1 /* ebrowse.c --- parsing files for the ebrowse C++ browser
2
3 Copyright (C) 1992-2021 Free Software Foundation, Inc.
4
5 This file is part of GNU Emacs.
6
7 GNU Emacs is free software: you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation, either version 3 of the License, or (at
10 your option) any later version.
11
12 GNU Emacs is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with GNU Emacs. If not, see <https://www.gnu.org/licenses/>. */
19
20
21 #include <config.h>
22 #include <stddef.h>
23 #include <stdlib.h>
24 #include <string.h>
25 #include <ctype.h>
26 #include <assert.h>
27 #include <getopt.h>
28
29 #include <flexmember.h>
30 #include <min-max.h>
31 #include <unlocked-io.h>
32
33 /* The SunOS compiler doesn't have SEEK_END. */
34 #ifndef SEEK_END
35 #define SEEK_END 2
36 #endif
37
38 /* Files are read in chunks of this number of bytes. */
39
40 enum { READ_CHUNK_SIZE = 100 * 1024 };
41
42 /* Value is true if strings X and Y compare equal. */
43
44 static bool
streq(char const * x,char const * y)45 streq (char const *x, char const *y)
46 {
47 return strcmp (x, y) == 0;
48 }
49
50 static bool
filename_eq(char const * x,char const * y)51 filename_eq (char const *x, char const *y)
52 {
53 #ifdef __MSDOS__
54 return strcasecmp (x, y) == 0;
55 #elif defined WINDOWSNT
56 return stricmp (x, y) == 0;
57 #else
58 return streq (x, y);
59 #endif
60 }
61
62 /* The default output file name. */
63
64 #define DEFAULT_OUTFILE "BROWSE"
65
66 /* A version string written to the output file. Change this whenever
67 the structure of the output file changes. */
68
69 #define EBROWSE_FILE_VERSION "ebrowse 5.0"
70
71 /* The output file consists of a tree of Lisp objects, with major
72 nodes built out of Lisp structures. These are the heads of the
73 Lisp structs with symbols identifying their type. */
74
75 #define TREE_HEADER_STRUCT "[ebrowse-hs "
76 #define TREE_STRUCT "[ebrowse-ts "
77 #define MEMBER_STRUCT "[ebrowse-ms "
78 #define CLASS_STRUCT "[ebrowse-cs "
79
80 /* The name of the symbol table entry for global functions, variables,
81 defines etc. This name also appears in the browser display. */
82
83 #define GLOBALS_NAME "*Globals*"
84
85 /* Token definitions. */
86
87 enum token
88 {
89 YYEOF = 0, /* end of file */
90 CSTRING = 256, /* string constant */
91 CCHAR, /* character constant */
92 CINT, /* integral constant */
93 CFLOAT, /* real constant */
94
95 ELLIPSIS, /* ... */
96 LSHIFTASGN, /* <<= */
97 RSHIFTASGN, /* >>= */
98 ARROWSTAR, /* ->* */
99 IDENT, /* identifier */
100 DIVASGN, /* /= */
101 INC, /* ++ */
102 ADDASGN, /* += */
103 DEC, /* -- */
104 ARROW, /* -> */
105 SUBASGN, /* -= */
106 MULASGN, /* *= */
107 MODASGN, /* %= */
108 LOR, /* || */
109 ORASGN, /* |= */
110 LAND, /* && */
111 ANDASGN, /* &= */
112 XORASGN, /* ^= */
113 POINTSTAR, /* .* */
114 DCOLON, /* :: */
115 EQ, /* == */
116 NE, /* != */
117 LE, /* <= */
118 LSHIFT, /* << */
119 GE, /* >= */
120 RSHIFT, /* >> */
121
122 /* Keywords. The undef's are there because these
123 three symbols are very likely to be defined somewhere. */
124 #undef BOOL
125 #undef TRUE
126 #undef FALSE
127
128 ASM, /* asm */
129 AUTO, /* auto */
130 BREAK, /* break */
131 CASE, /* case */
132 CATCH, /* catch */
133 CHAR, /* char */
134 CLASS, /* class */
135 CONST, /* const */
136 CONTINUE, /* continue */
137 DEFAULT, /* default */
138 DELETE, /* delete */
139 DO, /* do */
140 DOUBLE, /* double */
141 ELSE, /* else */
142 ENUM, /* enum */
143 EXTERN, /* extern */
144 FLOAT, /* float */
145 FOR, /* for */
146 FRIEND, /* friend */
147 GOTO, /* goto */
148 IF, /* if */
149 T_INLINE, /* inline */
150 INT, /* int */
151 LONG, /* long */
152 NEW, /* new */
153 OPERATOR, /* operator */
154 PRIVATE, /* private */
155 PROTECTED, /* protected */
156 PUBLIC, /* public */
157 REGISTER, /* register */
158 RETURN, /* return */
159 SHORT, /* short */
160 SIGNED, /* signed */
161 SIZEOF, /* sizeof */
162 STATIC, /* static */
163 STRUCT, /* struct */
164 SWITCH, /* switch */
165 TEMPLATE, /* template */
166 THIS, /* this */
167 THROW, /* throw */
168 TRY, /* try */
169 TYPEDEF, /* typedef */
170 UNION, /* union */
171 UNSIGNED, /* unsigned */
172 VIRTUAL, /* virtual */
173 VOID, /* void */
174 VOLATILE, /* volatile */
175 WHILE, /* while */
176 MUTABLE, /* mutable */
177 BOOL, /* bool */
178 TRUE, /* true */
179 FALSE, /* false */
180 SIGNATURE, /* signature (GNU extension) */
181 NAMESPACE, /* namespace */
182 EXPLICIT, /* explicit */
183 TYPENAME, /* typename */
184 CONST_CAST, /* const_cast */
185 DYNAMIC_CAST, /* dynamic_cast */
186 REINTERPRET_CAST, /* reinterpret_cast */
187 STATIC_CAST, /* static_cast */
188 TYPEID, /* typeid */
189 USING, /* using */
190 WCHAR, /* wchar_t */
191 FINAL /* final */
192 };
193
194 /* Storage classes, in a wider sense. */
195
196 enum sc
197 {
198 SC_UNKNOWN,
199 SC_MEMBER, /* Is an instance member. */
200 SC_STATIC, /* Is static member. */
201 SC_FRIEND, /* Is friend function. */
202 SC_TYPE /* Is a type definition. */
203 };
204
205 /* Member visibility. */
206
207 enum visibility
208 {
209 V_PUBLIC,
210 V_PROTECTED,
211 V_PRIVATE
212 };
213
214 /* Member flags. */
215
216 #define F_VIRTUAL 1 /* Is virtual function. */
217 #define F_INLINE 2 /* Is inline function. */
218 #define F_CONST 4 /* Is const. */
219 #define F_PURE 8 /* Is pure virtual function. */
220 #define F_MUTABLE 16 /* Is mutable. */
221 #define F_TEMPLATE 32 /* Is a template. */
222 #define F_EXPLICIT 64 /* Is explicit constructor. */
223 #define F_THROW 128 /* Has a throw specification. */
224 #define F_EXTERNC 256 /* Is declared extern "C". */
225 #define F_DEFINE 512 /* Is a #define. */
226
227 /* Set and test a bit in an int. */
228
229 static void
set_flag(int * f,int flag)230 set_flag (int *f, int flag)
231 {
232 *f |= flag;
233 }
234
235 static bool
has_flag(int f,int flag)236 has_flag (int f, int flag)
237 {
238 return (f & flag) != 0;
239 }
240
241 /* Structure describing a class member. */
242
243 struct member
244 {
245 struct member *next; /* Next in list of members. */
246 struct member *anext; /* Collision chain in member_table. */
247 struct member **list; /* Pointer to list in class. */
248 unsigned param_hash; /* Hash value for parameter types. */
249 int vis; /* Visibility (public, ...). */
250 int flags; /* See F_* above. */
251 char *regexp; /* Matching regular expression. */
252 const char *filename; /* Don't free this shared string. */
253 int pos; /* Buffer position of occurrence. */
254 char *def_regexp; /* Regular expression matching definition. */
255 const char *def_filename; /* File name of definition. */
256 int def_pos; /* Buffer position of definition. */
257 char name[FLEXIBLE_ARRAY_MEMBER]; /* Member name. */
258 };
259
260 /* Structures of this type are used to connect class structures with
261 their super and subclasses. */
262
263 struct link
264 {
265 struct sym *sym; /* The super or subclass. */
266 struct link *next; /* Next in list or NULL. */
267 };
268
269 /* Structure used to record namespace aliases. */
270
271 struct alias
272 {
273 struct alias *next; /* Next in list. */
274 struct sym *namesp; /* Namespace in which defined. */
275 struct link *aliasee; /* List of aliased namespaces (A::B::C...). */
276 char name[FLEXIBLE_ARRAY_MEMBER]; /* Alias name. */
277 };
278
279 /* The structure used to describe a class in the symbol table,
280 or a namespace in all_namespaces. */
281
282 struct sym
283 {
284 int flags; /* Is class a template class?. */
285 unsigned char visited; /* Used to find circles. */
286 struct sym *next; /* Hash collision list. */
287 struct link *subs; /* List of subclasses. */
288 struct link *supers; /* List of superclasses. */
289 struct member *vars; /* List of instance variables. */
290 struct member *fns; /* List of instance functions. */
291 struct member *static_vars; /* List of static variables. */
292 struct member *static_fns; /* List of static functions. */
293 struct member *friends; /* List of friend functions. */
294 struct member *types; /* List of local types. */
295 char *regexp; /* Matching regular expression. */
296 int pos; /* Buffer position. */
297 const char *filename; /* File in which it can be found. */
298 const char *sfilename; /* File in which members can be found. */
299 struct sym *namesp; /* Namespace in which defined. . */
300 char name[FLEXIBLE_ARRAY_MEMBER]; /* Name of the class. */
301 };
302
303 /* Experimental: Print info for `--position-info'. We print
304 '(CLASS-NAME SCOPE MEMBER-NAME). */
305
306 #define P_DEFN 1
307 #define P_DECL 2
308
309 static int info_where;
310 static struct sym *info_cls = NULL;
311 static struct member *info_member = NULL;
312
313 /* Experimental. For option `--position-info', the buffer position we
314 are interested in. When this position is reached, print out
315 information about what we know about that point. */
316
317 static int info_position = -1;
318
319 /* Command line options structure for getopt_long. */
320
321 static struct option const options[] =
322 {
323 {"append", no_argument, NULL, 'a'},
324 {"files", required_argument, NULL, 'f'},
325 {"help", no_argument, NULL, -2},
326 {"min-regexp-length", required_argument, NULL, 'm'},
327 {"max-regexp-length", required_argument, NULL, 'M'},
328 {"no-nested-classes", no_argument, NULL, 'n'},
329 {"no-regexps", no_argument, NULL, 'x'},
330 {"no-structs-or-unions", no_argument, NULL, 's'},
331 {"output-file", required_argument, NULL, 'o'},
332 {"position-info", required_argument, NULL, 'p'},
333 {"search-path", required_argument, NULL, 'I'},
334 {"verbose", no_argument, NULL, 'v'},
335 {"version", no_argument, NULL, -3},
336 {"very-verbose", no_argument, NULL, 'V'},
337 {NULL, 0, NULL, 0}
338 };
339
340 /* Semantic values of tokens. Set by yylex.. */
341
342 static unsigned yyival; /* Set for token CINT. */
343 static char *yytext; /* Set for token IDENT. */
344 static char *yytext_end;
345
346 /* Output file. */
347
348 static FILE *yyout;
349
350 /* Current line number. */
351
352 static int yyline;
353
354 /* The name of the current input file. */
355
356 static const char *filename;
357
358 /* Three character class vectors, and macros to test membership
359 of characters. */
360
361 static char is_ident[255];
362 static char is_digit[255];
363 static char is_white[255];
364
365 #define IDENTP(C) is_ident[(unsigned char) (C)]
366 #define DIGITP(C) is_digit[(unsigned char) (C)]
367 #define WHITEP(C) is_white[(unsigned char) (C)]
368
369 /* Command line flags. */
370
371 static int f_append;
372 static int f_verbose;
373 static int f_very_verbose;
374 static int f_structs = 1;
375 static int f_regexps = 1;
376 static int f_nested_classes = 1;
377
378 /* Maximum and minimum lengths of regular expressions matching a
379 member, class etc., for writing them to the output file. These are
380 overridable from the command line. */
381
382 static int min_regexp = 5;
