392
393 /* Return the current buffer position in the input file. */
394
395 #define BUFFER_POS() (in - inbuffer)
396
397 /* If current lookahead is CSTRING, the following points to the
398 first character in the string constant. Used for recognizing
399 extern "C". */
400
401 char *string_start;
402
403 /* The size of the hash tables for classes.and members. Should be
404 prime. */
405
406 #define TABLE_SIZE 1001
407
408 /* The hash table for class symbols. */
409
410 struct sym *class_table[TABLE_SIZE];
411
412 /* Hash table containing all member structures. This is generally
413 faster for member lookup than traversing the member lists of a
414 `struct sym'. */
415
416 struct member *member_table[TABLE_SIZE];
417
418 /* The special class symbol used to hold global functions,
419 variables etc. */
420
421 struct sym *global_symbols;
422
423 /* The current namespace. */
424
425 struct sym *current_namespace;
426
427 /* The list of all known namespaces. */
428
429 struct sym *all_namespaces;
430
431 /* Stack of namespaces we're currently nested in, during the parse. */
432
433 struct sym **namespace_stack;
434 int namespace_stack_size;
435 int namespace_sp;
436
437 /* The current lookahead token. */
438
439 int tk = -1;
440
441 /* Structure describing a keyword. */
442
443 struct kw
444 {
445 char *name; /* Spelling. */
446 int tk; /* Token value. */
447 struct kw *next; /* Next in collision chain. */
448 };
449
450 /* Keywords are lookup up in a hash table of their own. */
451
452 #define KEYWORD_TABLE_SIZE 1001
453 struct kw *keyword_table[KEYWORD_TABLE_SIZE];
454
455 /* Search path. */
456
457 struct search_path
458 {
459 char *path;
460 struct search_path *next;
461 };
462
463 struct search_path *search_path;
464 struct search_path *search_path_tail;
465
466 /* Function prototypes. */
467
468 int yylex P_ ((void));
469 void yyparse P_ ((void));
470 void re_init_parser P_ ((void));
471 char *token_string P_ ((int));
472 char *matching_regexp P_ ((void));
473 void init_sym P_ ((void));
474 struct sym *add_sym P_ ((char *, struct sym *));
475 void add_link P_ ((struct sym *, struct sym *));
476 void add_member_defn P_ ((struct sym *, char *, char *,
477 int, unsigned, int, int, int));
478 void add_member_decl P_ ((struct sym *, char *, char *, int,
479 unsigned, int, int, int, int));
480 void dump_roots P_ ((FILE *));
481 void *xmalloc P_ ((int));
482 void xfree P_ ((void *));
483 void add_global_defn P_ ((char *, char *, int, unsigned, int, int, int));
484 void add_global_decl P_ ((char *, char *, int, unsigned, int, int, int));
485 void add_define P_ ((char *, char *, int));
486 void mark_inherited_virtual P_ ((void));
487 void leave_namespace P_ ((void));
488 void enter_namespace P_ ((char *));
489 void register_namespace_alias P_ ((char *, char *));
490 void insert_keyword P_ ((char *, int));
491 void re_init_scanner P_ ((void));
492 void init_scanner P_ ((void));
493 void usage P_ ((int));
494 void version P_ ((void));
495 void process_file P_ ((char *));
496 void add_search_path P_ ((char *));
497 FILE *open_file P_ ((char *));
498 int process_pp_line P_ ((void));
499 int dump_members P_ ((FILE *, struct member *));
500 void dump_sym P_ ((FILE *, struct sym *));
501 int dump_tree P_ ((FILE *, struct sym *));
502 struct member *find_member P_ ((struct sym *, char *, int, int, unsigned));
503 struct member *add_member P_ ((struct sym *, char *, int, int, unsigned));
504 void mark_virtual P_ ((struct sym *));
505 void mark_virtual P_ ((struct sym *));
506 struct sym *make_namespace P_ ((char *));
507 char *sym_scope P_ ((struct sym *));
508 char *sym_scope_1 P_ ((struct sym *));
509 int skip_to P_ ((int));
510 void skip_matching P_ ((void));
511 void member P_ ((struct sym *, int));
512 void class_body P_ ((struct sym *, int));
513 void class_definition P_ ((struct sym *, int, int, int));
514 void declaration P_ ((int));
515 unsigned parm_list P_ ((int *));
516 char *operator_name P_ ((int *));
517 struct sym *parse_classname P_ ((void));
518 struct sym *parse_qualified_ident_or_type P_ ((char **));
519 void parse_qualified_param_ident_or_type P_ ((char **));
520 int globals P_ ((int));
521 void yyerror P_ ((char *, char *));
522
523
524 \f
525 /***********************************************************************
526 Utilities
527 ***********************************************************************/
528
529 /* Print an error in a printf-like style with the current input file
530 name and line number. */
531
532 void
533 yyerror (format, s)
534 char *format, *s;
535 {
536 fprintf (stderr, "%s:%d: ", filename, yyline);
537 fprintf (stderr, format, s);
538 putc ('\n', stderr);
539 }
540
541
542 /* Like malloc but print an error and exit if not enough memory is
543 available. */
544
545 void *
546 xmalloc (nbytes)
547 int nbytes;
548 {
549 void *p = malloc (nbytes);
550 if (p == NULL)
551 {
552 yyerror ("out of memory", NULL);
553 exit (1);
554 }
555 return p;
556 }
557
558
559 /* Like realloc but print an error and exit if out of memory. */
560
561 void *
562 xrealloc (p, sz)
563 void *p;
564 int sz;
565 {
566 p = realloc (p, sz);
567 if (p == NULL)
568 {
569 yyerror ("out of memory", NULL);
570 exit (1);
571 }
572 return p;
573 }
574
575
576 /* Like free but always check for null pointers.. */
577
578 void
579 xfree (p)
580 void *p;
581 {
582 if (p)
583 free (p);
584 }
585
586
587 /* Like strdup, but print an error and exit if not enough memory is
588 available.. If S is null, return null. */
589
590 char *
591 xstrdup (s)
592 char *s;
593 {
594 if (s)
595 s = strcpy (xmalloc (strlen (s) + 1), s);
596 return s;
597 }
598
599
600 \f
601 /***********************************************************************
602 Symbols
603 ***********************************************************************/
604
605 /* Initialize the symbol table. This currently only sets up the
606 special symbol for globals (`*Globals*'). */
607
608 void
609 init_sym ()
610 {
611 global_symbols = add_sym (GLOBALS_NAME, NULL);
612 }
613
614
615 /* Add a symbol for class NAME to the symbol table. NESTED_IN_CLASS
616 is the class in which class NAME was found. If it is null,
617 this means the scope of NAME is the current namespace.
618
619 If a symbol for NAME already exists, return that. Otherwise
620 create a new symbol and set it to default values. */
621
622 struct sym *
623 add_sym (name, nested_in_class)
624 char *name;
625 struct sym *nested_in_class;
626 {
627 struct sym *sym;
628 unsigned h;
629 char *s;
630 struct sym *scope = nested_in_class ? nested_in_class : current_namespace;
631
632 for (s = name, h = 0; *s; ++s)
633 h = (h << 1) ^ *s;
634 h %= TABLE_SIZE;
635
636 for (sym = class_table[h]; sym; sym = sym->next)
637 if (streq (name, sym->name) && sym->namesp == scope)
638 break;
639
640 if (sym == NULL)
641 {
642 if (f_very_verbose)
643 {
644 putchar ('\t');
645 puts (name);
646 }
647
648 sym = (struct sym *) xmalloc (sizeof *sym + strlen (name));
649 bzero (sym, sizeof *sym);
650 strcpy (sym->name, name);
651 sym->namesp = scope;
652 sym->next = class_table[h];
653 class_table[h] = sym;
654 }
655
656 return sym;
657 }
658
659
660 /* Add links between superclass SUPER and subclass SUB. */
661
662 void
663 add_link (super, sub)
664 struct sym *super, *sub;
665 {
666 struct link *lnk, *lnk2, *p, *prev;
667
668 /* See if a link already exists. */
669 for (p = super->subs, prev = NULL;
670 p && strcmp (sub->name, p->sym->name) > 0;
671 prev = p, p = p->next)
672 ;
673
674 /* Avoid duplicates. */
675 if (p == NULL || p->sym != sub)
676 {
677 lnk = (struct link *) xmalloc (sizeof *lnk);
678 lnk2 = (struct link *) xmalloc (sizeof *lnk2);
679
680 lnk->sym = sub;
681 lnk->next = p;
682
683 if (prev)
684 prev->next = lnk;
685 else
686 super->subs = lnk;
687
688 lnk2->sym = super;
689 lnk2->next = sub->supers;
690 sub->supers = lnk2;
691 }
692 }
693
694
695 /* Find in class CLS member NAME.
696
697 VAR non-zero means look for a member variable; otherwise a function
698 is searched. SC specifies what kind of member is searched---a
699 static, or per-instance member etc. HASH is a hash code for the
700 parameter types of functions. Value is a pointer to the member
701 found or null if not found. */
702
703 struct member *
704 find_member (cls, name, var, sc, hash)
705 struct sym *cls;
706 char *name;
707 int var, sc;
708 unsigned hash;
709 {
710 struct member **list;
711 struct member *p;
712 unsigned name_hash = 0;
713 char *s;
714 int i;
715
716 switch (sc)
717 {
718 case SC_FRIEND:
719 list = &cls->friends;
720 break;
721
722 case SC_TYPE:
723 list = &cls->types;
724 break;
725
726 case SC_STATIC:
727 list = var ? &cls->static_vars : &cls->static_fns;
728 break;
729
730 default:
731 list = var ? &cls->vars : &cls->fns;
732 break;
733 }
734
735 for (s = name; *s; ++s)
736 name_hash = (name_hash << 1) ^ *s;
737 i = name_hash % TABLE_SIZE;
738
739 for (p = member_table[i]; p; p = p->anext)
740 if (p->list == list && p->param_hash == hash && streq (name, p->name))
741 break;
742
743 return p;
744 }
745
746
747 /* Add to class CLS information for the declaration of member NAME.
748 REGEXP is a regexp matching the declaration, if non-null. POS is
749 the position in the source where the declaration is found. HASH is
750 a hash code for the parameter list of the member, if it's a
751 function. VAR non-zero means member is a variable or type. SC
752 specifies the type of member (instance member, static, ...). VIS
753 is the member's visibility (public, protected, private). FLAGS is
754 a bit set giving additional information about the member (see the
755 F_* defines). */
756
757 void
758 add_member_decl (cls, name, regexp, pos, hash, var, sc, vis, flags)
759 struct sym *cls;
760 char *name;
761 char *regexp;
762 int pos;
763 unsigned hash;
764 int var;
765 int sc;
766 int vis;
767 int flags;
768 {
769 struct member *m;
770
771 m = find_member (cls, name, var, sc, hash);
772 if (m == NULL)
773 m = add_member (cls, name, var, sc, hash);
774
775 /* Have we seen a new filename? If so record that. */
776 if (!cls->filename || !FILENAME_EQ (cls->filename, filename))
777 m->filename = filename;
778
779 m->regexp = regexp;
780 m->pos = pos;
781 m->flags = flags;
782
783 switch (vis)
784 {
785 case PRIVATE:
786 m->vis = V_PRIVATE;
787 break;
788
789 case PROTECTED:
790 m->vis = V_PROTECTED;
791 break;
792
793 case PUBLIC:
794 m->vis = V_PUBLIC;
795 break;
796 }
797
798 info_where = P_DECL;
799 info_cls = cls;
800 info_member = m;
801 }
802
803
804 /* Add to class CLS information for the definition of member NAME.