383 static int max_regexp = 50;
384
385 /* Input buffer. */
386
387 static char *inbuffer;
388 static char *in;
389 static size_t inbuffer_size;
390
391 /* Return the current buffer position in the input file. */
392
393 #define BUFFER_POS() (in - inbuffer)
394
395 /* If current lookahead is CSTRING, the following points to the
396 first character in the string constant. Used for recognizing
397 extern "C". */
398
399 static char *string_start;
400
401 /* The size of the hash tables for classes.and members. Should be
402 prime. */
403
404 #define TABLE_SIZE 1001
405
406 /* The hash table for class symbols. */
407
408 static struct sym *class_table[TABLE_SIZE];
409
410 /* Hash table containing all member structures. This is generally
411 faster for member lookup than traversing the member lists of a
412 `struct sym'. */
413
414 static struct member *member_table[TABLE_SIZE];
415
416 /* Hash table for namespace aliases */
417
418 static struct alias *namespace_alias_table[TABLE_SIZE];
419
420 /* The special class symbol used to hold global functions,
421 variables etc. */
422
423 static struct sym *global_symbols;
424
425 /* The current namespace. */
426
427 static struct sym *current_namespace;
428
429 /* The list of all known namespaces. */
430
431 static struct sym *all_namespaces;
432
433 /* Stack of namespaces we're currently nested in, during the parse. */
434
435 static struct sym **namespace_stack;
436 static int namespace_stack_size;
437 static int namespace_sp;
438
439 /* The current lookahead token. */
440
441 static int tk = -1;
442
443 /* Structure describing a keyword. */
444
445 struct kw
446 {
447 const char *name; /* Spelling. */
448 int tk; /* Token value. */
449 struct kw *next; /* Next in collision chain. */
450 };
451
452 /* Keywords are lookup up in a hash table of their own. */
453
454 #define KEYWORD_TABLE_SIZE 1001
455 static struct kw *keyword_table[KEYWORD_TABLE_SIZE];
456
457 /* Search path. */
458
459 struct search_path
460 {
461 char *path;
462 struct search_path *next;
463 };
464
465 static struct search_path *search_path;
466 static struct search_path *search_path_tail;
467
468 /* Function prototypes. */
469
470 static char *matching_regexp (void);
471 static struct sym *add_sym (const char *, struct sym *);
472 static void add_global_defn (char *, char *, int, unsigned, int, int, int);
473 static void add_global_decl (char *, char *, int, unsigned, int, int, int);
474 static struct member *add_member (struct sym *, char *, int, int, unsigned);
475 static void class_definition (struct sym *, const char *, int, int, int);
476 static char *operator_name (int *);
477 static void parse_qualified_param_ident_or_type (char **);
478
479 /***********************************************************************
480 Utilities
481 ***********************************************************************/
482
483 /* Print an error in a printf-like style with the current input file
484 name and line number. */
485
486 static void
yyerror(const char * format,const char * s)487 yyerror (const char *format, const char *s)
488 {
489 fprintf (stderr, "%s:%d: ", filename, yyline);
490 fprintf (stderr, format, s);
491 putc ('\n', stderr);
492 }
493
494
495 /* Like malloc but print an error and exit if not enough memory is
496 available. */
497
498 static void * ATTRIBUTE_MALLOC
xmalloc(size_t nbytes)499 xmalloc (size_t nbytes)
500 {
501 void *p = malloc (nbytes);
502 if (p == NULL)
503 {
504 yyerror ("out of memory", NULL);
505 exit (EXIT_FAILURE);
506 }
507 return p;
508 }
509
510
511 /* Like realloc but print an error and exit if out of memory. */
512
513 static void *
xrealloc(void * p,size_t sz)514 xrealloc (void *p, size_t sz)
515 {
516 p = realloc (p, sz);
517 if (p == NULL)
518 {
519 yyerror ("out of memory", NULL);
520 exit (EXIT_FAILURE);
521 }
522 return p;
523 }
524
525
526 /* Like strdup, but print an error and exit if not enough memory is
527 available.. If S is null, return null. */
528
529 static char *
xstrdup(char * s)530 xstrdup (char *s)
531 {
532 if (s)
533 return strcpy (xmalloc (strlen (s) + 1), s);
534 return s;
535 }
536
537
538
539 /***********************************************************************
540 Symbols
541 ***********************************************************************/
542
543 /* Initialize the symbol table. This currently only sets up the
544 special symbol for globals (`*Globals*'). */
545
546 static void
init_sym(void)547 init_sym (void)
548 {
549 global_symbols = add_sym (GLOBALS_NAME, NULL);
550 }
551
552
553 /* Add a symbol for class NAME to the symbol table. NESTED_IN_CLASS
554 is the class in which class NAME was found. If it is null,
555 this means the scope of NAME is the current namespace.
556
557 If a symbol for NAME already exists, return that. Otherwise
558 create a new symbol and set it to default values. */
559
560 static struct sym *
add_sym(const char * name,struct sym * nested_in_class)561 add_sym (const char *name, struct sym *nested_in_class)
562 {
563 struct sym *sym;
564 unsigned h;
565 const char *s;
566 struct sym *scope = nested_in_class ? nested_in_class : current_namespace;
567
568 for (s = name, h = 0; *s; ++s)
569 h = (h << 1) ^ *s;
570 h %= TABLE_SIZE;
571
572 for (sym = class_table[h]; sym; sym = sym->next)
573 if (streq (name, sym->name)
574 && ((!sym->namesp && !scope)
575 || (sym->namesp && scope
576 && streq (sym->namesp->name, scope->name))))
577 break;
578
579 if (sym == NULL)
580 {
581 if (f_very_verbose)
582 {
583 putchar ('\t');
584 puts (name);
585 }
586
587 sym = xmalloc (FLEXSIZEOF (struct sym, name, strlen (name) + 1));
588 memset (sym, 0, offsetof (struct sym, name));
589 strcpy (sym->name, name);
590 sym->namesp = scope;
591 sym->next = class_table[h];
592 class_table[h] = sym;
593 }
594
595 return sym;
596 }
597
598
599 /* Add links between superclass SUPER and subclass SUB. */
600
601 static void
add_link(struct sym * super,struct sym * sub)602 add_link (struct sym *super, struct sym *sub)
603 {
604 struct link *lnk, *lnk2, *p, *prev;
605
606 /* See if a link already exists. */
607 for (p = super->subs, prev = NULL;
608 p && strcmp (sub->name, p->sym->name) > 0;
609 prev = p, p = p->next)
610 ;
611
612 /* Avoid duplicates. */
613 if (p == NULL || p->sym != sub)
614 {
615 lnk = (struct link *) xmalloc (sizeof *lnk);
616 lnk2 = (struct link *) xmalloc (sizeof *lnk2);
617
618 lnk->sym = sub;
619 lnk->next = p;
620
621 if (prev)
622 prev->next = lnk;
623 else
624 super->subs = lnk;
625
626 lnk2->sym = super;
627 lnk2->next = sub->supers;
628 sub->supers = lnk2;
629 }
630 }
631
632
633 /* Find in class CLS member NAME.
634
635 VAR non-zero means look for a member variable; otherwise a function
636 is searched. SC specifies what kind of member is searched---a
637 static, or per-instance member etc. HASH is a hash code for the
638 parameter types of functions. Value is a pointer to the member
639 found or null if not found. */
640
641 static struct member *
find_member(struct sym * cls,char * name,int var,int sc,unsigned int hash)642 find_member (struct sym *cls, char *name, int var, int sc, unsigned int hash)
643 {
644 struct member **list;
645 struct member *p;
646 unsigned name_hash = 0;
647 char *s;
648 int i;
649
650 switch (sc)
651 {
652 case SC_FRIEND:
653 list = &cls->friends;
654 break;
655
656 case SC_TYPE:
657 list = &cls->types;
658 break;
659
660 case SC_STATIC:
661 list = var ? &cls->static_vars : &cls->static_fns;
662 break;
663
664 default:
665 list = var ? &cls->vars : &cls->fns;
666 break;
667 }
668
669 for (s = name; *s; ++s)
670 name_hash = (name_hash << 1) ^ *s;
671 i = name_hash % TABLE_SIZE;
672
673 for (p = member_table[i]; p; p = p->anext)
674 if (p->list == list && p->param_hash == hash && streq (name, p->name))
675 break;
676
677 return p;
678 }
679
680
681 /* Add to class CLS information for the declaration of member NAME.
682 REGEXP is a regexp matching the declaration, if non-null. POS is
683 the position in the source where the declaration is found. HASH is
684 a hash code for the parameter list of the member, if it's a
685 function. VAR non-zero means member is a variable or type. SC
686 specifies the type of member (instance member, static, ...). VIS
687 is the member's visibility (public, protected, private). FLAGS is
688 a bit set giving additional information about the member (see the
689 F_* defines). */
690
691 static void
add_member_decl(struct sym * cls,char * name,char * regexp,int pos,unsigned int hash,int var,int sc,int vis,int flags)692 add_member_decl (struct sym *cls, char *name, char *regexp, int pos, unsigned int hash, int var, int sc, int vis, int flags)
693 {
694 struct member *m;
695
696 m = find_member (cls, name, var, sc, hash);
697 if (m == NULL)
698 m = add_member (cls, name, var, sc, hash);
699
700 /* Have we seen a new filename? If so record that. */
701 if (!cls->filename || !filename_eq (cls->filename, filename))
702 m->filename = filename;
703
704 m->regexp = regexp;
705 m->pos = pos;
706 m->flags = flags;
707
708 switch (vis)
709 {
710 case PRIVATE:
711 m->vis = V_PRIVATE;
712 break;
713
714 case PROTECTED:
715 m->vis = V_PROTECTED;
716 break;
717
718 case PUBLIC:
719 m->vis = V_PUBLIC;
720 break;
721 }
722
723 info_where = P_DECL;
724 info_cls = cls;
725 info_member = m;
726 }
727
728
729 /* Add to class CLS information for the definition of member NAME.
730 REGEXP is a regexp matching the declaration, if non-null. POS is
731 the position in the source where the declaration is found. HASH is
732 a hash code for the parameter list of the member, if it's a
733 function. VAR non-zero means member is a variable or type. SC
734 specifies the type of member (instance member, static, ...). VIS
735 is the member's visibility (public, protected, private). FLAGS is
736 a bit set giving additional information about the member (see the
737 F_* defines). */
738
739 static void
add_member_defn(struct sym * cls,char * name,char * regexp,int pos,unsigned int hash,int var,int sc,int flags)740 add_member_defn (struct sym *cls, char *name, char *regexp, int pos, unsigned int hash, int var, int sc, int flags)
741 {
742 struct member *m;
743
744 if (sc == SC_UNKNOWN)
745 {
746 m = find_member (cls, name, var, SC_MEMBER, hash);
747 if (m == NULL)
748 {
749 m = find_member (cls, name, var, SC_STATIC, hash);
750 if (m == NULL)
751 m = add_member (cls, name, var, sc, hash);
752 }
753 }
754 else
755 {
756 m = find_member (cls, name, var, sc, hash);
757 if (m == NULL)
758 m = add_member (cls, name, var, sc, hash);
759 }
760
761 if (!cls->sfilename)
762 cls->sfilename = filename;
763
764 if (!filename_eq (cls->sfilename, filename))
765 m->def_filename = filename;
766
767 m->def_regexp = regexp;
768 m->def_pos = pos;
769 m->flags |= flags;
770
771 info_where = P_DEFN;
772 info_cls = cls;
773 info_member = m;
774 }
775
776
777 /* Add a symbol for a define named NAME to the symbol table.
778 REGEXP is a regular expression matching the define in the source,
779 if it is non-null. POS is the position in the file. */
780
781 static void
add_define(char * name,char * regexp,int pos)782 add_define (char *name, char *regexp, int pos)
783 {
784 add_global_defn (name, regexp, pos, 0, 1, SC_FRIEND, F_DEFINE);
785 add_global_decl (name, regexp, pos, 0, 1, SC_FRIEND, F_DEFINE);
786 }
787
788
789 /* Add information for the global definition of NAME.
790 REGEXP is a regexp matching the declaration, if non-null. POS is
791 the position in the source where the declaration is found. HASH is
792 a hash code for the parameter list of the member, if it's a
793 function. VAR non-zero means member is a variable or type. SC
794 specifies the type of member (instance member, static, ...). VIS
795 is the member's visibility (public, protected, private). FLAGS is
796 a bit set giving additional information about the member (see the
797 F_* defines). */
798
799 static void
add_global_defn(char * name,char * regexp,int pos,unsigned int hash,int var,int sc,int flags)800 add_global_defn (char *name, char *regexp, int pos, unsigned int hash, int var, int sc, int flags)
801 {
802 int i;
803 struct sym *sym;
804
805 /* Try to find out for which classes a function is a friend, and add
806 what we know about it to them. */
807 if (!var)
808 for (i = 0; i < TABLE_SIZE; ++i)
809 for (sym = class_table[i]; sym; sym = sym->next)
810 if (sym != global_symbols && sym->friends)
811 if (find_member (sym, name, 0, SC_FRIEND, hash))
812 add_member_defn (sym, name, regexp, pos, hash, 0,
813 SC_FRIEND, flags);
814
815 /* Add to global symbols. */
816 add_member_defn (global_symbols, name, regexp, pos, hash, var, sc, flags);
817 }
818
819
820 /* Add information for the global declaration of NAME.
821 REGEXP is a regexp matching the declaration, if non-null. POS is
822 the position in the source where the declaration is found. HASH is
823 a hash code for the parameter list of the member, if it's a
824 function. VAR non-zero means member is a variable or type. SC
825 specifies the type of member (instance member, static, ...). VIS
826 is the member's visibility (public, protected, private). FLAGS is
827 a bit set giving additional information about the member (see the
828 F_* defines). */
829
830 static void
add_global_decl(char * name,char * regexp,int pos,unsigned int hash,int var,int sc,int flags)831 add_global_decl (char *name, char *regexp, int pos, unsigned int hash, int var, int sc, int flags)
832 {
833 /* Add declaration only if not already declared. Header files must
834 be processed before source files for this to have the right effect.
835 I do not want to handle implicit declarations at the moment. */
836 struct member *m;
837 struct member *found;
838
839 m = found = find_member (global_symbols, name, var, sc, hash);
840 if (m == NULL)
841 m = add_member (global_symbols, name, var, sc, hash);
842
843 /* Definition already seen => probably last declaration implicit.
844 Override. This means that declarations must always be added to
845 the symbol table before definitions. */
846 if (!found)
847 {
848 if (!global_symbols->filename
849 || !filename_eq (global_symbols->filename, filename))
850 m->filename = filename;
851
852 m->regexp = regexp;
853 m->pos = pos;
854 m->vis = V_PUBLIC;
855 m->flags = flags;
856
857 info_where = P_DECL;
858 info_cls = global_symbols;
859 info_member = m;
860 }
861 }
862
863
864 /* Add a symbol for member NAME to class CLS.
865 VAR non-zero means it's a variable. SC specifies the kind of
866 member. HASH is a hash code for the parameter types of a function.
867 Value is a pointer to the member's structure. */
868
869 static struct member *
add_member(struct sym * cls,char * name,int var,int sc,unsigned int hash)870 add_member (struct sym *cls, char *name, int var, int sc, unsigned int hash)
871 {
872 struct member *m = xmalloc (FLEXSIZEOF (struct member, name,
873 strlen (name) + 1));
874 struct member **list;
875 struct member *p;
876 struct member *prev;
877 unsigned name_hash = 0;
878 int i;
879 char *s;
880
881 strcpy (m->name, name);
882 m->param_hash = hash;
883
884 m->vis = 0;
885 m->flags = 0;
886 m->regexp = NULL;
887 m->filename = NULL;
888 m->pos = 0;
889 m->def_regexp = NULL;
890 m->def_filename = NULL;
891 m->def_pos = 0;
892
893 assert (cls != NULL);
894
895 switch (sc)
896 {
897 case SC_FRIEND:
898 list = &cls->friends;
899 break;
900
901 case SC_TYPE:
902 list = &cls->types;
903 break;
904
905 case SC_STATIC:
906 list = var ? &cls->static_vars : &cls->static_fns;
907 break;
908
909 default:
910 list = var ? &cls->vars : &cls->fns;
911 break;
912 }
913
914 for (s = name; *s; ++s)
915 name_hash = (name_hash << 1) ^ *s;
916 i = name_hash % TABLE_SIZE;
917 m->anext = member_table[i];
918 member_table[i] = m;
919 m->list = list;
920
921 /* Keep the member list sorted. It's cheaper to do it here than to
922 sort them in Lisp. */
923 for (prev = NULL, p = *list;
924 p && strcmp (name, p->name) > 0;
925 prev = p, p = p->next)
926 ;
927
928 m->next = p;
929 if (prev)
930 prev->next = m;
931 else
932 *list = m;
933 return m;
934 }
935
936
937 /* Given the root R of a class tree, step through all subclasses
938 recursively, marking functions as virtual that are declared virtual
939 in base classes. */
940
941 static void
mark_virtual(struct sym * r)942 mark_virtual (struct sym *r)
943 {
944 struct link *p;
945 struct member *m, *m2;
946
947 for (p = r->subs; p; p = p->next)
948 {
949 for (m = r->fns; m; m = m->next)
950 if (has_flag (m->flags, F_VIRTUAL))
951 {
952 for (m2 = p->sym->fns; m2; m2 = m2->next)
953 if (m->param_hash == m2->param_hash && streq (m->name, m2->name))
954 set_flag (&m2->flags, F_VIRTUAL);
955 }
956
957 mark_virtual (p->sym);
958 }
959 }
960
961
962 /* For all roots of the class tree, mark functions as virtual that
963 are virtual because of a virtual declaration in a base class. */
964
965 static void
mark_inherited_virtual(void)966 mark_inherited_virtual (void)
967 {
968 struct sym *r;
969 int i;
970
971 for (i = 0; i < TABLE_SIZE; ++i)
972 for (r = class_table[i]; r; r = r->next)
973 if (r->supers == NULL)
974 mark_virtual (r);
975 }
976
977
978 /* Create and return a symbol for a namespace with name NAME. */
979
980 static struct sym *
make_namespace(char * name,struct sym * context)981 make_namespace (char *name, struct sym *context)
982 {
983 struct sym *s = xmalloc (FLEXSIZEOF (struct sym, name, strlen (name) + 1));
984 memset (s, 0, offsetof (struct sym, name));
985 strcpy (s->name, name);
986 s->next = all_namespaces;
987 s->namesp = context;
988 all_namespaces = s;
989 return s;
990 }
991
992
993 /* Find the symbol for namespace NAME. If not found, return NULL */
994
995 static struct sym *
check_namespace(char * name,struct sym * context)996 check_namespace (char *name, struct sym *context)
997 {
998 struct sym *p = NULL;
999
1000 for (p = all_namespaces; p; p = p->next)
1001 {
1002 if (streq (p->name, name) && (p->namesp == context))
1003 break;
1004 }
1005
1006 return p;
1007 }
1008
1009 /* Find the symbol for namespace NAME. If not found, add a new symbol
1010 for NAME to all_namespaces. */
1011
1012 static struct sym *
find_namespace(char * name,struct sym * context)1013 find_namespace (char *name, struct sym *context)
1014 {
1015 struct sym *p = check_namespace (name, context);
1016
1017 if (p == NULL)
1018 p = make_namespace (name, context);
1019
1020 return p;
1021 }
1022
1023
1024 /* Find namespace alias with name NAME. If not found return NULL. */
1025
1026 static struct link *
check_namespace_alias(char * name)1027 check_namespace_alias (char *name)
1028 {
1029 struct link *p = NULL;
1030 struct alias *al;
1031 unsigned h;
1032 char *s;
1033
1034 for (s = name, h = 0; *s; ++s)
1035 h = (h << 1) ^ *s;
1036 h %= TABLE_SIZE;
1037
1038 for (al = namespace_alias_table[h]; al; al = al->next)
1039 if (streq (name, al->name) && (al->namesp == current_namespace))
1040 {
1041 p = al->aliasee;
1042 break;
1043 }
1044
1045 return p;
1046 }
1047
1048 /* Register the name NEW_NAME as an alias for namespace list OLD_NAME. */
1049
1050 static void
register_namespace_alias(char * new_name,struct link * old_name)1051 register_namespace_alias (char *new_name, struct link *old_name)
1052 {
1053 unsigned h;
1054 char *s;
1055 struct alias *al;
1056
1057 for (s = new_name, h = 0; *s; ++s)
1058 h = (h << 1) ^ *s;
1059 h %= TABLE_SIZE;
1060
1061
1062 /* Is it already in the table of aliases? */
1063 for (al = namespace_alias_table[h]; al; al = al->next)
1064 if (streq (new_name, al->name) && (al->namesp == current_namespace))
1065 return;
1066
1067 al = xmalloc (FLEXSIZEOF (struct alias, name, strlen (new_name) + 1));
1068 strcpy (al->name, new_name);
1069 al->next = namespace_alias_table[h];
1070 al->namesp = current_namespace;
1071 al->aliasee = old_name;
1072 namespace_alias_table[h] = al;
1073 }
1074
1075
1076 /* Enter namespace with name NAME. */
1077
1078 static void
enter_namespace(char * name)1079 enter_namespace (char *name)
1080 {
1081 struct sym *p = find_namespace (name, current_namespace);
1082
1083 if (namespace_sp == namespace_stack_size)
1084 {
1085 int size = max (10, 2 * namespace_stack_size);
1086 namespace_stack
1087 = (struct sym **) xrealloc ((void *)namespace_stack,
1088 size * sizeof *namespace_stack);
1089 namespace_stack_size = size;
1090 }
1091
1092 namespace_stack[namespace_sp++] = current_namespace;
1093 current_namespace = p;
1094 }
1095
1096
1097 /* Leave the current namespace. */
1098
1099 static void
leave_namespace(void)1100 leave_namespace (void)
1101 {
1102 assert (namespace_sp > 0);
1103 current_namespace = namespace_stack[--namespace_sp];
1104 }
1105
1106
1107
1108 /***********************************************************************
1109 Writing the Output File
1110 ***********************************************************************/
1111
1112 /* Write string S to the output file FP in a Lisp-readable form.
1113 If S is null, write out `()'. */
1114
1115 static void
putstr(const char * s,FILE * fp)1116 putstr (const char *s, FILE *fp)
1117 {
1118 if (!s)
1119 {
1120 putc ('(', fp);
1121 putc (')', fp);
1122 putc (' ', fp);
1123 }
1124 else
1125 {
1126 putc ('"', fp);
1127 fputs (s, fp);
1128 putc ('"', fp);
1129 putc (' ', fp);
1130 }
1131 }
1132
1133 /* A dynamically allocated buffer for constructing a scope name. */
1134
1135 static char *scope_buffer;
1136 static int scope_buffer_size;
1137 static int scope_buffer_len;
1138
1139
1140 /* Make sure scope_buffer has enough room to add LEN chars to it. */
1141
1142 static void
ensure_scope_buffer_room(int len)1143 ensure_scope_buffer_room (int len)
1144 {
1145 if (scope_buffer_len + len >= scope_buffer_size)
1146 {
1147 int new_size = max (2 * scope_buffer_size, scope_buffer_len + len);
1148 scope_buffer = (char *) xrealloc (scope_buffer, new_size);
1149 scope_buffer_size = new_size;
1150 }
1151 }
1152
1153
1154 /* Recursively add the scope names of symbol P and the scopes of its
1155 namespaces to scope_buffer. Value is a pointer to the complete
1156 scope name constructed. */
1157
1158 static char *
sym_scope_1(struct sym * p)1159 sym_scope_1 (struct sym *p)
1160 {
1161 int len;
1162
1163 if (p->namesp)
1164 sym_scope_1 (p->namesp);
1165
1166 if (*scope_buffer)
1167 {
1168 ensure_scope_buffer_room (3);
1169 strcpy (scope_buffer + scope_buffer_len, "::");
1170 scope_buffer_len += 2;
1171 }
1172
1173 len = strlen (p->name);
1174 ensure_scope_buffer_room (len + 1);
1175 strcpy (scope_buffer + scope_buffer_len, p->name);
1176 scope_buffer_len += len;
1177
1178 if (has_flag (p->flags, F_TEMPLATE))
1179 {
1180 ensure_scope_buffer_room (3);
1181 strcpy (scope_buffer + scope_buffer_len, "<>");
1182 scope_buffer_len += 2;
1183 }
1184
1185 return scope_buffer;
1186 }
1187
1188
1189 /* Return the scope of symbol P in printed representation, i.e.