805 REGEXP is a regexp matching the declaration, if non-null. POS is
806 the position in the source where the declaration is found. HASH is
807 a hash code for the parameter list of the member, if it's a
808 function. VAR non-zero means member is a variable or type. SC
809 specifies the type of member (instance member, static, ...). VIS
810 is the member's visibility (public, protected, private). FLAGS is
811 a bit set giving additional information about the member (see the
812 F_* defines). */
813
814 void
815 add_member_defn (cls, name, regexp, pos, hash, var, sc, flags)
816 struct sym *cls;
817 char *name;
818 char *regexp;
819 int pos;
820 unsigned hash;
821 int var;
822 int sc;
823 int flags;
824 {
825 struct member *m;
826
827 if (sc == SC_UNKNOWN)
828 {
829 m = find_member (cls, name, var, SC_MEMBER, hash);
830 if (m == NULL)
831 {
832 m = find_member (cls, name, var, SC_STATIC, hash);
833 if (m == NULL)
834 m = add_member (cls, name, var, sc, hash);
835 }
836 }
837 else
838 {
839 m = find_member (cls, name, var, sc, hash);
840 if (m == NULL)
841 m = add_member (cls, name, var, sc, hash);
842 }
843
844 if (!cls->sfilename)
845 cls->sfilename = filename;
846
847 if (!FILENAME_EQ (cls->sfilename, filename))
848 m->def_filename = filename;
849
850 m->def_regexp = regexp;
851 m->def_pos = pos;
852 m->flags |= flags;
853
854 info_where = P_DEFN;
855 info_cls = cls;
856 info_member = m;
857 }
858
859
860 /* Add a symbol for a define named NAME to the symbol table.
861 REGEXP is a regular expression matching the define in the source,
862 if it is non-null. POS is the position in the file. */
863
864 void
865 add_define (name, regexp, pos)
866 char *name, *regexp;
867 int pos;
868 {
869 add_global_defn (name, regexp, pos, 0, 1, SC_FRIEND, F_DEFINE);
870 add_global_decl (name, regexp, pos, 0, 1, SC_FRIEND, F_DEFINE);
871 }
872
873
874 /* Add information for the global definition of NAME.
875 REGEXP is a regexp matching the declaration, if non-null. POS is
876 the position in the source where the declaration is found. HASH is
877 a hash code for the parameter list of the member, if it's a
878 function. VAR non-zero means member is a variable or type. SC
879 specifies the type of member (instance member, static, ...). VIS
880 is the member's visibility (public, protected, private). FLAGS is
881 a bit set giving additional information about the member (see the
882 F_* defines). */
883
884 void
885 add_global_defn (name, regexp, pos, hash, var, sc, flags)
886 char *name, *regexp;
887 int pos;
888 unsigned hash;
889 int var;
890 int sc;
891 int flags;
892 {
893 int i;
894 struct sym *sym;
895
896 /* Try to find out for which classes a function is a friend, and add
897 what we know about it to them. */
898 if (!var)
899 for (i = 0; i < TABLE_SIZE; ++i)
900 for (sym = class_table[i]; sym; sym = sym->next)
901 if (sym != global_symbols && sym->friends)
902 if (find_member (sym, name, 0, SC_FRIEND, hash))
903 add_member_defn (sym, name, regexp, pos, hash, 0,
904 SC_FRIEND, flags);
905
906 /* Add to global symbols. */
907 add_member_defn (global_symbols, name, regexp, pos, hash, var, sc, flags);
908 }
909
910
911 /* Add information for the global declaration of NAME.
912 REGEXP is a regexp matching the declaration, if non-null. POS is
913 the position in the source where the declaration is found. HASH is
914 a hash code for the parameter list of the member, if it's a
915 function. VAR non-zero means member is a variable or type. SC
916 specifies the type of member (instance member, static, ...). VIS
917 is the member's visibility (public, protected, private). FLAGS is
918 a bit set giving additional information about the member (see the
919 F_* defines). */
920
921 void
922 add_global_decl (name, regexp, pos, hash, var, sc, flags)
923 char *name, *regexp;
924 int pos;
925 unsigned hash;
926 int var;
927 int sc;
928 int flags;
929 {
930 /* Add declaration only if not already declared. Header files must
931 be processed before source files for this to have the right effect.
932 I do not want to handle implicit declarations at the moment. */
933 struct member *m;
934 struct member *found;
935
936 m = found = find_member (global_symbols, name, var, sc, hash);
937 if (m == NULL)
938 m = add_member (global_symbols, name, var, sc, hash);
939
940 /* Definition already seen => probably last declaration implicit.
941 Override. This means that declarations must always be added to
942 the symbol table before definitions. */
943 if (!found)
944 {
945 if (!global_symbols->filename
946 || !FILENAME_EQ (global_symbols->filename, filename))
947 m->filename = filename;
948
949 m->regexp = regexp;
950 m->pos = pos;
951 m->vis = V_PUBLIC;
952 m->flags = flags;
953
954 info_where = P_DECL;
955 info_cls = global_symbols;
956 info_member = m;
957 }
958 }
959
960
961 /* Add a symbol for member NAME to class CLS.
962 VAR non-zero means it's a variable. SC specifies the kind of
963 member. HASH is a hash code for the parameter types of a function.
964 Value is a pointer to the member's structure. */
965
966 struct member *
967 add_member (cls, name, var, sc, hash)
968 struct sym *cls;
969 char *name;
970 int var;
971 int sc;
972 unsigned hash;
973 {
974 struct member *m = (struct member *) xmalloc (sizeof *m + strlen (name));
975 struct member **list;
976 struct member *p;
977 struct member *prev;
978 unsigned name_hash = 0;
979 int i;
980 char *s;
981
982 strcpy (m->name, name);
983 m->param_hash = hash;
984
985 m->vis = 0;
986 m->flags = 0;
987 m->regexp = NULL;
988 m->filename = NULL;
989 m->pos = 0;
990 m->def_regexp = NULL;
991 m->def_filename = NULL;
992 m->def_pos = 0;
993
994 assert (cls != NULL);
995
996 switch (sc)
997 {
998 case SC_FRIEND:
999 list = &cls->friends;
1000 break;
1001
1002 case SC_TYPE:
1003 list = &cls->types;
1004 break;
1005
1006 case SC_STATIC:
1007 list = var ? &cls->static_vars : &cls->static_fns;
1008 break;
1009
1010 default:
1011 list = var ? &cls->vars : &cls->fns;
1012 break;
1013 }
1014
1015 for (s = name; *s; ++s)
1016 name_hash = (name_hash << 1) ^ *s;
1017 i = name_hash % TABLE_SIZE;
1018 m->anext = member_table[i];
1019 member_table[i] = m;
1020 m->list = list;
1021
1022 /* Keep the member list sorted. It's cheaper to do it here than to
1023 sort them in Lisp. */
1024 for (prev = NULL, p = *list;
1025 p && strcmp (name, p->name) > 0;
1026 prev = p, p = p->next)
1027 ;
1028
1029 m->next = p;
1030 if (prev)
1031 prev->next = m;
1032 else
1033 *list = m;
1034 return m;
1035 }
1036
1037
1038 /* Given the root R of a class tree, step through all subclasses
1039 recursively, marking functions as virtual that are declared virtual
1040 in base classes. */
1041
1042 void
1043 mark_virtual (r)
1044 struct sym *r;
1045 {
1046 struct link *p;
1047 struct member *m, *m2;
1048
1049 for (p = r->subs; p; p = p->next)
1050 {
1051 for (m = r->fns; m; m = m->next)
1052 if (HAS_FLAG (m->flags, F_VIRTUAL))
1053 {
1054 for (m2 = p->sym->fns; m2; m2 = m2->next)
1055 if (m->param_hash == m2->param_hash && streq (m->name, m2->name))
1056 SET_FLAG (m2->flags, F_VIRTUAL);
1057 }
1058
1059 mark_virtual (p->sym);
1060 }
1061 }
1062
1063
1064 /* For all roots of the class tree, mark functions as virtual that
1065 are virtual because of a virtual declaration in a base class. */
1066
1067 void
1068 mark_inherited_virtual ()
1069 {
1070 struct sym *r;
1071 int i;
1072
1073 for (i = 0; i < TABLE_SIZE; ++i)
1074 for (r = class_table[i]; r; r = r->next)
1075 if (r->supers == NULL)
1076 mark_virtual (r);
1077 }
1078
1079
1080 /* Create and return a symbol for a namespace with name NAME. */
1081
1082 struct sym *
1083 make_namespace (name)
1084 char *name;
1085 {
1086 struct sym *s = (struct sym *) xmalloc (sizeof *s + strlen (name));
1087 bzero (s, sizeof *s);
1088 strcpy (s->name, name);
1089 s->next = all_namespaces;
1090 s->namesp = current_namespace;
1091 all_namespaces = s;
1092 return s;
1093 }
1094
1095
1096 /* Find the symbol for namespace NAME. If not found, add a new symbol
1097 for NAME to all_namespaces. */
1098
1099 struct sym *
1100 find_namespace (name)
1101 char *name;
1102 {
1103 struct sym *p;
1104
1105 for (p = all_namespaces; p; p = p->next)
1106 {
1107 if (streq (p->name, name))
1108 break;
1109 else
1110 {
1111 struct alias *p2;
1112 for (p2 = p->namesp_aliases; p2; p2 = p2->next)
1113 if (streq (p2->name, name))
1114 break;
1115 if (p2)
1116 break;
1117 }
1118 }
1119
1120 if (p == NULL)
1121 p = make_namespace (name);
1122
1123 return p;
1124 }
1125
1126
1127 /* Register the name NEW_NAME as an alias for namespace OLD_NAME. */
1128
1129 void
1130 register_namespace_alias (new_name, old_name)
1131 char *new_name, *old_name;
1132 {
1133 struct sym *p = find_namespace (old_name);
1134 struct alias *al;
1135
1136 /* Is it already in the list of aliases? */
1137 for (al = p->namesp_aliases; al; al = al->next)
1138 if (streq (new_name, p->name))
1139 return;
1140
1141 al = (struct alias *) xmalloc (sizeof *al + strlen (new_name));
1142 strcpy (al->name, new_name);
1143 al->next = p->namesp_aliases;
1144 p->namesp_aliases = al;
1145 }
1146
1147
1148 /* Enter namespace with name NAME. */
1149
1150 void
1151 enter_namespace (name)
1152 char *name;
1153 {
1154 struct sym *p = find_namespace (name);
1155
1156 if (namespace_sp == namespace_stack_size)
1157 {
1158 int size = max (10, 2 * namespace_stack_size);
1159 namespace_stack
1160 = (struct sym **) xrealloc ((void *)namespace_stack,
1161 size * sizeof *namespace_stack);
1162 namespace_stack_size = size;
1163 }
1164
1165 namespace_stack[namespace_sp++] = current_namespace;
1166 current_namespace = p;
1167 }
1168
1169
1170 /* Leave the current namespace. */
1171
1172 void
1173 leave_namespace ()
1174 {
1175 assert (namespace_sp > 0);
1176 current_namespace = namespace_stack[--namespace_sp];
1177 }
1178
1179
1180 \f
1181 /***********************************************************************
1182 Writing the Output File
1183 ***********************************************************************/
1184
1185 /* Write string S to the output file FP in a Lisp-readable form.
1186 If S is null, write out `()'. */
1187
1188 #define PUTSTR(s, fp) \
1189 do { \
1190 if (!s) \
1191 { \
1192 putc ('(', fp); \
1193 putc (')', fp); \
1194 putc (' ', fp); \
1195 } \
1196 else \
1197 { \
1198 putc ('"', fp); \
1199 fputs (s, fp); \
1200 putc ('"', fp); \
1201 putc (' ', fp); \
1202 } \
1203 } while (0)
1204
1205 /* A dynamically allocated buffer for constructing a scope name. */
1206
1207 char *scope_buffer;
1208 int scope_buffer_size;
1209 int scope_buffer_len;
1210
1211
1212 /* Make sure scope_buffer has enough room to add LEN chars to it. */
1213
1214 void
1215 ensure_scope_buffer_room (len)
1216 int len;
1217 {
1218 if (scope_buffer_len + len >= scope_buffer_size)
1219 {
1220 int new_size = max (2 * scope_buffer_size, scope_buffer_len + len);
1221 scope_buffer = (char *) xrealloc (scope_buffer, new_size);
1222 scope_buffer_size = new_size;
1223 }
1224 }
1225
1226
1227 /* Recursively add the scope names of symbol P and the scopes of its
1228 namespaces to scope_buffer. Value is a pointer to the complete
1229 scope name constructed. */
1230
1231 char *
1232 sym_scope_1 (p)
1233 struct sym *p;
1234 {
1235 int len;
1236
1237 if (p->namesp)
1238 sym_scope_1 (p->namesp);
1239
1240 if (*scope_buffer)
1241 {
1242 ensure_scope_buffer_room (3);
1243 strcat (scope_buffer, "::");
1244 scope_buffer_len += 2;
1245 }
1246
1247 len = strlen (p->name);
1248 ensure_scope_buffer_room (len + 1);
1249 strcat (scope_buffer, p->name);
1250 scope_buffer_len += len;
1251
1252 if (HAS_FLAG (p->flags, F_TEMPLATE))
1253 {
1254 ensure_scope_buffer_room (3);
1255 strcat (scope_buffer, "<>");
1256 scope_buffer_len += 2;
1257 }
1258
1259 return scope_buffer;
1260 }
1261
1262
1263 /* Return the scope of symbol P in printed representation, i.e.