1190 as it would appear in a C*+ source file. */
1191
1192 static char *
sym_scope(struct sym * p)1193 sym_scope (struct sym *p)
1194 {
1195 if (!scope_buffer)
1196 {
1197 scope_buffer_size = 1024;
1198 scope_buffer = (char *) xmalloc (scope_buffer_size);
1199 }
1200
1201 *scope_buffer = '\0';
1202 scope_buffer_len = 0;
1203
1204 if (p->namesp)
1205 sym_scope_1 (p->namesp);
1206
1207 return scope_buffer;
1208 }
1209
1210
1211 /* Dump the list of members M to file FP. Value is the length of the
1212 list. */
1213
1214 static int
dump_members(FILE * fp,struct member * m)1215 dump_members (FILE *fp, struct member *m)
1216 {
1217 int n;
1218
1219 putc ('(', fp);
1220
1221 for (n = 0; m; m = m->next, ++n)
1222 {
1223 fputs (MEMBER_STRUCT, fp);
1224 putstr (m->name, fp);
1225 putstr (NULL, fp); /* FIXME? scope for globals */
1226 fprintf (fp, "%u ", (unsigned) m->flags);
1227 putstr (m->filename, fp);
1228 putstr (m->regexp, fp);
1229 fprintf (fp, "%u ", (unsigned) m->pos);
1230 fprintf (fp, "%u ", (unsigned) m->vis);
1231 putc (' ', fp);
1232 putstr (m->def_filename, fp);
1233 putstr (m->def_regexp, fp);
1234 fprintf (fp, "%u", (unsigned) m->def_pos);
1235 putc (']', fp);
1236 putc ('\n', fp);
1237 }
1238
1239 putc (')', fp);
1240 putc ('\n', fp);
1241 return n;
1242 }
1243
1244
1245 /* Dump class ROOT to stream FP. */
1246
1247 static void
dump_sym(FILE * fp,struct sym * root)1248 dump_sym (FILE *fp, struct sym *root)
1249 {
1250 fputs (CLASS_STRUCT, fp);
1251 putstr (root->name, fp);
1252
1253 /* Print scope, if any. */
1254 if (root->namesp)
1255 putstr (sym_scope (root), fp);
1256 else
1257 putstr (NULL, fp);
1258
1259 /* Print flags. */
1260 fprintf (fp, "%d", root->flags);
1261 putstr (root->filename, fp);
1262 putstr (root->regexp, fp);
1263 fprintf (fp, "%u", (unsigned) root->pos);
1264 putstr (root->sfilename, fp);
1265 putc (']', fp);
1266 putc ('\n', fp);
1267 }
1268
1269
1270 /* Dump class ROOT and its subclasses to file FP. Value is the
1271 number of classes written. */
1272
1273 static int
dump_tree(FILE * fp,struct sym * root)1274 dump_tree (FILE *fp, struct sym *root)
1275 {
1276 struct link *lk;
1277 unsigned n = 0;
1278
1279 dump_sym (fp, root);
1280
1281 if (f_verbose)
1282 {
1283 putchar ('+');
1284 fflush (stdout);
1285 }
1286
1287 putc ('(', fp);
1288
1289 for (lk = root->subs; lk; lk = lk->next)
1290 {
1291 fputs (TREE_STRUCT, fp);
1292 n += dump_tree (fp, lk->sym);
1293 putc (']', fp);
1294 }
1295
1296 putc (')', fp);
1297
1298 dump_members (fp, root->vars);
1299 n += dump_members (fp, root->fns);
1300 dump_members (fp, root->static_vars);
1301 n += dump_members (fp, root->static_fns);
1302 n += dump_members (fp, root->friends);
1303 dump_members (fp, root->types);
1304
1305 /* Superclasses. */
1306 putc ('(', fp);
1307 putc (')', fp);
1308
1309 /* Mark slot. */
1310 putc ('(', fp);
1311 putc (')', fp);
1312
1313 putc ('\n', fp);
1314 return n;
1315 }
1316
1317
1318 /* Dump the entire class tree to file FP. */
1319
1320 static void
dump_roots(FILE * fp)1321 dump_roots (FILE *fp)
1322 {
1323 int i, n = 0;
1324 struct sym *r;
1325
1326 /* Output file header containing version string, command line
1327 options etc. */
1328 if (!f_append)
1329 {
1330 fputs (TREE_HEADER_STRUCT, fp);
1331 putstr (EBROWSE_FILE_VERSION, fp);
1332
1333 putc ('\"', fp);
1334 if (!f_structs)
1335 fputs (" -s", fp);
1336 if (f_regexps)
1337 fputs (" -x", fp);
1338 putc ('\"', fp);
1339 fputs (" ()", fp);
1340 fputs (" ()", fp);
1341 putc (']', fp);
1342 }
1343
1344 /* Mark functions as virtual that are so because of functions
1345 declared virtual in base classes. */
1346 mark_inherited_virtual ();
1347
1348 /* Dump the roots of the graph. */
1349 for (i = 0; i < TABLE_SIZE; ++i)
1350 for (r = class_table[i]; r; r = r->next)
1351 if (!r->supers)
1352 {
1353 fputs (TREE_STRUCT, fp);
1354 n += dump_tree (fp, r);
1355 putc (']', fp);
1356 }
1357
1358 if (f_verbose)
1359 putchar ('\n');
1360 }
1361
1362
1363
1364 /***********************************************************************
1365 Scanner
1366 ***********************************************************************/
1367
1368 #ifdef DEBUG
1369 #define INCREMENT_LINENO \
1370 do { \
1371 if (f_very_verbose) \
1372 { \
1373 ++yyline; \
1374 printf ("%d:\n", yyline); \
1375 } \
1376 else \
1377 ++yyline; \
1378 } while (0)
1379 #else
1380 #define INCREMENT_LINENO ++yyline
1381 #endif
1382
1383 /* Define two macros for accessing the input buffer (current input
1384 file). GET(C) sets C to the next input character and advances the
1385 input pointer. UNGET retracts the input pointer. */
1386
1387 #define GET(C) ((C) = *in++)
1388 #define UNGET() (--in)
1389
1390
1391 /* Process a preprocessor line. Value is the next character from the
1392 input buffer not consumed. */
1393
1394 static int
process_pp_line(void)1395 process_pp_line (void)
1396 {
1397 int in_comment = 0, in_string = 0;
1398 int c;
1399 char *p = yytext;
1400
1401 /* Skip over white space. The `#' has been consumed already. */
1402 while (WHITEP (GET (c)))
1403 ;
1404
1405 /* Read the preprocessor command (if any). */
1406 while (IDENTP (c))
1407 {
1408 *p++ = c;
1409 GET (c);
1410 }
1411
1412 /* Is it a `define'? */
1413 *p = '\0';
1414
1415 if (*yytext && streq (yytext, "define"))
1416 {
1417 p = yytext;
1418 while (WHITEP (c))
1419 GET (c);
1420 while (IDENTP (c))
1421 {
1422 *p++ = c;
1423 GET (c);
1424 }
1425
1426 *p = '\0';
1427
1428 if (*yytext)
1429 {
1430 char *regexp = matching_regexp ();
1431 int pos = BUFFER_POS ();
1432 add_define (yytext, regexp, pos);
1433 }
1434 }
1435
1436 while (c && (c != '\n' || in_comment || in_string))
1437 {
1438 if (c == '\\')
1439 GET (c);
1440 else if (c == '/' && !in_comment)
1441 {
1442 if (GET (c) == '*')
1443 in_comment = 1;
1444 }
1445 else if (c == '*' && in_comment)
1446 {
1447 if (GET (c) == '/')
1448 in_comment = 0;
1449 }
1450 else if (c == '"')
1451 in_string = !in_string;
1452
1453 if (c == '\n')
1454 INCREMENT_LINENO;
1455
1456 GET (c);
1457 }
1458
1459 return c;
1460 }
1461
1462
1463 /* Value is the next token from the input buffer. */
1464
1465 static int
yylex(void)1466 yylex (void)
1467 {
1468 int c;
1469 char end_char;
1470 char *p;
1471
1472 for (;;)
1473 {
1474 while (WHITEP (GET (c)))
1475 ;
1476
1477 switch (c)
1478 {
1479 case '\n':
1480 INCREMENT_LINENO;
1481 break;
1482
1483 case '\r':
1484 break;
1485
1486 case 0:
1487 /* End of file. */
1488 return YYEOF;
1489
1490 case '\\':
1491 GET (c);
1492 break;
1493
1494 case '"':
1495 case '\'':
1496 /* String and character constants. */
1497 end_char = c;
1498 string_start = in;
1499 while (GET (c) && c != end_char)
1500 {
1501 switch (c)
1502 {
1503 case '\\':
1504 /* Escape sequences. */
1505 if (!GET (c))
1506 {
1507 if (end_char == '\'')
1508 yyerror ("EOF in character constant", NULL);
1509 else
1510 yyerror ("EOF in string constant", NULL);
1511 goto end_string;
1512 }
1513 else switch (c)
1514 {
1515 case '\n':
1516 INCREMENT_LINENO;
1517 case 'a':
1518 case 'b':
1519 case 'f':
1520 case 'n':
1521 case 'r':
1522 case 't':
1523 case 'v':
1524 break;
1525
1526 case 'x':
1527 {
1528 /* Hexadecimal escape sequence. */
1529 int i;
1530 for (i = 0; i < 2; ++i)
1531 {
1532 GET (c);
1533
1534 if (c >= '0' && c <= '7')
1535 ;
1536 else if (c >= 'a' && c <= 'f')
1537 ;
1538 else if (c >= 'A' && c <= 'F')
1539 ;
1540 else
1541 {
1542 UNGET ();
1543 break;
1544 }
1545 }
1546 }
1547 break;
1548
1549 case '0':
1550 {
1551 /* Octal escape sequence. */
1552 int i;
1553 for (i = 0; i < 3; ++i)
1554 {
1555 GET (c);
1556
1557 if (c >= '0' && c <= '7')
1558 ;
1559 else
1560 {
1561 UNGET ();
1562 break;
1563 }
1564 }
1565 }
1566 break;
1567
1568 default:
1569 break;
1570 }
1571 break;
1572
1573 case '\n':
1574 if (end_char == '\'')
1575 yyerror ("newline in character constant", NULL);
1576 else
1577 yyerror ("newline in string constant", NULL);
1578 INCREMENT_LINENO;
1579 break;
1580
1581 default:
1582 break;
1583 }
1584 }
1585
1586 end_string:
1587 return end_char == '\'' ? CCHAR : CSTRING;
1588
1589 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': case 'g':
1590 case 'h': case 'i': case 'j': case 'k': case 'l': case 'm': case 'n':
1591 case 'o': case 'p': case 'q': case 'r': case 's': case 't': case 'u':
1592 case 'v': case 'w': case 'x': case 'y': case 'z':
1593 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': case 'G':
1594 case 'H': case 'I': case 'J': case 'K': case 'L': case 'M': case 'N':
1595 case 'O': case 'P': case 'Q': case 'R': case 'S': case 'T': case 'U':
1596 case 'V': case 'W': case 'X': case 'Y': case 'Z': case '_':
1597 {
1598 /* Identifier and keywords. */
1599 unsigned hash;
1600 struct kw *k;
1601
1602 p = yytext;
1603 *p++ = hash = c;
1604
1605 while (IDENTP (GET (*p)))
1606 {
1607 hash = (hash << 1) ^ *p++;
1608 if (p == yytext_end - 1)
1609 {
1610 int size = yytext_end - yytext;
1611 yytext = (char *) xrealloc (yytext, 2 * size);
1612 yytext_end = yytext + 2 * size;
1613 p = yytext + size - 1;
1614 }
1615 }
1616
1617 UNGET ();
1618 *p = 0;
1619
1620 for (k = keyword_table[hash % KEYWORD_TABLE_SIZE]; k; k = k->next)
1621 if (streq (k->name, yytext))
1622 return k->tk;
1623
1624 return IDENT;
1625 }
1626
1627 case '/':
1628 /* C and C++ comments, '/' and '/='. */
1629 switch (GET (c))
1630 {
1631 case '*':
1632 while (GET (c))
1633 {
1634 switch (c)
1635 {
1636 case '*':
1637 if (GET (c) == '/')
1638 goto comment_end;
1639 UNGET ();
1640 break;
1641 case '\\':
1642 GET (c);
1643 break;
1644 case '\n':
1645 INCREMENT_LINENO;
1646 break;
1647 }
1648 }
1649 comment_end:;
1650 break;
1651
1652 case '=':
1653 return DIVASGN;
1654
1655 case '/':
1656 while (GET (c) && c != '\n')
1657 ;
1658 /* Don't try to read past the end of the input buffer if
1659 the file ends in a C++ comment without a newline. */
1660 if (c == 0)
1661 return YYEOF;
1662
1663 INCREMENT_LINENO;
1664 break;
1665
1666 default:
1667 UNGET ();
1668 return '/';
1669 }
1670 break;
1671
1672 case '+':
1673 if (GET (c) == '+')
1674 return INC;
1675 else if (c == '=')
1676 return ADDASGN;
1677 UNGET ();
1678 return '+';
1679
1680 case '-':
1681 switch (GET (c))
1682 {
1683 case '-':
1684 return DEC;
1685 case '>':
1686 if (GET (c) == '*')
1687 return ARROWSTAR;
1688 UNGET ();
1689 return ARROW;
1690 case '=':
1691 return SUBASGN;
1692 }
1693 UNGET ();
1694 return '-';
1695
1696 case '*':
1697 if (GET (c) == '=')
1698 return MULASGN;
1699 UNGET ();
1700 return '*';
1701
1702 case '%':
1703 if (GET (c) == '=')
1704 return MODASGN;
1705 UNGET ();
1706 return '%';
1707
1708 case '|':
1709 if (GET (c) == '|')
1710 return LOR;
1711 else if (c == '=')
1712 return ORASGN;
1713 UNGET ();
1714 return '|';
1715
1716 case '&':
1717 if (GET (c) == '&')
1718 return LAND;
1719 else if (c == '=')
1720 return ANDASGN;
1721 UNGET ();
1722 return '&';
1723
1724 case '^':
1725 if (GET (c) == '=')
1726 return XORASGN;
1727 UNGET ();
1728 return '^';
1729
1730 case '.':
1731 if (GET (c) == '*')
1732 return POINTSTAR;
1733 else if (c == '.')
1734 {
1735 if (GET (c) != '.')
1736 yyerror ("invalid token '..' ('...' assumed)", NULL);
1737 UNGET ();
1738 return ELLIPSIS;
1739 }
1740 else if (!DIGITP (c))
1741 {
1742 UNGET ();
1743 return '.';
1744 }
1745 goto mantissa;
1746
1747 case ':':
1748 if (GET (c) == ':')
1749 return DCOLON;
1750 UNGET ();
1751 return ':';
1752
1753 case '=':
1754 if (GET (c) == '=')
1755 return EQ;
1756 UNGET ();
1757 return '=';
1758
1759 case '!':
1760 if (GET (c) == '=')
1761 return NE;
1762 UNGET ();
1763 return '!';
1764
1765 case '<':
1766 switch (GET (c))
1767 {
1768 case '=':
1769 return LE;
1770 case '<':
1771 if (GET (c) == '=')
1772 return LSHIFTASGN;
1773 UNGET ();
1774 return LSHIFT;
1775 }
1776 UNGET ();
1777 return '<';
1778
1779 case '>':
1780 switch (GET (c))
1781 {
1782 case '=':
1783 return GE;
1784 case '>':
1785 if (GET (c) == '=')
1786 return RSHIFTASGN;
1787 UNGET ();
1788 return RSHIFT;
1789 }
1790 UNGET ();
1791 return '>';
1792
1793 case '#':
1794 c = process_pp_line ();
1795 if (c == 0)
1796 return YYEOF;
1797 break;
1798
1799 case '(': case ')': case '[': case ']': case '{': case '}':
1800 case ';': case ',': case '?': case '~':
1801 return c;
1802
1803 case '0':
1804 yyival = 0;
1805
1806 if (GET (c) == 'x' || c == 'X')
1807 {
1808 while (GET (c))
1809 {
1810 if (DIGITP (c))
1811 yyival = yyival * 16 + c - '0';
1812 else if (c >= 'a' && c <= 'f')
1813 yyival = yyival * 16 + c - 'a' + 10;
1814 else if (c >= 'A' && c <= 'F')
1815 yyival = yyival * 16 + c - 'A' + 10;
1816 else
1817 break;
1818 }
1819
1820 goto int_suffixes;
1821 }
1822 else if (c == '.')
1823 goto mantissa;
1824
1825 while (c >= '0' && c <= '7')
1826 {
1827 yyival = (yyival << 3) + c - '0';
1828 GET (c);
1829 }
1830
1831 int_suffixes:
1832 /* Integer suffixes. */
1833 while (isalpha (c))
1834 GET (c);
1835 UNGET ();
1836 return CINT;
1837
1838 case '1': case '2': case '3': case '4': case '5': case '6':
1839 case '7': case '8': case '9':
1840 /* Integer or floating constant, part before '.'. */
1841 yyival = c - '0';
1842
1843 while (GET (c) && DIGITP (c))
1844 yyival = 10 * yyival + c - '0';
1845
1846 if (c != '.')
1847 goto int_suffixes;
1848
1849 mantissa:
1850 /* Digits following '.'. */
1851 while (DIGITP (c))
1852 GET (c);
1853
1854 /* Optional exponent. */
1855 if (c == 'E' || c == 'e')
1856 {
1857 if (GET (c) == '-' || c == '+')
1858 GET (c);
1859
1860 while (DIGITP (c))
1861 GET (c);
1862 }
1863
1864 /* Optional type suffixes. */
1865 while (isalpha (c))
1866 GET (c);
1867 UNGET ();
1868 return CFLOAT;
1869
1870 default:
1871 break;
1872 }
1873 }
1874 }
1875
1876
1877 /* Actually local to matching_regexp. These variables must be in
1878 global scope for the case that `static' gets defined away. */
1879
1880 static char *matching_regexp_buffer, *matching_regexp_end_buf;
1881
1882
1883 /* Value is the string from the start of the line to the current
1884 position in the input buffer, or maybe a bit more if that string is
1885 shorter than min_regexp. */
1886
1887 static char *
matching_regexp(void)1888 matching_regexp (void)
1889 {
1890 char *p;
1891 char *s;
1892 char *t;
1893
1894 if (!f_regexps)
1895 return NULL;
1896
1897 if (matching_regexp_buffer == NULL)
1898 {
1899 matching_regexp_buffer = (char *) xmalloc (max_regexp);
1900 matching_regexp_end_buf = &matching_regexp_buffer[max_regexp] - 1;
1901 }
1902
1903 /* Scan back to previous newline of buffer start. */
1904 for (p = in - 1; p > inbuffer && *p != '\n'; --p)
1905 ;
1906
1907 if (*p == '\n')
1908 {
1909 while (in - p < min_regexp && p > inbuffer)
1910 {
1911 /* Line probably not significant enough */
1912 for (--p; p > inbuffer && *p != '\n'; --p)
1913 ;
1914 }
1915 if (*p == '\n')
1916 ++p;
1917 }
1918
1919 /* Copy from end to make sure significant portions are included.
1920 This implies that in the browser a regular expressing of the form
1921 `^.*{regexp}' has to be used. */
1922 for (s = matching_regexp_end_buf - 1, t = in;
1923 s > matching_regexp_buffer && t > p;)
1924 {
1925 *--s = *--t;
1926
1927 if (*s == '"' || *s == '\\')
1928 *--s = '\\';
1929 }
1930
1931 *(matching_regexp_end_buf - 1) = '\0';
1932 return xstrdup (s);
1933 }
1934
1935
1936 /* Return a printable representation of token T. */
1937
1938 static const char *
token_string(int t)1939 token_string (int t)
1940 {
1941 static char b[3];
1942
1943 switch (t)
1944 {
1945 case CSTRING: return "string constant";
1946 case CCHAR: return "char constant";
1947 case CINT: return "int constant";
1948 case CFLOAT: return "floating constant";
1949 case ELLIPSIS: return "...";
1950 case LSHIFTASGN: return "<<=";
1951 case RSHIFTASGN: return ">>=";
1952 case ARROWSTAR: return "->*";
1953 case IDENT: return "identifier";
1954 case DIVASGN: return "/=";
1955 case INC: return "++";
1956 case ADDASGN: return "+=";
1957 case DEC: return "--";
1958 case ARROW: return "->";
1959 case SUBASGN: return "-=";
1960 case MULASGN: return "*=";
1961 case MODASGN: return "%=";
1962 case LOR: return "||";
1963 case ORASGN: return "|=";
1964 case LAND: return "&&";
1965 case ANDASGN: return "&=";
1966 case XORASGN: return "^=";
1967 case POINTSTAR: return ".*";
1968 case DCOLON: return "::";
1969 case EQ: return "==";
1970 case NE: return "!=";
1971 case LE: return "<=";
1972 case LSHIFT: return "<<";
1973 case GE: return ">=";
1974 case RSHIFT: return ">>";
1975 case ASM: return "asm";
1976 case AUTO: return "auto";
1977 case BREAK: return "break";
1978 case CASE: return "case";
1979 case CATCH: return "catch";
1980 case CHAR: return "char";
1981 case CLASS: return "class";
1982 case CONST: return "const";
1983 case CONTINUE: return "continue";
1984 case DEFAULT: return "default";
1985 case DELETE: return "delete";
1986 case DO: return "do";
1987 case DOUBLE: return "double";
1988 case ELSE: return "else";
1989 case ENUM: return "enum";
1990 case EXTERN: return "extern";
1991 case FLOAT: return "float";
1992 case FOR: return "for";
1993 case FRIEND: return "friend";
1994 case GOTO: return "goto";
1995 case IF: return "if";
1996 case T_INLINE: return "inline";
1997 case INT: return "int";
1998 case LONG: return "long";
1999 case NEW: return "new";
2000 case OPERATOR: return "operator";
2001 case PRIVATE: return "private";
2002 case PROTECTED: return "protected";
2003 case PUBLIC: return "public";
2004 case REGISTER: return "register";
2005 case RETURN: return "return";
2006 case SHORT: return "short";
2007 case SIGNED: return "signed";
2008 case SIZEOF: return "sizeof";
2009 case STATIC: return "static";
2010 case STRUCT: return "struct";
2011 case SWITCH: return "switch";
2012 case TEMPLATE: return "template";
2013 case THIS: return "this";
2014 case THROW: return "throw";
2015 case TRY: return "try";
2016 case TYPEDEF: return "typedef";
2017 case UNION: return "union";
2018 case UNSIGNED: return "unsigned";
2019 case VIRTUAL: return "virtual";
2020 case VOID: return "void";
2021 case VOLATILE: return "volatile";
2022 case WHILE: return "while";
2023 case MUTABLE: return "mutable";
2024 case BOOL: return "bool";
2025 case TRUE: return "true";
2026 case FALSE: return "false";
2027 case SIGNATURE: return "signature";
2028 case NAMESPACE: return "namespace";
2029 case EXPLICIT: return "explicit";
2030 case TYPENAME: return "typename";
2031 case CONST_CAST: return "const_cast";
2032 case DYNAMIC_CAST: return "dynamic_cast";
2033 case REINTERPRET_CAST: return "reinterpret_cast";
2034 case STATIC_CAST: return "static_cast";
2035 case TYPEID: return "typeid";
2036 case USING: return "using";
2037 case WCHAR: return "wchar_t";
2038 case YYEOF: return "EOF";
2039 case FINAL: return "final";
2040
2041 default:
2042 if (t < 255)
2043 {
2044 b[0] = t;
2045 b[1] = '\0';
2046 return b;
2047 }
2048 else
2049 return "???";
2050 }
2051 }
2052
2053
2054 /* Reinitialize the scanner for a new input file. */
2055
2056 static void
re_init_scanner(void)2057 re_init_scanner (void)
2058 {
2059 in = inbuffer;
2060 yyline = 1;
2061
2062 if (yytext == NULL)
2063 {
2064 int size = 256;