1264 as it would appear in a C*+ source file. */
1265
1266 char *
1267 sym_scope (p)
1268 struct sym *p;
1269 {
1270 if (!scope_buffer)
1271 {
1272 scope_buffer_size = 1024;
1273 scope_buffer = (char *) xmalloc (scope_buffer_size);
1274 }
1275
1276 *scope_buffer = '\0';
1277 scope_buffer_len = 0;
1278
1279 if (p->namesp)
1280 sym_scope_1 (p->namesp);
1281
1282 return scope_buffer;
1283 }
1284
1285
1286 /* Dump the list of members M to file FP. Value is the length of the
1287 list. */
1288
1289 int
1290 dump_members (fp, m)
1291 FILE *fp;
1292 struct member *m;
1293 {
1294 int n;
1295
1296 putc ('(', fp);
1297
1298 for (n = 0; m; m = m->next, ++n)
1299 {
1300 fputs (MEMBER_STRUCT, fp);
1301 PUTSTR (m->name, fp);
1302 PUTSTR (NULL, fp); /* FIXME? scope for globals */
1303 fprintf (fp, "%u ", (unsigned) m->flags);
1304 PUTSTR (m->filename, fp);
1305 PUTSTR (m->regexp, fp);
1306 fprintf (fp, "%u ", (unsigned) m->pos);
1307 fprintf (fp, "%u ", (unsigned) m->vis);
1308 putc (' ', fp);
1309 PUTSTR (m->def_filename, fp);
1310 PUTSTR (m->def_regexp, fp);
1311 fprintf (fp, "%u", (unsigned) m->def_pos);
1312 putc (']', fp);
1313 putc ('\n', fp);
1314 }
1315
1316 putc (')', fp);
1317 putc ('\n', fp);
1318 return n;
1319 }
1320
1321
1322 /* Dump class ROOT to stream FP. */
1323
1324 void
1325 dump_sym (fp, root)
1326 FILE *fp;
1327 struct sym *root;
1328 {
1329 fputs (CLASS_STRUCT, fp);
1330 PUTSTR (root->name, fp);
1331
1332 /* Print scope, if any. */
1333 if (root->namesp)
1334 PUTSTR (sym_scope (root), fp);
1335 else
1336 PUTSTR (NULL, fp);
1337
1338 /* Print flags. */
1339 fprintf (fp, "%u", root->flags);
1340 PUTSTR (root->filename, fp);
1341 PUTSTR (root->regexp, fp);
1342 fprintf (fp, "%u", (unsigned) root->pos);
1343 PUTSTR (root->sfilename, fp);
1344 putc (']', fp);
1345 putc ('\n', fp);
1346 }
1347
1348
1349 /* Dump class ROOT and its subclasses to file FP. Value is the
1350 number of classes written. */
1351
1352 int
1353 dump_tree (fp, root)
1354 FILE *fp;
1355 struct sym *root;
1356 {
1357 struct link *lk;
1358 unsigned n = 0;
1359
1360 dump_sym (fp, root);
1361
1362 if (f_verbose)
1363 {
1364 putchar ('+');
1365 fflush (stdout);
1366 }
1367
1368 putc ('(', fp);
1369
1370 for (lk = root->subs; lk; lk = lk->next)
1371 {
1372 fputs (TREE_STRUCT, fp);
1373 n += dump_tree (fp, lk->sym);
1374 putc (']', fp);
1375 }
1376
1377 putc (')', fp);
1378
1379 dump_members (fp, root->vars);
1380 n += dump_members (fp, root->fns);
1381 dump_members (fp, root->static_vars);
1382 n += dump_members (fp, root->static_fns);
1383 n += dump_members (fp, root->friends);
1384 dump_members (fp, root->types);
1385
1386 /* Superclasses. */
1387 putc ('(', fp);
1388 putc (')', fp);
1389
1390 /* Mark slot. */
1391 putc ('(', fp);
1392 putc (')', fp);
1393
1394 putc ('\n', fp);
1395 return n;
1396 }
1397
1398
1399 /* Dump the entire class tree to file FP. */
1400
1401 void
1402 dump_roots (fp)
1403 FILE *fp;
1404 {
1405 int i, n = 0;
1406 struct sym *r;
1407
1408 /* Output file header containing version string, command line
1409 options etc. */
1410 if (!f_append)
1411 {
1412 fputs (TREE_HEADER_STRUCT, fp);
1413 PUTSTR (EBROWSE_FILE_VERSION, fp);
1414
1415 putc ('\"', fp);
1416 if (!f_structs)
1417 fputs (" -s", fp);
1418 if (f_regexps)
1419 fputs (" -x", fp);
1420 putc ('\"', fp);
1421 fputs (" ()", fp);
1422 fputs (" ()", fp);
1423 putc (']', fp);
1424 }
1425
1426 /* Mark functions as virtual that are so because of functions
1427 declared virtual in base classes. */
1428 mark_inherited_virtual ();
1429
1430 /* Dump the roots of the graph. */
1431 for (i = 0; i < TABLE_SIZE; ++i)
1432 for (r = class_table[i]; r; r = r->next)
1433 if (!r->supers)
1434 {
1435 fputs (TREE_STRUCT, fp);
1436 n += dump_tree (fp, r);
1437 putc (']', fp);
1438 }
1439
1440 if (f_verbose)
1441 putchar ('\n');
1442 }
1443
1444
1445 \f
1446 /***********************************************************************
1447 Scanner
1448 ***********************************************************************/
1449
1450 #ifdef DEBUG
1451 #define INCREMENT_LINENO \
1452 do { \
1453 if (f_very_verbose) \
1454 { \
1455 ++yyline; \
1456 printf ("%d:\n", yyline); \
1457 } \
1458 else \
1459 ++yyline; \
1460 } while (0)
1461 #else
1462 #define INCREMENT_LINENO ++yyline
1463 #endif
1464
1465 /* Define two macros for accessing the input buffer (current input
1466 file). GET(C) sets C to the next input character and advances the
1467 input pointer. UNGET retracts the input pointer. */
1468
1469 #define GET(C) ((C) = *in++)
1470 #define UNGET() (--in)
1471
1472
1473 /* Process a preprocessor line. Value is the next character from the
1474 input buffer not consumed. */
1475
1476 int
1477 process_pp_line ()
1478 {
1479 int in_comment = 0, in_string = 0;
1480 int c;
1481 char *p = yytext;
1482
1483 /* Skip over white space. The `#' has been consumed already. */
1484 while (WHITEP (GET (c)))
1485 ;
1486
1487 /* Read the preprocessor command (if any). */
1488 while (IDENTP (c))
1489 {
1490 *p++ = c;
1491 GET (c);
1492 }
1493
1494 /* Is it a `define'? */
1495 *p = '\0';
1496
1497 if (*yytext && streq (yytext, "define"))
1498 {
1499 p = yytext;
1500 while (WHITEP (c))
1501 GET (c);
1502 while (IDENTP (c))
1503 {
1504 *p++ = c;
1505 GET (c);
1506 }
1507
1508 *p = '\0';
1509
1510 if (*yytext)
1511 {
1512 char *regexp = matching_regexp ();
1513 int pos = BUFFER_POS ();
1514 add_define (yytext, regexp, pos);
1515 }
1516 }
1517
1518 while (c && (c != '\n' || in_comment || in_string))
1519 {
1520 if (c == '\\')
1521 GET (c);
1522 else if (c == '/' && !in_comment)
1523 {
1524 if (GET (c) == '*')
1525 in_comment = 1;
1526 }
1527 else if (c == '*' && in_comment)
1528 {
1529 if (GET (c) == '/')
1530 in_comment = 0;
1531 }
1532 else if (c == '"')
1533 in_string = !in_string;
1534
1535 if (c == '\n')
1536 INCREMENT_LINENO;
1537
1538 GET (c);
1539 }
1540
1541 return c;
1542 }
1543
1544
1545 /* Value is the next token from the input buffer. */
1546
1547 int
1548 yylex ()
1549 {
1550 int c;
1551 char end_char;
1552 char *p;
1553
1554 for (;;)
1555 {
1556 while (WHITEP (GET (c)))
1557 ;
1558
1559 switch (c)
1560 {
1561 case '\n':
1562 INCREMENT_LINENO;
1563 break;
1564
1565 case '\r':
1566 break;
1567
1568 case 0:
1569 /* End of file. */
1570 return YYEOF;
1571
1572 case '\\':
1573 GET (c);
1574 break;
1575
1576 case '"':
1577 case '\'':
1578 /* String and character constants. */
1579 end_char = c;
1580 string_start = in;
1581 while (GET (c) && c != end_char)
1582 {
1583 switch (c)
1584 {
1585 case '\\':
1586 /* Escape sequences. */
1587 if (!GET (c))
1588 {
1589 if (end_char == '\'')
1590 yyerror ("EOF in character constant", NULL);
1591 else
1592 yyerror ("EOF in string constant", NULL);
1593 goto end_string;
1594 }
1595 else switch (c)
1596 {
1597 case '\n':
1598 INCREMENT_LINENO;
1599 case 'a':
1600 case 'b':
1601 case 'f':
1602 case 'n':
1603 case 'r':
1604 case 't':
1605 case 'v':
1606 break;
1607
1608 case 'x':
1609 {
1610 /* Hexadecimal escape sequence. */
1611 int i;
1612 for (i = 0; i < 2; ++i)
1613 {
1614 GET (c);
1615
1616 if (c >= '0' && c <= '7')
1617 ;
1618 else if (c >= 'a' && c <= 'f')
1619 ;
1620 else if (c >= 'A' && c <= 'F')
1621 ;
1622 else
1623 {
1624 UNGET ();
1625 break;
1626 }
1627 }
1628 }
1629 break;
1630
1631 case '0':
1632 {
1633 /* Octal escape sequence. */
1634 int i;
1635 for (i = 0; i < 3; ++i)
1636 {
1637 GET (c);
1638
1639 if (c >= '0' && c <= '7')
1640 ;
1641 else
1642 {
1643 UNGET ();
1644 break;
1645 }
1646 }
1647 }
1648 break;
1649
1650 default:
1651 break;
1652 }
1653 break;
1654
1655 case '\n':
1656 if (end_char == '\'')
1657 yyerror ("newline in character constant", NULL);
1658 else
1659 yyerror ("newline in string constant", NULL);
1660 INCREMENT_LINENO;
1661 break;
1662
1663 default:
1664 break;
1665 }
1666 }
1667
1668 end_string:
1669 return end_char == '\'' ? CCHAR : CSTRING;
1670
1671 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': case 'g':
1672 case 'h': case 'i': case 'j': case 'k': case 'l': case 'm': case 'n':
1673 case 'o': case 'p': case 'q': case 'r': case 's': case 't': case 'u':
1674 case 'v': case 'w': case 'x': case 'y': case 'z':
1675 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': case 'G':
1676 case 'H': case 'I': case 'J': case 'K': case 'L': case 'M': case 'N':
1677 case 'O': case 'P': case 'Q': case 'R': case 'S': case 'T': case 'U':
1678 case 'V': case 'W': case 'X': case 'Y': case 'Z': case '_':
1679 {
1680 /* Identifier and keywords. */
1681 unsigned hash;
1682 struct kw *k;
1683
1684 p = yytext;
1685 *p++ = hash = c;
1686
1687 while (IDENTP (GET (*p)))
1688 {
1689 hash = (hash << 1) ^ *p++;
1690 if (p == yytext_end - 1)
1691 {
1692 int size = yytext_end - yytext;
1693 yytext = (char *) xrealloc (yytext, 2 * size);
1694 yytext_end = yytext + 2 * size;
1695 p = yytext + size - 1;
1696 }
1697 }
1698
1699 UNGET ();
1700 *p = 0;
1701
1702 for (k = keyword_table[hash % KEYWORD_TABLE_SIZE]; k; k = k->next)
1703 if (streq (k->name, yytext))
1704 return k->tk;
1705
1706 return IDENT;
1707 }
1708
1709 case '/':
1710 /* C and C++ comments, '/' and '/='. */
1711 switch (GET (c))
1712 {
1713 case '*':
1714 while (GET (c))
1715 {
1716 switch (c)
1717 {
1718 case '*':
1719 if (GET (c) == '/')
1720 goto comment_end;
1721 UNGET ();
1722 break;
1723 case '\\':
1724 GET (c);
1725 break;
1726 case '\n':
1727 INCREMENT_LINENO;
1728 break;
1729 }
1730 }
1731 comment_end:;
1732 break;
1733
1734 case '=':
1735 return DIVASGN;
1736
1737 case '/':
1738 while (GET (c) && c != '\n')
1739 ;
1740 INCREMENT_LINENO;
1741 break;
1742
1743 default:
1744 UNGET ();
1745 return '/';
1746 }
1747 break;
1748
1749 case '+':
1750 if (GET (c) == '+')
1751 return INC;
1752 else if (c == '=')
1753 return ADDASGN;
1754 UNGET ();
1755 return '+';
1756
1757 case '-':
1758 switch (GET (c))
1759 {
1760 case '-':
1761 return DEC;
1762 case '>':
1763 if (GET (c) == '*')
1764 return ARROWSTAR;
1765 UNGET ();
1766 return ARROW;
1767 case '=':
1768 return SUBASGN;
1769 }
1770 UNGET ();
1771 return '-';
1772
1773 case '*':
1774 if (GET (c) == '=')
1775 return MULASGN;
1776 UNGET ();
1777 return '*';
1778
1779 case '%':
1780 if (GET (c) == '=')
1781 return MODASGN;
1782 UNGET ();
1783 return '%';
1784
1785 case '|':
1786 if (GET (c) == '|')
1787 return LOR;
1788 else if (c == '=')
1789 return ORASGN;
1790 UNGET ();
1791 return '|';
1792
1793 case '&':
1794 if (GET (c) == '&')
1795 return LAND;
1796 else if (c == '=')
1797 return ANDASGN;
1798 UNGET ();
1799 return '&';
1800
1801 case '^':
1802 if (GET (c) == '=')
1803 return XORASGN;
1804 UNGET ();
1805 return '^';
1806
1807 case '.':
1808 if (GET (c) == '*')
1809 return POINTSTAR;
1810 else if (c == '.')