2065 yytext = (char *) xmalloc (size * sizeof *yytext);
2066 yytext_end = yytext + size;
2067 }
2068 }
2069
2070
2071 /* Insert a keyword NAME with token value TKV into the keyword hash
2072 table. */
2073
2074 static void
insert_keyword(const char * name,int tkv)2075 insert_keyword (const char *name, int tkv)
2076 {
2077 const char *s;
2078 unsigned h = 0;
2079 struct kw *k = (struct kw *) xmalloc (sizeof *k);
2080
2081 for (s = name; *s; ++s)
2082 h = (h << 1) ^ *s;
2083
2084 h %= KEYWORD_TABLE_SIZE;
2085 k->name = name;
2086 k->tk = tkv;
2087 k->next = keyword_table[h];
2088 keyword_table[h] = k;
2089 }
2090
2091
2092 /* Initialize the scanner for the first file. This sets up the
2093 character class vectors and fills the keyword hash table. */
2094
2095 static void
init_scanner(void)2096 init_scanner (void)
2097 {
2098 int i;
2099
2100 /* Allocate the input buffer */
2101 inbuffer_size = READ_CHUNK_SIZE + 1;
2102 inbuffer = in = (char *) xmalloc (inbuffer_size);
2103 yyline = 1;
2104
2105 /* Set up character class vectors. */
2106 for (i = 0; i < sizeof is_ident; ++i)
2107 {
2108 if (i == '_' || isalnum (i))
2109 is_ident[i] = 1;
2110
2111 if (i >= '0' && i <= '9')
2112 is_digit[i] = 1;
2113
2114 if (i == ' ' || i == '\t' || i == '\f' || i == '\v')
2115 is_white[i] = 1;
2116 }
2117
2118 /* Fill keyword hash table. */
2119 insert_keyword ("and", LAND);
2120 insert_keyword ("and_eq", ANDASGN);
2121 insert_keyword ("asm", ASM);
2122 insert_keyword ("auto", AUTO);
2123 insert_keyword ("bitand", '&');
2124 insert_keyword ("bitor", '|');
2125 insert_keyword ("bool", BOOL);
2126 insert_keyword ("break", BREAK);
2127 insert_keyword ("case", CASE);
2128 insert_keyword ("catch", CATCH);
2129 insert_keyword ("char", CHAR);
2130 insert_keyword ("class", CLASS);
2131 insert_keyword ("compl", '~');
2132 insert_keyword ("const", CONST);
2133 insert_keyword ("const_cast", CONST_CAST);
2134 insert_keyword ("continue", CONTINUE);
2135 insert_keyword ("default", DEFAULT);
2136 insert_keyword ("delete", DELETE);
2137 insert_keyword ("do", DO);
2138 insert_keyword ("double", DOUBLE);
2139 insert_keyword ("dynamic_cast", DYNAMIC_CAST);
2140 insert_keyword ("else", ELSE);
2141 insert_keyword ("enum", ENUM);
2142 insert_keyword ("explicit", EXPLICIT);
2143 insert_keyword ("extern", EXTERN);
2144 insert_keyword ("false", FALSE);
2145 insert_keyword ("final", FINAL);
2146 insert_keyword ("float", FLOAT);
2147 insert_keyword ("for", FOR);
2148 insert_keyword ("friend", FRIEND);
2149 insert_keyword ("goto", GOTO);
2150 insert_keyword ("if", IF);
2151 insert_keyword ("inline", T_INLINE);
2152 insert_keyword ("int", INT);
2153 insert_keyword ("long", LONG);
2154 insert_keyword ("mutable", MUTABLE);
2155 insert_keyword ("namespace", NAMESPACE);
2156 insert_keyword ("new", NEW);
2157 insert_keyword ("not", '!');
2158 insert_keyword ("not_eq", NE);
2159 insert_keyword ("operator", OPERATOR);
2160 insert_keyword ("or", LOR);
2161 insert_keyword ("or_eq", ORASGN);
2162 insert_keyword ("private", PRIVATE);
2163 insert_keyword ("protected", PROTECTED);
2164 insert_keyword ("public", PUBLIC);
2165 insert_keyword ("register", REGISTER);
2166 insert_keyword ("reinterpret_cast", REINTERPRET_CAST);
2167 insert_keyword ("return", RETURN);
2168 insert_keyword ("short", SHORT);
2169 insert_keyword ("signed", SIGNED);
2170 insert_keyword ("sizeof", SIZEOF);
2171 insert_keyword ("static", STATIC);
2172 insert_keyword ("static_cast", STATIC_CAST);
2173 insert_keyword ("struct", STRUCT);
2174 insert_keyword ("switch", SWITCH);
2175 insert_keyword ("template", TEMPLATE);
2176 insert_keyword ("this", THIS);
2177 insert_keyword ("throw", THROW);
2178 insert_keyword ("true", TRUE);
2179 insert_keyword ("try", TRY);
2180 insert_keyword ("typedef", TYPEDEF);
2181 insert_keyword ("typeid", TYPEID);
2182 insert_keyword ("typename", TYPENAME);
2183 insert_keyword ("union", UNION);
2184 insert_keyword ("unsigned", UNSIGNED);
2185 insert_keyword ("using", USING);
2186 insert_keyword ("virtual", VIRTUAL);
2187 insert_keyword ("void", VOID);
2188 insert_keyword ("volatile", VOLATILE);
2189 insert_keyword ("wchar_t", WCHAR);
2190 insert_keyword ("while", WHILE);
2191 insert_keyword ("xor", '^');
2192 insert_keyword ("xor_eq", XORASGN);
2193 }
2194
2195
2196
2197 /***********************************************************************
2198 Parser
2199 ***********************************************************************/
2200
2201 /* Match the current lookahead token and set it to the next token. */
2202
2203 #define MATCH() (tk = yylex ())
2204
2205 /* Return the lookahead token. If current lookahead token is cleared,
2206 read a new token. */
2207
2208 #define LA1 (tk == -1 ? (tk = yylex ()) : tk)
2209
2210 /* Is the current lookahead equal to the token T? */
2211
2212 #define LOOKING_AT(T) (tk == (T))
2213
2214 /* Is the current lookahead one of T1 or T2? */
2215
2216 #define LOOKING_AT2(T1, T2) (tk == (T1) || tk == (T2))
2217
2218 /* Is the current lookahead one of T1, T2 or T3? */
2219
2220 #define LOOKING_AT3(T1, T2, T3) (tk == (T1) || tk == (T2) || tk == (T3))
2221
2222 /* Is the current lookahead one of T1...T4? */
2223
2224 #define LOOKING_AT4(T1, T2, T3, T4) \
2225 (tk == (T1) || tk == (T2) || tk == (T3) || tk == (T4))
2226
2227 /* Match token T if current lookahead is T. */
2228
2229 #define MATCH_IF(T) if (LOOKING_AT (T)) MATCH (); else ((void) 0)
2230
2231 /* Skip to matching token if current token is T. */
2232
2233 #define SKIP_MATCHING_IF(T) \
2234 if (LOOKING_AT (T)) skip_matching (); else ((void) 0)
2235
2236
2237 /* Skip forward until a given token TOKEN or YYEOF is seen and return
2238 the current lookahead token after skipping. */
2239
2240 static int
skip_to(int token)2241 skip_to (int token)
2242 {
2243 while (!LOOKING_AT2 (YYEOF, token))
2244 MATCH ();
2245 return tk;
2246 }
2247
2248 /* Skip over pairs of tokens (parentheses, square brackets,
2249 angle brackets, curly brackets) matching the current lookahead. */
2250
2251 static void
skip_matching(void)2252 skip_matching (void)
2253 {
2254 int open, close, n;
2255
2256 switch (open = LA1)
2257 {
2258 case '{':
2259 close = '}';
2260 break;
2261
2262 case '(':
2263 close = ')';
2264 break;
2265
2266 case '<':
2267 close = '>';
2268 break;
2269
2270 case '[':
2271 close = ']';
2272 break;
2273
2274 default:
2275 abort ();
2276 }
2277
2278 for (n = 0;;)
2279 {
2280 if (LOOKING_AT (open))
2281 ++n;
2282 else if (LOOKING_AT (close))
2283 --n;
2284 else if (LOOKING_AT (YYEOF))
2285 break;
2286
2287 MATCH ();
2288
2289 if (n == 0)
2290 break;
2291 }
2292 }
2293
2294 static void
skip_initializer(void)2295 skip_initializer (void)
2296 {
2297 for (;;)
2298 {
2299 switch (LA1)
2300 {
2301 case ';':
2302 case ',':
2303 case YYEOF:
2304 return;
2305
2306 case '{':
2307 case '[':
2308 case '(':
2309 skip_matching ();
2310 break;
2311
2312 default:
2313 MATCH ();
2314 break;
2315 }
2316 }
2317 }
2318
2319 /* Build qualified namespace alias (A::B::c) and return it. */
2320
2321 static struct link *
match_qualified_namespace_alias(void)2322 match_qualified_namespace_alias (void)
2323 {
2324 struct link *head = NULL;
2325 struct link *cur = NULL;
2326 struct link *tmp = NULL;
2327
2328 for (;;)
2329 {
2330 MATCH ();
2331 switch (LA1)
2332 {
2333 case IDENT:
2334 tmp = (struct link *) xmalloc (sizeof *cur);
2335 tmp->sym = find_namespace (yytext, cur ? cur->sym : NULL);
2336 tmp->next = NULL;
2337 if (head)
2338 {
2339 cur = cur->next = tmp;
2340 }
2341 else
2342 {
2343 head = cur = tmp;
2344 }
2345 break;
2346 case DCOLON:
2347 /* Just skip */
2348 break;
2349 default:
2350 return head;
2351 break;
2352 }
2353 }
2354 }
2355
2356 /* Re-initialize the parser by resetting the lookahead token. */
2357
2358 static void
re_init_parser(void)2359 re_init_parser (void)
2360 {
2361 tk = -1;
2362 }
2363
2364
2365 /* Parse a parameter list, including the const-specifier,
2366 pure-specifier, and throw-list that may follow a parameter list.
2367 Return in FLAGS what was seen following the parameter list.
2368 Returns a hash code for the parameter types. This value is used to
2369 distinguish between overloaded functions. */
2370
2371 static unsigned
parm_list(int * flags)2372 parm_list (int *flags)
2373 {
2374 unsigned hash = 0;
2375 int type_seen = 0;
2376
2377 while (!LOOKING_AT2 (YYEOF, ')'))
2378 {
2379 switch (LA1)
2380 {
2381 /* Skip over grouping parens or parameter lists in parameter
2382 declarations. */
2383 case '(':
2384 skip_matching ();
2385 break;
2386
2387 /* Next parameter. */
2388 case ',':
2389 MATCH ();
2390 type_seen = 0;
2391 break;
2392
2393 /* Ignore the scope part of types, if any. This is because
2394 some types need scopes when defined outside of a class body,
2395 and don't need them inside the class body. This means that
2396 we have to look for the last IDENT in a sequence of
2397 IDENT::IDENT::... */
2398 case IDENT:
2399 if (!type_seen)
2400 {
2401 char *last_id;
2402 unsigned ident_type_hash = 0;
2403
2404 parse_qualified_param_ident_or_type (&last_id);
2405 if (last_id)
2406 {
2407 /* LAST_ID null means something like `X::*'. */
2408 for (; *last_id; ++last_id)
2409 ident_type_hash = (ident_type_hash << 1) ^ *last_id;
2410 hash = (hash << 1) ^ ident_type_hash;
2411 type_seen = 1;
2412 }
2413 }
2414 else
2415 MATCH ();
2416 break;
2417
2418 case VOID:
2419 /* This distinction is made to make `func (void)' equivalent
2420 to `func ()'. */
2421 type_seen = 1;
2422 MATCH ();
2423 if (!LOOKING_AT (')'))
2424 hash = (hash << 1) ^ VOID;
2425 break;
2426
2427 case BOOL: case CHAR: case CLASS: case CONST:
2428 case DOUBLE: case ENUM: case FLOAT: case INT:
2429 case LONG: case SHORT: case SIGNED: case STRUCT:
2430 case UNION: case UNSIGNED: case VOLATILE: case WCHAR:
2431 case ELLIPSIS:
2432 type_seen = 1;
2433 hash = (hash << 1) ^ LA1;
2434 MATCH ();
2435 break;
2436
2437 case '*': case '&': case '[': case ']':
2438 hash = (hash << 1) ^ LA1;
2439 MATCH ();
2440 break;
2441
2442 default:
2443 MATCH ();
2444 break;
2445 }
2446 }
2447
2448 if (LOOKING_AT (')'))
2449 {
2450 MATCH ();
2451
2452 if (LOOKING_AT (CONST))
2453 {
2454 /* We can overload the same function on `const' */
2455 hash = (hash << 1) ^ CONST;
2456 set_flag (flags, F_CONST);
2457 MATCH ();
2458 }
2459
2460 if (LOOKING_AT (THROW))
2461 {
2462 MATCH ();
2463 SKIP_MATCHING_IF ('(');
2464 set_flag (flags, F_THROW);
2465 }
2466
2467 if (LOOKING_AT ('='))
2468 {
2469 MATCH ();
2470 if (LOOKING_AT (CINT) && yyival == 0)
2471 {
2472 MATCH ();
2473 set_flag (flags, F_PURE);
2474 }
2475 }
2476 }
2477
2478 return hash;
2479 }
2480
2481
2482 /* Print position info to stdout. */
2483
2484 static void
print_info(void)2485 print_info (void)
2486 {
2487 if (info_position >= 0 && BUFFER_POS () <= info_position)
2488 if (info_cls)
2489 printf ("(\"%s\" \"%s\" \"%s\" %d)\n",
2490 info_cls->name, sym_scope (info_cls),
2491 info_member->name, info_where);
2492 }
2493
2494
2495 /* Parse a member declaration within the class body of CLS. VIS is
2496 the access specifier for the member (private, protected,
2497 public). */
2498
2499 static void
member(struct sym * cls,int vis)2500 member (struct sym *cls, int vis)
2501 {
2502 char *id = NULL;
2503 int sc = SC_MEMBER;
2504 char *regexp = NULL;
2505 int pos;
2506 int is_constructor;
2507 int flags = 0;
2508 int class_tag;
2509 char *class_name;
2510 int type_seen = 0;
2511 int paren_seen = 0;
2512 unsigned hash = 0;
2513 int tilde = 0;
2514
2515 while (!LOOKING_AT4 (';', '{', '}', YYEOF))
2516 {
2517 switch (LA1)
2518 {
2519 default:
2520 MATCH ();
2521 break;
2522
2523 /* A function or class may follow. */
2524 case TEMPLATE:
2525 MATCH ();
2526 set_flag (&flags, F_TEMPLATE);
2527 /* Skip over template argument list */
2528 SKIP_MATCHING_IF ('<');
2529 break;
2530
2531 case EXPLICIT:
2532 set_flag (&flags, F_EXPLICIT);
2533 goto typeseen;
2534
2535 case MUTABLE:
2536 set_flag (&flags, F_MUTABLE);
2537 goto typeseen;
2538
2539 case T_INLINE:
2540 set_flag (&flags, F_INLINE);
2541 goto typeseen;
2542
2543 case VIRTUAL:
2544 set_flag (&flags, F_VIRTUAL);
2545 goto typeseen;
2546
2547 case '[':
2548 skip_matching ();
2549 break;
2550
2551 case ENUM:
2552 sc = SC_TYPE;
2553 goto typeseen;
2554
2555 case TYPEDEF:
2556 sc = SC_TYPE;
2557 goto typeseen;
2558
2559 case FRIEND:
2560 sc = SC_FRIEND;
2561 goto typeseen;
2562
2563 case STATIC:
2564 sc = SC_STATIC;
2565 goto typeseen;
2566
2567 case '~':
2568 tilde = 1;
2569 MATCH ();
2570 break;
2571
2572 case IDENT:
2573 /* Remember IDENTS seen so far. Among these will be the member
2574 name. */
2575 id = (char *) xrealloc (id, strlen (yytext) + 2);
2576 if (tilde)
2577 {
2578 *id = '~';
2579 strcpy (id + 1, yytext);
2580 }
2581 else
2582 strcpy (id, yytext);
2583 MATCH ();
2584 break;
2585
2586 case OPERATOR:
2587 {
2588 char *s = operator_name (&sc);
2589 id = (char *) xrealloc (id, strlen (s) + 1);
2590 strcpy (id, s);
2591 }
2592 break;
2593
2594 case '(':
2595 /* Most probably the beginning of a parameter list. */
2596 MATCH ();
2597 paren_seen = 1;
2598
2599 if (id && cls)
2600 {
2601 if (!(is_constructor = streq (id, cls->name)))
2602 regexp = matching_regexp ();
2603 }
2604 else
2605 is_constructor = 0;
2606
2607 pos = BUFFER_POS ();
2608 hash = parm_list (&flags);
2609
2610 if (is_constructor)
2611 regexp = matching_regexp ();
2612
2613 if (id && cls != NULL)
2614 add_member_decl (cls, id, regexp, pos, hash, 0, sc, vis, flags);
2615
2616 while (!LOOKING_AT3 (';', '{', YYEOF))
2617 MATCH ();
2618
2619 if (LOOKING_AT ('{') && id && cls)
2620 add_member_defn (cls, id, regexp, pos, hash, 0, sc, flags);
2621
2622 free (id);
2623 id = NULL;
2624 sc = SC_MEMBER;
2625 break;
2626
2627 case STRUCT: case UNION: case CLASS:
2628 /* Nested class */
2629 class_tag = LA1;
2630 type_seen = 1;
2631 MATCH ();
2632 class_name = NULL;
2633
2634 /* More than one ident here to allow for MS-DOS specialties
2635 like `_export class' etc. The last IDENT seen counts
2636 as the class name. */
2637 while (!LOOKING_AT4 (YYEOF, ';', ':', '{'))
2638 {
2639 if (LOOKING_AT (IDENT))
2640 {
2641 if (class_name)
2642 {
2643 int size = strlen (yytext);
2644
2645 if(strlen (class_name) < size)
2646 {
2647 class_name = (char *) xrealloc(class_name, size + 1);
2648 }
2649
2650 memcpy(class_name, yytext, size + 1);
2651 }
2652 else
2653 {
2654 class_name = xstrdup(yytext);
2655 }
2656 }
2657
2658 MATCH ();
2659 }
2660
2661 if (LOOKING_AT2 (':', '{'))
2662 class_definition (class_name ? cls : NULL, class_name ? class_name : yytext, class_tag, flags, 1);
2663 else
2664 skip_to (';');
2665
2666 free(class_name);
2667 break;
2668
2669 case INT: case CHAR: case LONG: case UNSIGNED:
2670 case SIGNED: case CONST: case DOUBLE: case VOID:
2671 case SHORT: case VOLATILE: case BOOL: case WCHAR:
2672 case TYPENAME:
2673 typeseen:
2674 type_seen = 1;
2675 MATCH ();
2676 break;
2677 }
2678 }
2679
2680 if (LOOKING_AT (';'))
2681 {
2682 /* The end of a member variable, a friend declaration or an access
2683 declaration. We don't want to add friend classes as members. */
2684 if (id && sc != SC_FRIEND && cls)
2685 {
2686 regexp = matching_regexp ();
2687 pos = BUFFER_POS ();
2688
2689 if (cls != NULL)
2690 {
2691 if (type_seen || !paren_seen)
2692 add_member_decl (cls, id, regexp, pos, 0, 1, sc, vis, 0);
2693 else
2694 add_member_decl (cls, id, regexp, pos, hash, 0, sc, vis, 0);
2695 }
2696 }
2697
2698 MATCH ();
2699 print_info ();
2700 }
2701 else if (LOOKING_AT ('{'))
2702 {
2703 /* A named enum. */
2704 if (sc == SC_TYPE && id && cls)
2705 {
2706 regexp = matching_regexp ();
2707 pos = BUFFER_POS ();
2708
2709 if (cls != NULL)
2710 {
2711 add_member_decl (cls, id, regexp, pos, 0, 1, sc, vis, 0);
2712 add_member_defn (cls, id, regexp, pos, 0, 1, sc, 0);
2713 }
2714 }
2715
2716 skip_matching ();
2717 print_info ();
2718 }
2719
2720 free (id);
2721 }
2722
2723
2724 /* Parse the body of class CLS. TAG is the tag of the class (struct,
2725 union, class). */
2726
2727 static void
class_body(struct sym * cls,int tag)2728 class_body (struct sym *cls, int tag)
2729 {
2730 int vis = tag == CLASS ? PRIVATE : PUBLIC;
2731 int temp;
2732
2733 while (!LOOKING_AT2 (YYEOF, '}'))
2734 {
2735 switch (LA1)
2736 {
2737 case PRIVATE: case PROTECTED: case PUBLIC:
2738 temp = LA1;
2739 MATCH ();
2740
2741 if (LOOKING_AT (':'))
2742 {
2743 vis = temp;
2744 MATCH ();
2745 }
2746 else
2747 {
2748 /* Probably conditional compilation for inheritance list.