1811 {
1812 if (GET (c) != '.')
1813 yyerror ("invalid token '..' ('...' assumed)", NULL);
1814 UNGET ();
1815 return ELLIPSIS;
1816 }
1817 else if (!DIGITP (c))
1818 {
1819 UNGET ();
1820 return '.';
1821 }
1822 goto mantissa;
1823
1824 case ':':
1825 if (GET (c) == ':')
1826 return DCOLON;
1827 UNGET ();
1828 return ':';
1829
1830 case '=':
1831 if (GET (c) == '=')
1832 return EQ;
1833 UNGET ();
1834 return '=';
1835
1836 case '!':
1837 if (GET (c) == '=')
1838 return NE;
1839 UNGET ();
1840 return '!';
1841
1842 case '<':
1843 switch (GET (c))
1844 {
1845 case '=':
1846 return LE;
1847 case '<':
1848 if (GET (c) == '=')
1849 return LSHIFTASGN;
1850 UNGET ();
1851 return LSHIFT;
1852 }
1853 UNGET ();
1854 return '<';
1855
1856 case '>':
1857 switch (GET (c))
1858 {
1859 case '=':
1860 return GE;
1861 case '>':
1862 if (GET (c) == '=')
1863 return RSHIFTASGN;
1864 UNGET ();
1865 return RSHIFT;
1866 }
1867 UNGET ();
1868 return '>';
1869
1870 case '#':
1871 c = process_pp_line ();
1872 if (c == 0)
1873 return YYEOF;
1874 break;
1875
1876 case '(': case ')': case '[': case ']': case '{': case '}':
1877 case ';': case ',': case '?': case '~':
1878 return c;
1879
1880 case '0':
1881 yyival = 0;
1882
1883 if (GET (c) == 'x' || c == 'X')
1884 {
1885 while (GET (c))
1886 {
1887 if (DIGITP (c))
1888 yyival = yyival * 16 + c - '0';
1889 else if (c >= 'a' && c <= 'f')
1890 yyival = yyival * 16 + c - 'a' + 10;
1891 else if (c >= 'A' && c <= 'F')
1892 yyival = yyival * 16 + c - 'A' + 10;
1893 else
1894 break;
1895 }
1896
1897 goto int_suffixes;
1898 }
1899 else if (c == '.')
1900 goto mantissa;
1901
1902 while (c >= '0' && c <= '7')
1903 {
1904 yyival = (yyival << 3) + c - '0';
1905 GET (c);
1906 }
1907
1908 int_suffixes:
1909 /* Integer suffixes. */
1910 while (isalpha (c))
1911 GET (c);
1912 UNGET ();
1913 return CINT;
1914
1915 case '1': case '2': case '3': case '4': case '5': case '6':
1916 case '7': case '8': case '9':
1917 /* Integer or floating constant, part before '.'. */
1918 yyival = c - '0';
1919
1920 while (GET (c) && DIGITP (c))
1921 yyival = 10 * yyival + c - '0';
1922
1923 if (c != '.')
1924 goto int_suffixes;
1925
1926 mantissa:
1927 /* Digits following '.'. */
1928 while (DIGITP (c))
1929 GET (c);
1930
1931 /* Optional exponent. */
1932 if (c == 'E' || c == 'e')
1933 {
1934 if (GET (c) == '-' || c == '+')
1935 GET (c);
1936
1937 while (DIGITP (c))
1938 GET (c);
1939 }
1940
1941 /* Optional type suffixes. */
1942 while (isalpha (c))
1943 GET (c);
1944 UNGET ();
1945 return CFLOAT;
1946
1947 default:
1948 break;
1949 }
1950 }
1951 }
1952
1953
1954 /* Actually local to matching_regexp. These variables must be in
1955 global scope for the case that `static' get's defined away. */
1956
1957 static char *matching_regexp_buffer, *matching_regexp_end_buf;
1958
1959
1960 /* Value is the string from the start of the line to the current
1961 position in the input buffer, or maybe a bit more if that string is
1962 shorter than min_regexp. */
1963
1964 char *
1965 matching_regexp ()
1966 {
1967 char *p;
1968 char *s;
1969 char *t;
1970
1971 if (!f_regexps)
1972 return NULL;
1973
1974 if (matching_regexp_buffer == NULL)
1975 {
1976 matching_regexp_buffer = (char *) xmalloc (max_regexp);
1977 matching_regexp_end_buf = &matching_regexp_buffer[max_regexp] - 1;
1978 }
1979
1980 /* Scan back to previous newline of buffer start. */
1981 for (p = in - 1; p > inbuffer && *p != '\n'; --p)
1982 ;
1983
1984 if (*p == '\n')
1985 {
1986 while (in - p < min_regexp && p > inbuffer)
1987 {
1988 /* Line probably not significant enough */
1989 for (--p; p >= inbuffer && *p != '\n'; --p)
1990 ;
1991 }
1992 if (*p == '\n')
1993 ++p;
1994 }
1995
1996 /* Copy from end to make sure significant portions are included.
1997 This implies that in the browser a regular expressing of the form
1998 `^.*{regexp}' has to be used. */
1999 for (s = matching_regexp_end_buf - 1, t = in;
2000 s > matching_regexp_buffer && t > p;)
2001 {
2002 *--s = *--t;
2003
2004 if (*s == '"')
2005 *--s = '\\';
2006 }
2007
2008 *(matching_regexp_end_buf - 1) = '\0';
2009 return xstrdup (s);
2010 }
2011
2012
2013 /* Return a printable representation of token T. */
2014
2015 char *
2016 token_string (t)
2017 int t;
2018 {
2019 static char b[3];
2020
2021 switch (t)
2022 {
2023 case CSTRING: return "string constant";
2024 case CCHAR: return "char constant";
2025 case CINT: return "int constant";
2026 case CFLOAT: return "floating constant";
2027 case ELLIPSIS: return "...";
2028 case LSHIFTASGN: return "<<=";
2029 case RSHIFTASGN: return ">>=";
2030 case ARROWSTAR: return "->*";
2031 case IDENT: return "identifier";
2032 case DIVASGN: return "/=";
2033 case INC: return "++";
2034 case ADDASGN: return "+=";
2035 case DEC: return "--";
2036 case ARROW: return "->";
2037 case SUBASGN: return "-=";
2038 case MULASGN: return "*=";
2039 case MODASGN: return "%=";
2040 case LOR: return "||";
2041 case ORASGN: return "|=";
2042 case LAND: return "&&";
2043 case ANDASGN: return "&=";
2044 case XORASGN: return "^=";
2045 case POINTSTAR: return ".*";
2046 case DCOLON: return "::";
2047 case EQ: return "==";
2048 case NE: return "!=";
2049 case LE: return "<=";
2050 case LSHIFT: return "<<";
2051 case GE: return ">=";
2052 case RSHIFT: return ">>";
2053 case ASM: return "asm";
2054 case AUTO: return "auto";
2055 case BREAK: return "break";
2056 case CASE: return "case";
2057 case CATCH: return "catch";
2058 case CHAR: return "char";
2059 case CLASS: return "class";
2060 case CONST: return "const";
2061 case CONTINUE: return "continue";
2062 case DEFAULT: return "default";
2063 case DELETE: return "delete";
2064 case DO: return "do";
2065 case DOUBLE: return "double";
2066 case ELSE: return "else";
2067 case ENUM: return "enum";
2068 case EXTERN: return "extern";
2069 case FLOAT: return "float";
2070 case FOR: return "for";
2071 case FRIEND: return "friend";
2072 case GOTO: return "goto";
2073 case IF: return "if";
2074 case T_INLINE: return "inline";
2075 case INT: return "int";
2076 case LONG: return "long";
2077 case NEW: return "new";
2078 case OPERATOR: return "operator";
2079 case PRIVATE: return "private";
2080 case PROTECTED: return "protected";
2081 case PUBLIC: return "public";
2082 case REGISTER: return "register";
2083 case RETURN: return "return";
2084 case SHORT: return "short";
2085 case SIGNED: return "signed";
2086 case SIZEOF: return "sizeof";
2087 case STATIC: return "static";
2088 case STRUCT: return "struct";
2089 case SWITCH: return "switch";
2090 case TEMPLATE: return "template";
2091 case THIS: return "this";
2092 case THROW: return "throw";
2093 case TRY: return "try";
2094 case TYPEDEF: return "typedef";
2095 case UNION: return "union";
2096 case UNSIGNED: return "unsigned";
2097 case VIRTUAL: return "virtual";
2098 case VOID: return "void";
2099 case VOLATILE: return "volatile";
2100 case WHILE: return "while";
2101 case MUTABLE: return "mutable";
2102 case BOOL: return "bool";
2103 case TRUE: return "true";
2104 case FALSE: return "false";
2105 case SIGNATURE: return "signature";
2106 case NAMESPACE: return "namespace";
2107 case EXPLICIT: return "explicit";
2108 case TYPENAME: return "typename";
2109 case CONST_CAST: return "const_cast";
2110 case DYNAMIC_CAST: return "dynamic_cast";
2111 case REINTERPRET_CAST: return "reinterpret_cast";
2112 case STATIC_CAST: return "static_cast";
2113 case TYPEID: return "typeid";
2114 case USING: return "using";
2115 case WCHAR: return "wchar_t";
2116 case YYEOF: return "EOF";
2117
2118 default:
2119 if (t < 255)
2120 {
2121 b[0] = t;
2122 b[1] = '\0';
2123 return b;
2124 }
2125 else
2126 return "???";
2127 }
2128 }
2129
2130
2131 /* Reinitialize the scanner for a new input file. */
2132
2133 void
2134 re_init_scanner ()
2135 {
2136 in = inbuffer;
2137 yyline = 1;
2138
2139 if (yytext == NULL)
2140 {
2141 int size = 256;
2142 yytext = (char *) xmalloc (size * sizeof *yytext);
2143 yytext_end = yytext + size;
2144 }
2145 }
2146
2147
2148 /* Insert a keyword NAME with token value TK into the keyword hash
2149 table. */
2150
2151 void
2152 insert_keyword (name, tk)
2153 char *name;
2154 int tk;
2155 {
2156 char *s;
2157 unsigned h = 0;
2158 struct kw *k = (struct kw *) xmalloc (sizeof *k);
2159
2160 for (s = name; *s; ++s)
2161 h = (h << 1) ^ *s;
2162
2163 h %= KEYWORD_TABLE_SIZE;
2164 k->name = name;
2165 k->tk = tk;
2166 k->next = keyword_table[h];
2167 keyword_table[h] = k;
2168 }
2169
2170
2171 /* Initialize the scanner for the first file. This sets up the
2172 character class vectors and fills the keyword hash table. */
2173
2174 void
2175 init_scanner ()
2176 {
2177 int i;
2178
2179 /* Allocate the input buffer */
2180 inbuffer_size = READ_CHUNK_SIZE + 1;
2181 inbuffer = in = (char *) xmalloc (inbuffer_size);
2182 yyline = 1;
2183
2184 /* Set up character class vectors. */
2185 for (i = 0; i < sizeof is_ident; ++i)
2186 {
2187 if (i == '_' || isalnum (i))
2188 is_ident[i] = 1;
2189
2190 if (i >= '0' && i <= '9')
2191 is_digit[i] = 1;
2192
2193 if (i == ' ' || i == '\t' || i == '\f' || i == '\v')
2194 is_white[i] = 1;
2195 }
2196
2197 /* Fill keyword hash table. */
2198 insert_keyword ("and", LAND);
2199 insert_keyword ("and_eq", ANDASGN);
2200 insert_keyword ("asm", ASM);
2201 insert_keyword ("auto", AUTO);
2202 insert_keyword ("bitand", '&');
2203 insert_keyword ("bitor", '|');
2204 insert_keyword ("bool", BOOL);
2205 insert_keyword ("break", BREAK);
2206 insert_keyword ("case", CASE);
2207 insert_keyword ("catch", CATCH);
2208 insert_keyword ("char", CHAR);
2209 insert_keyword ("class", CLASS);
2210 insert_keyword ("compl", '~');
2211 insert_keyword ("const", CONST);
2212 insert_keyword ("const_cast", CONST_CAST);
2213 insert_keyword ("continue", CONTINUE);
2214 insert_keyword ("default", DEFAULT);
2215 insert_keyword ("delete", DELETE);
2216 insert_keyword ("do", DO);
2217 insert_keyword ("double", DOUBLE);
2218 insert_keyword ("dynamic_cast", DYNAMIC_CAST);
2219 insert_keyword ("else", ELSE);
2220 insert_keyword ("enum", ENUM);
2221 insert_keyword ("explicit", EXPLICIT);
2222 insert_keyword ("extern", EXTERN);
2223 insert_keyword ("false", FALSE);
2224 insert_keyword ("float", FLOAT);
2225 insert_keyword ("for", FOR);
2226 insert_keyword ("friend", FRIEND);
2227 insert_keyword ("goto", GOTO);
2228 insert_keyword ("if", IF);
2229 insert_keyword ("inline", T_INLINE);
2230 insert_keyword ("int", INT);
2231 insert_keyword ("long", LONG);
2232 insert_keyword ("mutable", MUTABLE);
2233 insert_keyword ("namespace", NAMESPACE);
2234 insert_keyword ("new", NEW);
2235 insert_keyword ("not", '!');
2236 insert_keyword ("not_eq", NE);
2237 insert_keyword ("operator", OPERATOR);
2238 insert_keyword ("or", LOR);
2239 insert_keyword ("or_eq", ORASGN);
2240 insert_keyword ("private", PRIVATE);
2241 insert_keyword ("protected", PROTECTED);
2242 insert_keyword ("public", PUBLIC);
2243 insert_keyword ("register", REGISTER);
2244 insert_keyword ("reinterpret_cast", REINTERPRET_CAST);
2245 insert_keyword ("return", RETURN);
2246 insert_keyword ("short", SHORT);
2247 insert_keyword ("signed", SIGNED);
2248 insert_keyword ("sizeof", SIZEOF);
2249 insert_keyword ("static", STATIC);
2250 insert_keyword ("static_cast", STATIC_CAST);
2251 insert_keyword ("struct", STRUCT);
2252 insert_keyword ("switch", SWITCH);
2253 insert_keyword ("template", TEMPLATE);
2254 insert_keyword ("this", THIS);
2255 insert_keyword ("throw", THROW);
2256 insert_keyword ("true", TRUE);
2257 insert_keyword ("try", TRY);
2258 insert_keyword ("typedef", TYPEDEF);
2259 insert_keyword ("typeid", TYPEID);
2260 insert_keyword ("typename", TYPENAME);
2261 insert_keyword ("union", UNION);
2262 insert_keyword ("unsigned", UNSIGNED);
2263 insert_keyword ("using", USING);
2264 insert_keyword ("virtual", VIRTUAL);
2265 insert_keyword ("void", VOID);
2266 insert_keyword ("volatile", VOLATILE);
2267 insert_keyword ("wchar_t", WCHAR);
2268 insert_keyword ("while", WHILE);
2269 insert_keyword ("xor", '^');
2270 insert_keyword ("xor_eq", XORASGN);
2271 }
2272
2273
2274 \f
2275 /***********************************************************************
2276 Parser
2277 ***********************************************************************/
2278
2279 /* Match the current lookahead token and set it to the next token. */
2280
2281 #define MATCH() (tk = yylex ())
2282
2283 /* Return the lookahead token. If current lookahead token is cleared,
2284 read a new token. */
2285
2286 #define LA1 (tk == -1 ? (tk = yylex ()) : tk)
2287
2288 /* Is the current lookahead equal to the token T? */
2289
2290 #define LOOKING_AT(T) (tk == (T))
2291
2292 /* Is the current lookahead one of T1 or T2? */
2293
2294 #define LOOKING_AT2(T1, T2) (tk == (T1) || tk == (T2))
2295
2296 /* Is the current lookahead one of T1, T2 or T3? */
2297
2298 #define LOOKING_AT3(T1, T2, T3) (tk == (T1) || tk == (T2) || tk == (T3))
2299
2300 /* Is the current lookahead one of T1...T4? */
2301
2302 #define LOOKING_AT4(T1, T2, T3, T4) \
2303 (tk == (T1) || tk == (T2) || tk == (T3) || tk == (T4))
2304
2305 /* Match token T if current lookahead is T. */
2306
2307 #define MATCH_IF(T) if (LOOKING_AT (T)) MATCH (); else ((void) 0)
2308
2309 /* Skip to matching token if current token is T. */
2310
2311 #define SKIP_MATCHING_IF(T) \
2312 if (LOOKING_AT (T)) skip_matching (); else ((void) 0)
2313
2314
2315 /* Skip forward until a given token TOKEN or YYEOF is seen and return
2316 the current lookahead token after skipping. */
2317
2318 int
2319 skip_to (token)
2320 int token;
2321 {
2322 while (!LOOKING_AT2 (YYEOF, token))
2323 MATCH ();
2324 return tk;
2325 }
2326
2327
2328 /* Skip over pairs of tokens (parentheses, square brackets,
2329 angle brackets, curly brackets) matching the current lookahead. */
2330
2331 void
2332 skip_matching ()
2333 {
2334 int open, close, n;
2335
2336 switch (open = LA1)
2337 {
2338 case '{':
2339 close = '}';
2340 break;
2341
2342 case '(':
2343 close = ')';
2344 break;
2345
2346 case '<':
2347 close = '>';
2348 break;
2349
2350 case '[':
2351 close = ']';
2352 break;
2353
2354 default:
2355 abort ();
2356 }
2357
2358 for (n = 0;;)
2359 {
2360 if (LOOKING_AT (open))
2361 ++n;
2362 else if (LOOKING_AT (close))
2363 --n;
2364 else if (LOOKING_AT (YYEOF))
2365 break;
2366
2367 MATCH ();
2368
2369 if (n == 0)
2370 break;
2371 }
2372 }
2373
2374
2375 /* Re-initialize the parser by resetting the lookahead token. */
2376
2377 void
2378 re_init_parser ()
2379 {
2380 tk = -1;
2381 }
2382
2383
2384 /* Parse a parameter list, including the const-specifier,
2385 pure-specifier, and throw-list that may follow a parameter list.
2386 Return in FLAGS what was seen following the parameter list.
2387 Returns a hash code for the parameter types. This value is used to
2388 distinguish between overloaded functions. */
2389
2390 unsigned
2391 parm_list (flags)
2392 int *flags;
2393 {
2394 unsigned hash = 0;
2395 int type_seen = 0;
2396
2397 while (!LOOKING_AT2 (YYEOF, ')'))
2398 {
2399 switch (LA1)
2400 {
2401 /* Skip over grouping parens or parameter lists in parameter
2402 declarations. */
2403 case '(':
2404 skip_matching ();
2405 break;
2406
2407 /* Next parameter. */
2408 case ',':
2409 MATCH ();
2410 type_seen = 0;
2411 break;
2412
2413 /* Ignore the scope part of types, if any. This is because
2414 some types need scopes when defined outside of a class body,
2415 and don't need them inside the class body. This means that
2416 we have to look for the last IDENT in a sequence of
2417 IDENT::IDENT::... */
2418 case IDENT:
2419 if (!type_seen)
2420 {
2421 char *last_id;
2422 unsigned ident_type_hash = 0;
2423
2424 parse_qualified_param_ident_or_type (&last_id);
2425 if (last_id)
2426 {
2427 /* LAST_ID null means something like `X::*'. */
2428 for (; *last_id; ++last_id)
2429 ident_type_hash = (ident_type_hash << 1) ^ *last_id;
2430 hash = (hash << 1) ^ ident_type_hash;
2431 type_seen = 1;
2432 }
2433 }
2434 else
2435 MATCH ();
2436 break;
2437
2438 case VOID:
2439 /* This distinction is made to make `func (void)' equivalent
2440 to `func ()'. */
2441 type_seen = 1;
2442 MATCH ();
2443 if (!LOOKING_AT (')'))
2444 hash = (hash << 1) ^ VOID;
2445 break;
2446
2447 case BOOL: case CHAR: case CLASS: case CONST:
2448 case DOUBLE: case ENUM: case FLOAT: case INT:
2449 case LONG: case SHORT: case SIGNED: case STRUCT:
2450 case UNION: case UNSIGNED: case VOLATILE: case WCHAR:
2451 case ELLIPSIS:
2452 type_seen = 1;
2453 hash = (hash << 1) ^ LA1;
2454 MATCH ();
2455 break;
2456
2457 case '*': case '&': case '[': case ']':
2458 hash = (hash << 1) ^ LA1;
2459 MATCH ();
2460 break;
2461
2462 default:
2463 MATCH ();
2464 break;
2465 }
2466 }
2467
2468 if (LOOKING_AT (')'))
2469 {
2470 MATCH ();
2471
2472 if (LOOKING_AT (CONST))
2473 {
2474 /* We can overload the same function on `const' */
2475 hash = (hash << 1) ^ CONST;
2476 SET_FLAG (*flags, F_CONST);
2477 MATCH ();
2478 }
2479
2480 if (LOOKING_AT (THROW))
2481 {
2482 MATCH ();
2483 SKIP_MATCHING_IF ('(');
2484 SET_FLAG (*flags, F_THROW);
2485 }
2486
2487 if (LOOKING_AT ('='))
2488 {
2489 MATCH ();
2490 if (LOOKING_AT (CINT) && yyival == 0)
2491 {
2492 MATCH ();
2493 SET_FLAG (*flags, F_PURE);
2494 }
2495 }
2496 }
2497
2498 return hash;
2499 }
2500
2501
2502 /* Print position info to stdout. */
2503
2504 void
2505 print_info ()
2506 {
2507 if (info_position >= 0 && BUFFER_POS () <= info_position)
2508 if (info_cls)
2509 printf ("(\"%s\" \"%s\" \"%s\" %d)\n",
2510 info_cls->name, sym_scope (info_cls),
2511 info_member->name, info_where);
2512 }
2513
2514
2515 /* Parse a member declaration within the class body of CLS. VIS is
2516 the access specifier for the member (private, protected,
2517 public). */
2518
2519 void
2520 member (cls, vis)
2521 struct sym *cls;
2522 int vis;
2523 {
2524 char *id = NULL;
2525 int sc = SC_MEMBER;
2526 char *regexp = NULL;
2527 int pos;
2528 int is_constructor;
2529 int anonymous = 0;
2530 int flags = 0;
2531 int class_tag;
2532 int type_seen = 0;
2533 int paren_seen = 0;
2534 unsigned hash = 0;
2535 int tilde = 0;
2536
2537 while (!LOOKING_AT4 (';', '{', '}', YYEOF))
2538 {
2539 switch (LA1)
2540 {
2541 default:
2542 MATCH ();
2543 break;
2544
2545 /* A function or class may follow. */
2546 case TEMPLATE:
2547 MATCH();
2548 SET_FLAG (flags, F_TEMPLATE);
2549 /* Skip over template argument list */
2550 SKIP_MATCHING_IF ('<');
2551 break;
2552
2553 case EXPLICIT:
2554 SET_FLAG (flags, F_EXPLICIT);
2555 goto typeseen;
2556
2557 case MUTABLE:
2558 SET_FLAG (flags, F_MUTABLE);
2559 goto typeseen;
2560
2561 case T_INLINE:
2562 SET_FLAG (flags, F_INLINE);
2563 goto typeseen;
2564
2565 case VIRTUAL:
2566 SET_FLAG (flags, F_VIRTUAL);
2567 goto typeseen;
2568
2569 case '[':
2570 skip_matching ();
2571 break;
2572
2573 case ENUM:
2574 sc = SC_TYPE;
2575 goto typeseen;
2576
2577 case TYPEDEF:
2578 sc = SC_TYPE;
2579 goto typeseen;
2580
2581 case FRIEND:
2582 sc = SC_FRIEND;
2583 goto typeseen;
2584
2585 case STATIC:
2586 sc = SC_STATIC;
2587 goto typeseen;
2588
2589 case '~':
2590 tilde = 1;
2591 MATCH ();
2592 break;
2593
2594 case IDENT:
2595 /* Remember IDENTS seen so far. Among these will be the member
2596 name. */
2597 id = (char *) xrealloc (id, strlen (yytext) + 2);
2598 if (tilde)
2599 {
2600 *id = '~';
2601 strcpy (id + 1, yytext);
2602 }
2603 else
2604 strcpy (id, yytext);
2605 MATCH ();
2606 break;
2607
2608 case OPERATOR:
2609 {
2610 char *s = operator_name (&sc);
2611 id = (char *) xrealloc (id, strlen (s) + 1);
2612 strcpy (id, s);
2613 }
2614 break;
2615
2616 case '(':
2617 /* Most probably the beginning of a parameter list. */
2618 MATCH ();
2619 paren_seen = 1;
2620
2621 if (id && cls)
2622 {
2623 if (!(is_constructor = streq (id, cls->name)))
2624 regexp = matching_regexp ();
2625 }
2626 else
2627 is_constructor = 0;
2628
2629 pos = BUFFER_POS ();
2630 hash = parm_list (&flags);
2631
2632 if (is_constructor)
2633 regexp = matching_regexp ();
2634
2635 if (id && cls != NULL)
2636 add_member_decl (cls, id, regexp, pos, hash, 0, sc, vis, flags);
2637
2638 while (!LOOKING_AT3 (';', '{', YYEOF))
2639 MATCH ();
2640
2641 if (LOOKING_AT ('{') && id && cls)
2642 add_member_defn (cls, id, regexp, pos, hash, 0, sc, flags);
2643
2644 xfree (id);
2645 id = NULL;
2646 sc = SC_MEMBER;
2647 break;
2648
2649 case STRUCT: case UNION: case CLASS:
2650 /* Nested class */
2651 class_tag = LA1;
2652 type_seen = 1;
2653 MATCH ();
2654 anonymous = 1;
2655
2656 /* More than one ident here to allow for MS-DOS specialties
2657 like `_export class' etc. The last IDENT seen counts
2658 as the class name. */
2659 while (!LOOKING_AT4 (YYEOF, ';', ':', '{'))
2660 {
2661 if (LOOKING_AT (IDENT))
2662 anonymous = 0;
2663 MATCH ();
2664 }
2665
2666 if (LOOKING_AT2 (':', '{'))
2667 class_definition (anonymous ? NULL : cls, class_tag, flags, 1);
2668 else
2669 skip_to (';');
2670 break;
2671
2672 case INT: case CHAR: case LONG: case UNSIGNED:
2673 case SIGNED: case CONST: case DOUBLE: case VOID:
2674 case SHORT: case VOLATILE: case BOOL: case WCHAR:
2675 case TYPENAME:
2676 typeseen:
2677 type_seen = 1;
2678 MATCH ();
2679 break;
2680 }
2681 }
2682
2683 if (LOOKING_AT (';'))
2684 {
2685 /* The end of a member variable, a friend declaration or an access
2686 declaration. We don't want to add friend classes as members. */
2687 if (id && sc != SC_FRIEND && cls)
2688 {
2689 regexp = matching_regexp ();
2690 pos = BUFFER_POS ();
2691
2692 if (cls != NULL)
2693 {
2694 if (type_seen || !paren_seen)
2695 add_member_decl (cls, id, regexp, pos, 0, 1, sc, vis, 0);
2696 else
2697 add_member_decl (cls, id, regexp, pos, hash, 0, sc, vis, 0);
2698 }
2699 }
2700
2701 MATCH ();
2702 print_info ();
2703 }
2704 else if (LOOKING_AT ('{'))
2705 {
2706 /* A named enum. */
2707 if (sc == SC_TYPE && id && cls)
2708 {
2709 regexp = matching_regexp ();
2710 pos = BUFFER_POS ();
2711
2712 if (cls != NULL)
2713 {
2714 add_member_decl (cls, id, regexp, pos, 0, 1, sc, vis, 0);
2715 add_member_defn (cls, id, regexp, pos, 0, 1, sc, 0);
2716 }
2717 }
2718
2719 skip_matching ();
2720 print_info ();
2721 }
2722
2723 xfree (id);
2724 }
2725
2726
2727 /* Parse the body of class CLS. TAG is the tag of the class (struct,
2728 union, class). */
2729
2730 void
2731 class_body (cls, tag)
2732 struct sym *cls;
2733 int tag;
2734 {
2735 int vis = tag == CLASS ? PRIVATE : PUBLIC;
2736 int temp;
2737
2738 while (!LOOKING_AT2 (YYEOF, '}'))
2739 {
2740 switch (LA1)
2741 {
2742 case PRIVATE: case PROTECTED: case PUBLIC:
2743 temp = LA1;
2744 MATCH ();
2745
2746 if (LOOKING_AT (':'))
2747 {
2748 vis = temp;
2749 MATCH ();
2750 }
2751 else
2752 {
2753 /* Probably conditional compilation for inheritance list.
2754 We don't known whether there comes more of this.
2755 This is only a crude fix that works most of the time. */
2756 do
2757 {
2758 MATCH ();
2759 }
2760 while (LOOKING_AT2 (IDENT, ',')
2761 || LOOKING_AT3 (PUBLIC, PROTECTED, PRIVATE));
2762 }
2763 break;
2764
2765 case TYPENAME:
2766 case USING:
2767 skip_to (';');
2768 break;
2769
2770 /* Try to synchronize */
2771 case CHAR: case CLASS: case CONST:
2772 case DOUBLE: case ENUM: case FLOAT: case INT:
2773 case LONG: case SHORT: case SIGNED: case STRUCT:
2774 case UNION: case UNSIGNED: case VOID: case VOLATILE:
2775 case TYPEDEF: case STATIC: case T_INLINE: case FRIEND:
2776 case VIRTUAL: case TEMPLATE: case IDENT: case '~':
2777 case BOOL: case WCHAR: case EXPLICIT: case MUTABLE:
2778 member (cls, vis);
2779 break;
2780
2781 default:
2782 MATCH ();
2783 break;
2784 }
2785 }
2786 }
2787
2788
2789 /* Parse a qualified identifier. Current lookahead is IDENT. A
2790 qualified ident has the form `X<..>::Y<...>::T<...>. Returns a
2791 symbol for that class. */
2792
2793 struct sym *
2794 parse_classname ()
2795 {
2796 struct sym *last_class = NULL;
2797
2798 while (LOOKING_AT (IDENT))
2799 {
2800 last_class = add_sym (yytext, last_class);
2801 MATCH ();
2802
2803 if (LOOKING_AT ('<'))
2804 {
2805 skip_matching ();
2806 SET_FLAG (last_class->flags, F_TEMPLATE);
2807 }
2808
2809 if (!LOOKING_AT (DCOLON))
2810 break;
2811
2812 MATCH ();
2813 }
2814
2815 return last_class;
2816 }
2817
2818
2819 /* Parse an operator name. Add the `static' flag to *SC if an
2820 implicitly static operator has been parsed. Value is a pointer to
2821 a static buffer holding the constructed operator name string. */
2822
2823 char *
2824 operator_name (sc)
2825 int *sc;
2826 {
2827 static int id_size = 0;
2828 static char *id = NULL;
2829 char *s;
2830 int len;
2831
2832 MATCH ();
2833
2834 if (LOOKING_AT2 (NEW, DELETE))
2835 {
2836 /* `new' and `delete' are implicitly static. */
2837 if (*sc != SC_FRIEND)
2838 *sc = SC_STATIC;
2839
2840 s = token_string (LA1);
2841 MATCH ();
2842
2843 len = strlen (s) + 10;
2844 if (len > id_size)
2845 {
2846 int new_size = max (len, 2 * id_size);
2847 id = (char *) xrealloc (id, new_size);
2848 id_size = new_size;
2849 }
2850 strcpy (id, s);
2851
2852 /* Vector new or delete? */
2853 if (LOOKING_AT ('['))
2854 {
2855 strcat (id, "[");
2856 MATCH ();
2857
2858 if (LOOKING_AT (']'))
2859 {
2860 strcat (id, "]");
2861 MATCH ();
2862 }
2863 }
2864 }
2865 else
2866 {
2867 int tokens_matched = 0;
2868
2869 len = 20;
2870 if (len > id_size)
2871 {
2872 int new_size = max (len, 2 * id_size);
2873 id = (char *) xrealloc (id, new_size);
2874 id_size = new_size;
2875 }
2876 strcpy (id, "operator");
2877
2878 /* Beware access declarations of the form "X::f;" Beware of
2879 `operator () ()'. Yet another difficulty is found in
2880 GCC 2.95's STL: `operator == __STL_NULL_TMPL_ARGS (...'. */
2881 while (!(LOOKING_AT ('(') && tokens_matched)
2882 && !LOOKING_AT2 (';', YYEOF))
2883 {
2884 s = token_string (LA1);
2885 len += strlen (s) + 2;
2886 if (len > id_size)
2887 {
2888 int new_size = max (len, 2 * id_size);
2889 id = (char *) xrealloc (id, new_size);
2890 id_size = new_size;
2891 }
2892
2893 if (*s != ')' && *s != ']')
2894 strcat (id, " ");
2895 strcat (id, s);
2896 MATCH ();
2897
2898 /* If this is a simple operator like `+', stop now. */
2899 if (!isalpha ((unsigned char) *s) && *s != '(' && *s != '[')
2900 break;
2901
2902 ++tokens_matched;
2903 }
2904 }
2905
2906 return id;
2907 }
2908
2909
2910 /* This one consumes the last IDENT of a qualified member name like
2911 `X::Y::z'. This IDENT is returned in LAST_ID. Value if the
2912 symbol structure for the ident. */
2913
2914 struct sym *
2915 parse_qualified_ident_or_type (last_id)
2916 char **last_id;
2917 {
2918 struct sym *cls = NULL;
2919 char *id = NULL;
2920 size_t id_size = 0;
2921
2922 while (LOOKING_AT (IDENT))
2923 {
2924 int len = strlen (yytext) + 1;
2925 if (len > id_size)
2926 {
2927 id = (char *) xrealloc (id, len);
2928 id_size = len;
2929 }
2930 strcpy (id, yytext);
2931 *last_id = id;
2932 MATCH ();
2933
2934 SKIP_MATCHING_IF ('<');
2935
2936 if (LOOKING_AT (DCOLON))
2937 {
2938 cls = add_sym (id, cls);
2939 *last_id = NULL;
2940 xfree (id);
2941 id = NULL;
2942 id_size = 0;
2943 MATCH ();
2944 }
2945 else
2946 break;
2947 }
2948
2949 return cls;
2950 }
2951
2952
2953 /* This one consumes the last IDENT of a qualified member name like
2954 `X::Y::z'. This IDENT is returned in LAST_ID. Value if the
2955 symbol structure for the ident. */
2956
2957 void
2958 parse_qualified_param_ident_or_type (last_id)
2959 char **last_id;
2960 {
2961 struct sym *cls = NULL;
2962 static char *id = NULL;
2963 static int id_size = 0;
2964
2965 while (LOOKING_AT (IDENT))
2966 {
2967 int len = strlen (yytext) + 1;
2968 if (len > id_size)
2969 {
2970 id = (char *) xrealloc (id, len);
2971 id_size = len;
2972 }
2973 strcpy (id, yytext);
2974 *last_id = id;
2975 MATCH ();
2976
2977 SKIP_MATCHING_IF ('<');
2978
2979 if (LOOKING_AT (DCOLON))
2980 {
2981 cls = add_sym (id, cls);
2982 *last_id = NULL;
2983 MATCH ();
2984 }
2985 else
2986 break;
2987 }
2988 }
2989
2990
2991 /* Parse a class definition.
2992
2993 CONTAINING is the class containing the class being parsed or null.
2994 This may also be null if NESTED != 0 if the containing class is
2995 anonymous. TAG is the tag of the class (struct, union, class).
2996 NESTED is non-zero if we are parsing a nested class.
2997
2998 Current lookahead is the class name. */
2999
3000 void
3001 class_definition (containing, tag, flags, nested)
3002 struct sym *containing;
3003 int tag;
3004 int flags;
3005 int nested;
3006 {
3007 struct sym *current;
3008 struct sym *base_class;
3009
3010 /* Set CURRENT to null if no entry has to be made for the class
3011 parsed. This is the case for certain command line flag
3012 settings. */
3013 if ((tag != CLASS && !f_structs) || (nested && !f_nested_classes))
3014 current = NULL;
3015 else
3016 {
3017 current = add_sym (yytext, containing);
3018 current->pos = BUFFER_POS ();
3019 current->regexp = matching_regexp ();
3020 current->filename = filename;
3021 current->flags = flags;
3022 }
3023
3024 /* If at ':', base class list follows. */
3025 if (LOOKING_AT (':'))
3026 {
3027 int done = 0;
3028 MATCH ();
3029
3030 while (!done)
3031 {
3032 switch (LA1)
3033 {
3034 case VIRTUAL: case PUBLIC: case PROTECTED: case PRIVATE:
3035 MATCH ();
3036 break;
3037
3038 case IDENT:
3039 base_class = parse_classname ();
3040 if (base_class && current && base_class != current)
3041 add_link (base_class, current);
3042 break;
3043
3044 /* The `,' between base classes or the end of the base
3045 class list. Add the previously found base class.
3046 It's done this way to skip over sequences of
3047 `A::B::C' until we reach the end.
3048
3049 FIXME: it is now possible to handle `class X : public B::X'
3050 because we have enough information. */
3051 case ',':
3052 MATCH ();
3053 break;
3054
3055 default:
3056 /* A syntax error, possibly due to preprocessor constructs
3057 like
3058
3059 #ifdef SOMETHING
3060 class A : public B
3061 #else
3062 class A : private B.
3063
3064 MATCH until we see something like `;' or `{'. */
3065 while (!LOOKING_AT3 (';', YYEOF, '{'))
3066 MATCH ();
3067 done = 1;
3068
3069 case '{':
3070 done = 1;
3071 break;
3072 }
3073 }
3074 }
3075
3076 /* Parse the class body if there is one. */
3077 if (LOOKING_AT ('{'))
3078 {
3079 if (tag != CLASS && !f_structs)
3080 skip_matching ();
3081 else
3082 {
3083 MATCH ();
3084 class_body (current, tag);
3085
3086 if (LOOKING_AT ('}'))
3087 {
3088 MATCH ();
3089 if (LOOKING_AT (';') && !nested)
3090 MATCH ();
3091 }
3092 }
3093 }
3094 }
3095
3096
3097 /* Parse a declaration. */
3098
3099 void
3100 declaration (flags)
3101 int flags;
3102 {
3103 char *id = NULL;
3104 struct sym *cls = NULL;
3105 char *regexp = NULL;
3106 int pos = 0;
3107 unsigned hash = 0;
3108 int is_constructor;
3109 int sc = 0;
3110
3111 while (!LOOKING_AT3 (';', '{', YYEOF))
3112 {
3113 switch (LA1)
3114 {
3115 default:
3116 MATCH ();
3117 break;
3118
3119 case '[':
3120 skip_matching ();
3121 break;
3122
3123 case ENUM:
3124 case TYPEDEF:
3125 sc = SC_TYPE;
3126 MATCH ();
3127 break;
3128
3129 case STATIC:
3130 sc = SC_STATIC;
3131 MATCH ();
3132 break;
3133
3134 case INT: case CHAR: case LONG: case UNSIGNED:
3135 case SIGNED: case CONST: case DOUBLE: case VOID:
3136 case SHORT: case VOLATILE: case BOOL: case WCHAR:
3137 MATCH ();
3138 break;
3139
3140 case CLASS: case STRUCT: case UNION:
3141 /* This is for the case `STARTWRAP class X : ...' or
3142 `declare (X, Y)\n class A : ...'. */
3143 if (id)
3144 {
3145 xfree (id);
3146 return;
3147 }
3148
3149 case '=':
3150 /* Assumed to be the start of an initialization in this context.
3151 Skip over everything up to ';'. */
3152 skip_to (';');
3153 break;
3154
3155 case OPERATOR:
3156 {
3157 char *s = operator_name (&sc);
3158 id = (char *) xrealloc (id, strlen (s) + 1);
3159 strcpy (id, s);
3160 }
3161 break;
3162
3163 case T_INLINE:
3164 SET_FLAG (flags, F_INLINE);
3165 MATCH ();
3166 break;
3167
3168 case '~':
3169 MATCH ();
3170 if (LOOKING_AT (IDENT))
3171 {
3172 id = (char *) xrealloc (id, strlen (yytext) + 2);
3173 *id = '~';
3174 strcpy (id + 1, yytext);
3175 MATCH ();
3176 }
3177 break;
3178
3179 case IDENT:
3180 cls = parse_qualified_ident_or_type (&id);
3181 break;
3182
3183 case '(':
3184 /* Most probably the beginning of a parameter list. */
3185 if (cls)
3186 {
3187 MATCH ();
3188
3189 if (id && cls)
3190 {
3191 if (!(is_constructor = streq (id, cls->name)))
3192 regexp = matching_regexp ();
3193 }
3194 else
3195 is_constructor = 0;
3196
3197 pos = BUFFER_POS ();
3198 hash = parm_list (&flags);
3199
3200 if (is_constructor)
3201 regexp = matching_regexp ();
3202
3203 if (id && cls)
3204 add_member_defn (cls, id, regexp, pos, hash, 0,
3205 SC_UNKNOWN, flags);
3206 }
3207 else
3208 {
3209 /* This may be a C functions, but also a macro
3210 call of the form `declare (A, B)' --- such macros
3211 can be found in some class libraries. */
3212 MATCH ();
3213
3214 if (id)
3215 {
3216 regexp = matching_regexp ();
3217 pos = BUFFER_POS ();
3218 hash = parm_list (&flags);
3219 add_global_decl (id, regexp, pos, hash, 0, sc, flags);
3220 }
3221
3222 /* This is for the case that the function really is
3223 a macro with no `;' following it. If a CLASS directly
3224 follows, we would miss it otherwise. */
3225 if (LOOKING_AT3 (CLASS, STRUCT, UNION))
3226 return;
3227 }
3228
3229 while (!LOOKING_AT3 (';', '{', YYEOF))
3230 MATCH ();
3231
3232 if (!cls && id && LOOKING_AT ('{'))
3233 add_global_defn (id, regexp, pos, hash, 0, sc, flags);
3234
3235 xfree (id);
3236 id = NULL;
3237 break;
3238 }
3239 }
3240
3241 if (LOOKING_AT (';'))
3242 {
3243 /* The end of a member variable or of an access declaration
3244 `X::f'. To distinguish between them we have to know whether
3245 type information has been seen. */
3246 if (id)
3247 {
3248 char *regexp = matching_regexp ();
3249 int pos = BUFFER_POS ();
3250
3251 if (cls)
3252 add_member_defn (cls, id, regexp, pos, 0, 1, SC_UNKNOWN, flags);
3253 else
3254 add_global_defn (id, regexp, pos, 0, 1, sc, flags);
3255 }
3256
3257 MATCH ();
3258 print_info ();
3259 }
3260 else if (LOOKING_AT ('{'))
3261 {
3262 if (sc == SC_TYPE && id)
3263 {
3264 /* A named enumeration. */
3265 regexp = matching_regexp ();
3266 pos = BUFFER_POS ();
3267 add_global_defn (id, regexp, pos, 0, 1, sc, flags);
3268 }
3269
3270 skip_matching ();
3271 print_info ();
3272 }
3273
3274 xfree (id);
3275 }
3276
3277
3278 /* Parse a list of top-level declarations/definitions. START_FLAGS
3279 says in which context we are parsing. If it is F_EXTERNC, we are
3280 parsing in an `extern "C"' block. Value is 1 if EOF is reached, 0
3281 otherwise. */
3282
3283 int
3284 globals (start_flags)
3285 int start_flags;
3286 {
3287 int anonymous;
3288 int class_tk;
3289 int flags = start_flags;
3290
3291 for (;;)
3292 {
3293 char *prev_in = in;
3294
3295 switch (LA1)
3296 {
3297 case NAMESPACE:
3298 {
3299 MATCH ();
3300
3301 if (LOOKING_AT (IDENT))
3302 {
3303 char *namespace_name = xstrdup (yytext);
3304 MATCH ();
3305
3306 if (LOOKING_AT ('='))
3307 {
3308 MATCH ();
3309 if (LOOKING_AT (IDENT))
3310 register_namespace_alias (namespace_name, yytext);
3311
3312 if (skip_to (';') == ';')
3313 MATCH ();
3314 }
3315 else if (LOOKING_AT ('{'))
3316 {
3317 MATCH ();
3318 enter_namespace (namespace_name);
3319 globals (0);
3320 leave_namespace ();
3321 MATCH_IF ('}');
3322 }
3323
3324 xfree (namespace_name);
3325 }
3326 }
3327 break;
3328
3329 case EXTERN:
3330 MATCH ();
3331 if (LOOKING_AT (CSTRING) && *string_start == 'C'
3332 && *(string_start + 1) == '"')
3333 {
3334 /* This is `extern "C"'. */
3335 MATCH ();
3336
3337 if (LOOKING_AT ('{'))
3338 {
3339 MATCH ();
3340 globals (F_EXTERNC);
3341 MATCH_IF ('}');
3342 }
3343 else
3344 SET_FLAG (flags, F_EXTERNC);
3345 }
3346 break;
3347
3348 case TEMPLATE:
3349 MATCH ();
3350 SKIP_MATCHING_IF ('<');
3351 SET_FLAG (flags, F_TEMPLATE);
3352 break;
3353
3354 case CLASS: case STRUCT: case UNION:
3355 class_tk = LA1;
3356 MATCH ();
3357 anonymous = 1;
3358
3359 /* More than one ident here to allow for MS-DOS and OS/2
3360 specialties like `far', `_Export' etc. Some C++ libs
3361 have constructs like `_OS_DLLIMPORT(_OS_CLIENT)' in front
3362 of the class name. */
3363 while (!LOOKING_AT4 (YYEOF, ';', ':', '{'))
3364 {
3365 if (LOOKING_AT (IDENT))
3366 anonymous = 0;
3367 MATCH ();
3368 }
3369
3370 /* Don't add anonymous unions. */
3371 if (LOOKING_AT2 (':', '{') && !anonymous)
3372 class_definition (NULL, class_tk, flags, 0);
3373 else
3374 {
3375 if (skip_to (';') == ';')
3376 MATCH ();
3377 }
3378
3379 flags = start_flags;
3380 break;
3381
3382 case YYEOF:
3383 return 1;
3384
3385 case '}':
3386 return 0;
3387
3388 default:
3389 declaration (flags);
3390 flags = start_flags;
3391 break;
3392 }
3393
3394 if (prev_in == in)
3395 yyerror ("parse error", NULL);
3396 }
3397 }
3398
3399
3400 /* Parse the current input file. */
3401
3402 void
3403 yyparse ()
3404 {
3405 while (globals (0) == 0)
3406 MATCH_IF ('}');
3407 }
3408
3409
3410 \f
3411 /***********************************************************************
3412 Main Program
3413 ***********************************************************************/
3414
3415 /* Add the list of paths PATH_LIST to the current search path for
3416 input files. */
3417
3418 void
3419 add_search_path (path_list)
3420 char *path_list;
3421 {
3422 while (*path_list)
3423 {
3424 char *start = path_list;
3425 struct search_path *p;
3426
3427 while (*path_list && *path_list != PATH_LIST_SEPARATOR)
3428 ++path_list;
3429
3430 p = (struct search_path *) xmalloc (sizeof *p);
3431 p->path = (char *) xmalloc (path_list - start + 1);
3432 memcpy (p->path, start, path_list - start);
3433 p->path[path_list - start] = '\0';
3434 p->next = NULL;
3435
3436 if (search_path_tail)
3437 {
3438 search_path_tail->next = p;
3439 search_path_tail = p;
3440 }
3441 else
3442 search_path = search_path_tail = p;
3443
3444 while (*path_list == PATH_LIST_SEPARATOR)
3445 ++path_list;
3446 }
3447 }
3448
3449
3450 /* Open FILE and return a file handle for it, or -1 if FILE cannot be
3451 opened. Try to find FILE in search_path first, then try the
3452 unchanged file name. */
3453
3454 FILE *
3455 open_file (file)
3456 char *file;
3457 {
3458 FILE *fp = NULL;
3459 static char *buffer;
3460 static int buffer_size;
3461 struct search_path *path;
3462 int flen = strlen (file) + 1; /* +1 for the slash */
3463
3464 filename = xstrdup (file);
3465
3466 for (path = search_path; path && fp == NULL; path = path->next)
3467 {
3468 int len = strlen (path->path) + flen;
3469
3470 if (len + 1 >= buffer_size)
3471 {
3472 buffer_size = max (len + 1, 2 * buffer_size);
3473 buffer = (char *) xrealloc (buffer, buffer_size);
3474 }
3475
3476 strcpy (buffer, path->path);
3477 strcat (buffer, "/");
3478 strcat (buffer, file);
3479 fp = fopen (buffer, "r");
3480 }
3481
3482 /* Try the original file name. */
3483 if (fp == NULL)
3484 fp = fopen (file, "r");
3485
3486 if (fp == NULL)
3487 yyerror ("cannot open", NULL);
3488
3489 return fp;
3490 }
3491
3492
3493 /* Display usage information and exit program. */
3494
3495 #define USAGE "\
3496 Usage: ebrowse [options] {files}\n\
3497 \n\
3498 -a, --append append output to existing file\n\
3499 -f, --files=FILES read input file names from FILE\n\
3500 -I, --search-path=LIST set search path for input files\n\
3501 -m, --min-regexp-length=N set minimum regexp length to N\n\
3502 -M, --max-regexp-length=N set maximum regexp length to N\n\
3503 -n, --no-nested-classes exclude nested classes\n\
3504 -o, --output-file=FILE set output file name to FILE\n\
3505 -p, --position-info print info about position in file\n\
3506 -s, --no-structs-or-unions don't record structs or unions\n\
3507 -v, --verbose be verbose\n\
3508 -V, --very-verbose be very verbose\n\
3509 -x, --no-regexps don't record regular expressions\n\
3510 --help display this help\n\
3511 --version display version info\n\
3512 "
3513
3514 void
3515 usage (error)
3516 int error;
3517 {
3518 puts (USAGE);
3519 exit (error ? 1 : 0);
3520 }
3521
3522
3523 /* Display version and copyright info. The VERSION macro is set
3524 from the Makefile and contains the Emacs version. */
3525
3526 #ifndef VERSION
3527 # define VERSION "21"
3528 #endif
3529
3530 void
3531 version ()
3532 {
3533 printf ("ebrowse %s\n", VERSION);
3534 puts ("Copyright (C) 1992-1999, 2000 Free Software Foundation, Inc.");
3535 puts ("This program is distributed under the same terms as Emacs.");
3536 exit (0);
3537 }
3538
3539
3540 /* Parse one input file FILE, adding classes and members to the symbol
3541 table. */
3542
3543 void
3544 process_file (file)
3545 char *file;
3546 {
3547 FILE *fp;
3548
3549 fp = open_file (file);
3550 if (fp)
3551 {
3552 int nread, nbytes;
3553
3554 /* Give a progress indication if needed. */
3555 if (f_very_verbose)
3556 {
3557 puts (filename);
3558 fflush (stdout);
3559 }
3560 else if (f_verbose)
3561 {
3562 putchar ('.');
3563 fflush (stdout);
3564 }
3565
3566 /* Read file to inbuffer. */
3567 for (nread = 0;;)
3568 {
3569 if (nread + READ_CHUNK_SIZE >= inbuffer_size)
3570 {
3571 inbuffer_size = nread + READ_CHUNK_SIZE + 1;
3572 inbuffer = (char *) xrealloc (inbuffer, inbuffer_size);
3573 }
3574
3575 nbytes = fread (inbuffer + nread, 1, READ_CHUNK_SIZE, fp);
3576 if (nbytes <= 0)
3577 break;
3578 nread += nbytes;
3579 }
3580 if (nread < 0)
3581 nread = 0;
3582 inbuffer[nread] = '\0';
3583
3584 /* Reinitialize scanner and parser for the new input file. */
3585 re_init_scanner ();
3586 re_init_parser ();
3587
3588 /* Parse it and close the file. */
3589 yyparse ();
3590 fclose (fp);
3591 }
3592 }
3593
3594
3595 /* Read a line from stream FP and return a pointer to a static buffer
3596 containing its contents without the terminating newline. Value
3597 is null when EOF is reached. */
3598
3599 char *
3600 read_line (fp)
3601 FILE *fp;
3602 {
3603 static char *buffer;
3604 static int buffer_size;
3605 int i = 0, c;
3606
3607 while ((c = getc (fp)) != EOF && c != '\n')
3608 {
3609 if (i >= buffer_size)
3610 {
3611 buffer_size = max (100, buffer_size * 2);
3612 buffer = (char *) xrealloc (buffer, buffer_size);
3613 }
3614
3615 buffer[i++] = c;
3616 }
3617
3618 if (c == EOF && i == 0)
3619 return NULL;
3620
3621 if (i == buffer_size)
3622 {
3623 buffer_size = max (100, buffer_size * 2);
3624 buffer = (char *) xrealloc (buffer, buffer_size);
3625 }
3626
3627 buffer[i] = '\0';
3628 if (i > 0 && buffer[i - 1] == '\r')
3629 buffer[i - 1] = '\0';
3630 return buffer;
3631 }
3632
3633
3634 /* Main entry point. */
3635
3636 int
3637 main (argc, argv)
3638 int argc;
3639 char **argv;
3640 {
3641 int i;
3642 int any_inputfiles = 0;
3643 static char *out_filename = DEFAULT_OUTFILE;
3644 static char **input_filenames = NULL;
3645 static int input_filenames_size = 0;
3646 static int n_input_files;
3647
3648 filename = "command line";
3649 yyout = stdout;
3650
3651 while ((i = getopt_long (argc, argv, "af:I:m:M:no:p:svVx",
3652 options, NULL)) != EOF)
3653 {
3654 switch (i)
3655 {
3656 /* Experimental. */
3657 case 'p':
3658 info_position = atoi (optarg);
3659 break;
3660
3661 case 'n':
3662 f_nested_classes = 0;
3663 break;
3664
3665 case 'x':
3666 f_regexps = 0;
3667 break;
3668
3669 /* Add the name of a file containing more input files. */
3670 case 'f':
3671 if (n_input_files == input_filenames_size)
3672 {
3673 input_filenames_size = max (10, 2 * input_filenames_size);
3674 input_filenames = (char **) xrealloc ((void *)input_filenames,
3675 input_filenames_size);
3676 }
3677 input_filenames[n_input_files++] = xstrdup (optarg);
3678 break;
3679
3680 /* Append new output to output file instead of truncating it. */
3681 case 'a':
3682 f_append = 1;
3683 break;
3684
3685 /* Include structs in the output */
3686 case 's':
3687 f_structs = 0;
3688 break;
3689
3690 /* Be verbose (give a progress indication). */
3691 case 'v':
3692 f_verbose = 1;
3693 break;
3694
3695 /* Be very verbose (print file names as they are processed). */
3696 case 'V':
3697 f_verbose = 1;
3698 f_very_verbose = 1;
3699 break;
3700
3701 /* Change the name of the output file. */
3702 case 'o':
3703 out_filename = optarg;
3704 break;
3705
3706 /* Set minimum length for regular expression strings
3707 when recorded in the output file. */
3708 case 'm':
3709 min_regexp = atoi (optarg);
3710 break;
3711
3712 /* Set maximum length for regular expression strings
3713 when recorded in the output file. */
3714 case 'M':
3715 max_regexp = atoi (optarg);
3716 break;
3717
3718 /* Add to search path. */
3719 case 'I':
3720 add_search_path (optarg);
3721 break;
3722
3723 /* Display help */
3724 case -2:
3725 usage (0);
3726 break;
3727
3728 case -3:
3729 version ();
3730 break;
3731 }
3732 }
3733
3734 /* Call init_scanner after command line flags have been processed to be
3735 able to add keywords depending on command line (not yet
3736 implemented). */
3737 init_scanner ();
3738 init_sym ();
3739
3740 /* Open output file */
3741 if (*out_filename)
3742 {
3743 if (f_append)
3744 {
3745 /* Check that the file to append to exists, and is not
3746 empty. More specifically, it should be a valid file
3747 produced by a previous run of ebrowse, but that's too
3748 difficult to check. */
3749 FILE *fp;
3750 int rc;
3751
3752 fp = fopen (out_filename, "r");
3753 if (fp == NULL)
3754 yyerror ("file `%s' must exist for --append", out_filename);
3755
3756 rc = fseek (fp, 0, SEEK_END);
3757 if (rc == -1)
3758 yyerror ("error seeking in file `%s'", out_filename);
3759
3760 rc = ftell (fp);
3761 if (rc == -1)
3762 yyerror ("error getting size of file `%s'", out_filename);
3763 else if (rc == 0)
3764 yyerror ("file `%s' is empty", out_filename);
3765
3766 fclose (fp);
3767 }
3768
3769 yyout = fopen (out_filename, f_append ? "a" : "w");
3770 if (yyout == NULL)
3771 {
3772 yyerror ("cannot open output file `%s'", out_filename);
3773 exit (1);
3774 }
3775 }
3776
3777 /* Process input files specified on the command line. */
3778 while (optind < argc)
3779 {
3780 process_file (argv[optind++]);
3781 any_inputfiles = 1;
3782 }
3783
3784 /* Process files given on stdin if no files specified. */
3785 if (!any_inputfiles && n_input_files == 0)
3786 {
3787 char *file;
3788 while ((file = read_line (stdin)) != NULL)
3789 process_file (file);
3790 }
3791 else
3792 {
3793 /* Process files from `--files=FILE'. Every line in FILE names
3794 one input file to process. */
3795 for (i = 0; i < n_input_files; ++i)
3796 {
3797 FILE *fp = fopen (input_filenames[i], "r");
3798
3799 if (fp == NULL)
3800 yyerror ("cannot open input file `%s'", input_filenames[i]);
3801 else
3802 {
3803 char *file;
3804 while ((file = read_line (fp)) != NULL)
3805 process_file (file);
3806 fclose (fp);
3807 }
3808 }
3809 }
3810
3811 /* Write output file. */
3812 dump_roots (yyout);
3813
3814 /* Close output file. */
3815 if (yyout != stdout)
3816 fclose (yyout);
3817
3818 return 0;
3819 }
3820
3821
3822 /* ebrowse.c ends here. */
Emacs repository mirror