2749 We don't known whether there comes more of this.
2750 This is only a crude fix that works most of the time. */
2751 do
2752 {
2753 MATCH ();
2754 }
2755 while (LOOKING_AT2 (IDENT, ',')
2756 || LOOKING_AT3 (PUBLIC, PROTECTED, PRIVATE));
2757 }
2758 break;
2759
2760 case TYPENAME:
2761 case USING:
2762 skip_to (';');
2763 break;
2764
2765 /* Try to synchronize */
2766 case CHAR: case CLASS: case CONST:
2767 case DOUBLE: case ENUM: case FLOAT: case INT:
2768 case LONG: case SHORT: case SIGNED: case STRUCT:
2769 case UNION: case UNSIGNED: case VOID: case VOLATILE:
2770 case TYPEDEF: case STATIC: case T_INLINE: case FRIEND:
2771 case VIRTUAL: case TEMPLATE: case IDENT: case '~':
2772 case BOOL: case WCHAR: case EXPLICIT: case MUTABLE:
2773 member (cls, vis);
2774 break;
2775
2776 default:
2777 MATCH ();
2778 break;
2779 }
2780 }
2781 }
2782
2783
2784 /* Parse a qualified identifier. Current lookahead is IDENT. A
2785 qualified ident has the form `X<..>::Y<...>::T<...>. Returns a
2786 symbol for that class. */
2787
2788 static struct sym *
parse_classname(void)2789 parse_classname (void)
2790 {
2791 struct sym *last_class = NULL;
2792
2793 while (LOOKING_AT (IDENT))
2794 {
2795 last_class = add_sym (yytext, last_class);
2796 MATCH ();
2797
2798 if (LOOKING_AT ('<'))
2799 {
2800 skip_matching ();
2801 set_flag (&last_class->flags, F_TEMPLATE);
2802 }
2803
2804 if (!LOOKING_AT (DCOLON))
2805 break;
2806
2807 MATCH ();
2808 }
2809
2810 return last_class;
2811 }
2812
2813
2814 /* Parse an operator name. Add the `static' flag to *SC if an
2815 implicitly static operator has been parsed. Value is a pointer to
2816 a static buffer holding the constructed operator name string. */
2817
2818 static char *
operator_name(int * sc)2819 operator_name (int *sc)
2820 {
2821 static size_t id_size = 0;
2822 static char *id = NULL;
2823 const char *s;
2824 size_t len;
2825
2826 MATCH ();
2827
2828 if (LOOKING_AT2 (NEW, DELETE))
2829 {
2830 /* `new' and `delete' are implicitly static. */
2831 if (*sc != SC_FRIEND)
2832 *sc = SC_STATIC;
2833
2834 s = token_string (LA1);
2835 MATCH ();
2836
2837 ptrdiff_t slen = strlen (s);
2838 len = slen + 10;
2839 if (len > id_size)
2840 {
2841 size_t new_size = max (len, 2 * id_size);
2842 id = (char *) xrealloc (id, new_size);
2843 id_size = new_size;
2844 }
2845 char *z = stpcpy (id, s);
2846
2847 /* Vector new or delete? */
2848 if (LOOKING_AT ('['))
2849 {
2850 z = stpcpy (z, "[");
2851 MATCH ();
2852
2853 if (LOOKING_AT (']'))
2854 {
2855 strcpy (z, "]");
2856 MATCH ();
2857 }
2858 }
2859 }
2860 else
2861 {
2862 size_t tokens_matched = 0;
2863
2864 len = 20;
2865 if (len > id_size)
2866 {
2867 int new_size = max (len, 2 * id_size);
2868 id = (char *) xrealloc (id, new_size);
2869 id_size = new_size;
2870 }
2871 char *z = stpcpy (id, "operator");
2872
2873 /* Beware access declarations of the form "X::f;" Beware of
2874 `operator () ()'. Yet another difficulty is found in
2875 GCC 2.95's STL: `operator == __STL_NULL_TMPL_ARGS (...'. */
2876 while (!(LOOKING_AT ('(') && tokens_matched)
2877 && !LOOKING_AT2 (';', YYEOF))
2878 {
2879 s = token_string (LA1);
2880 len += strlen (s) + 2;
2881 if (len > id_size)
2882 {
2883 ptrdiff_t idlen = z - id;
2884 size_t new_size = max (len, 2 * id_size);
2885 id = (char *) xrealloc (id, new_size);
2886 id_size = new_size;
2887 z = id + idlen;
2888 }
2889
2890 if (*s != ')' && *s != ']')
2891 *z++ = ' ';
2892 z = stpcpy (z, s);
2893 MATCH ();
2894
2895 /* If this is a simple operator like `+', stop now. */
2896 if (!isalpha ((unsigned char) *s) && *s != '(' && *s != '[')
2897 break;
2898
2899 ++tokens_matched;
2900 }
2901 }
2902
2903 return id;
2904 }
2905
2906
2907 /* This one consumes the last IDENT of a qualified member name like
2908 `X::Y::z'. This IDENT is returned in LAST_ID. Value is the
2909 symbol structure for the ident. */
2910
2911 static struct sym *
parse_qualified_ident_or_type(char ** last_id)2912 parse_qualified_ident_or_type (char **last_id)
2913 {
2914 struct sym *cls = NULL;
2915 char *id = NULL;
2916 size_t id_size = 0;
2917 int enter = 0;
2918
2919 while (LOOKING_AT (IDENT))
2920 {
2921 int len = strlen (yytext) + 1;
2922 if (len > id_size)
2923 {
2924 id = (char *) xrealloc (id, len);
2925 id_size = len;
2926 }
2927 strcpy (id, yytext);
2928 *last_id = id;
2929 MATCH ();
2930
2931 SKIP_MATCHING_IF ('<');
2932
2933 if (LOOKING_AT (DCOLON))
2934 {
2935 struct sym *pcn = NULL;
2936 struct link *pna = check_namespace_alias (id);
2937 if (pna)
2938 {
2939 do
2940 {
2941 enter_namespace (pna->sym->name);
2942 enter++;
2943 pna = pna->next;
2944 }
2945 while (pna);
2946 }
2947 else if ((pcn = check_namespace (id, current_namespace)))
2948 {
2949 enter_namespace (pcn->name);
2950 enter++;
2951 }
2952 else
2953 cls = add_sym (id, cls);
2954
2955 *last_id = NULL;
2956 free (id);
2957 id = NULL;
2958 id_size = 0;
2959 MATCH ();
2960 }
2961 else
2962 break;
2963 }
2964
2965 while (enter--)
2966 leave_namespace ();
2967
2968 return cls;
2969 }
2970
2971
2972 /* This one consumes the last IDENT of a qualified member name like
2973 `X::Y::z'. This IDENT is returned in LAST_ID. Value is the
2974 symbol structure for the ident. */
2975
2976 static void
parse_qualified_param_ident_or_type(char ** last_id)2977 parse_qualified_param_ident_or_type (char **last_id)
2978 {
2979 struct sym *cls = NULL;
2980 static char *id = NULL;
2981 static int id_size = 0;
2982
2983 assert (LOOKING_AT (IDENT));
2984
2985 do
2986 {
2987 int len = strlen (yytext) + 1;
2988 if (len > id_size)
2989 {
2990 id = (char *) xrealloc (id, len);
2991 id_size = len;
2992 }
2993 strcpy (id, yytext);
2994 *last_id = id;
2995 MATCH ();
2996
2997 SKIP_MATCHING_IF ('<');
2998
2999 if (LOOKING_AT (DCOLON))
3000 {
3001 cls = add_sym (id, cls);
3002 *last_id = NULL;
3003 MATCH ();
3004 }
3005 else
3006 break;
3007 }
3008 while (LOOKING_AT (IDENT));
3009 }
3010
3011
3012 /* Parse a class definition.
3013
3014 CONTAINING is the class containing the class being parsed or null.
3015 This may also be null if NESTED != 0 if the containing class is
3016 anonymous. TAG is the tag of the class (struct, union, class).
3017 NESTED is non-zero if we are parsing a nested class.
3018
3019 Current lookahead is the class name. */
3020
3021 static void
class_definition(struct sym * containing,const char * class_name,int tag,int flags,int nested)3022 class_definition (struct sym *containing, const char *class_name, int tag, int flags, int nested)
3023 {
3024 struct sym *current;
3025 struct sym *base_class;
3026
3027 /* Set CURRENT to null if no entry has to be made for the class
3028 parsed. This is the case for certain command line flag
3029 settings. */
3030 if ((tag != CLASS && !f_structs) || (nested && !f_nested_classes))
3031 current = NULL;
3032 else
3033 {
3034 current = add_sym (class_name, containing);
3035 current->pos = BUFFER_POS ();
3036 current->regexp = matching_regexp ();
3037 current->filename = filename;
3038 current->flags = flags;
3039 }
3040
3041 /* If at ':', base class list follows. */
3042 if (LOOKING_AT (':'))
3043 {
3044 int done = 0;
3045 MATCH ();
3046
3047 while (!done)
3048 {
3049 switch (LA1)
3050 {
3051 case VIRTUAL: case PUBLIC: case PROTECTED: case PRIVATE:
3052 MATCH ();
3053 break;
3054
3055 case IDENT:
3056 base_class = parse_classname ();
3057 if (base_class && current && base_class != current)
3058 add_link (base_class, current);
3059 break;
3060
3061 /* The `,' between base classes or the end of the base
3062 class list. Add the previously found base class.
3063 It's done this way to skip over sequences of
3064 `A::B::C' until we reach the end.
3065
3066 FIXME: it is now possible to handle `class X : public B::X'
3067 because we have enough information. */
3068 case ',':
3069 MATCH ();
3070 break;
3071
3072 default:
3073 /* A syntax error, possibly due to preprocessor constructs
3074 like
3075
3076 #ifdef SOMETHING
3077 class A : public B
3078 #else
3079 class A : private B.
3080
3081 MATCH until we see something like `;' or `{'. */
3082 while (!LOOKING_AT3 (';', YYEOF, '{'))
3083 MATCH ();
3084 FALLTHROUGH;
3085 case '{':
3086 done = 1;
3087 break;
3088 }
3089 }
3090 }
3091
3092 /* Parse the class body if there is one. */
3093 if (LOOKING_AT ('{'))
3094 {
3095 if (tag != CLASS && !f_structs)
3096 skip_matching ();
3097 else
3098 {
3099 MATCH ();
3100 class_body (current, tag);
3101
3102 if (LOOKING_AT ('}'))
3103 {
3104 MATCH ();
3105 if (LOOKING_AT (';') && !nested)
3106 MATCH ();
3107 }
3108 }
3109 }
3110 }
3111
3112 /* Add to class *CLS information for the declaration of variable or
3113 type *ID. If *CLS is null, this means a global declaration. SC is
3114 the storage class of *ID. FLAGS is a bit set giving additional
3115 information about the member (see the F_* defines). */
3116
3117 static void
add_declarator(struct sym ** cls,char ** id,int flags,int sc)3118 add_declarator (struct sym **cls, char **id, int flags, int sc)
3119 {
3120 if (LOOKING_AT2 (';', ','))
3121 {
3122 /* The end of a member variable or of an access declaration
3123 `X::f'. To distinguish between them we have to know whether
3124 type information has been seen. */
3125 if (*id)
3126 {
3127 char *regexp = matching_regexp ();
3128 int pos = BUFFER_POS ();
3129
3130 if (*cls)
3131 add_member_defn (*cls, *id, regexp, pos, 0, 1, SC_UNKNOWN, flags);
3132 else
3133 add_global_defn (*id, regexp, pos, 0, 1, sc, flags);
3134 }
3135
3136 MATCH ();
3137 print_info ();
3138 }
3139 else if (LOOKING_AT ('{'))
3140 {
3141 if (sc == SC_TYPE && *id)
3142 {
3143 /* A named enumeration. */
3144 char *regexp = matching_regexp ();
3145 int pos = BUFFER_POS ();
3146 add_global_defn (*id, regexp, pos, 0, 1, sc, flags);
3147 }
3148
3149 skip_matching ();
3150 print_info ();
3151 }
3152
3153 free (*id);
3154 *id = NULL;
3155 *cls = NULL;
3156 }
3157
3158 /* Parse a declaration. */
3159
3160 static void
declaration(int flags)3161 declaration (int flags)
3162 {
3163 char *id = NULL;
3164 struct sym *cls = NULL;
3165 char *regexp = NULL;
3166 int pos = 0;
3167 unsigned hash = 0;
3168 int is_constructor;
3169 int sc = 0;
3170
3171 while (!LOOKING_AT3 (';', '{', YYEOF))
3172 {
3173 switch (LA1)
3174 {
3175 default:
3176 MATCH ();
3177 break;
3178
3179 case '[':
3180 skip_matching ();
3181 break;
3182
3183 case ENUM:
3184 case TYPEDEF:
3185 sc = SC_TYPE;
3186 MATCH ();
3187 break;
3188
3189 case STATIC:
3190 sc = SC_STATIC;
3191 MATCH ();
3192 break;
3193
3194 case INT: case CHAR: case LONG: case UNSIGNED:
3195 case SIGNED: case CONST: case DOUBLE: case VOID:
3196 case SHORT: case VOLATILE: case BOOL: case WCHAR:
3197 MATCH ();
3198 break;
3199
3200 case CLASS: case STRUCT: case UNION:
3201 /* This is for the case `STARTWRAP class X : ...' or
3202 `declare (X, Y)\n class A : ...'. */
3203 if (id)
3204 {
3205 free (id);
3206 return;
3207 }
3208 FALLTHROUGH;
3209 case '=':
3210 /* Assumed to be the start of an initialization in this
3211 context. */
3212 skip_initializer ();
3213 break;
3214
3215 case ',':
3216 add_declarator (&cls, &id, flags, sc);
3217 break;
3218
3219 case OPERATOR:
3220 {
3221 char *s = operator_name (&sc);
3222 id = (char *) xrealloc (id, strlen (s) + 1);
3223 strcpy (id, s);
3224 }
3225 break;
3226
3227 case T_INLINE:
3228 set_flag (&flags, F_INLINE);
3229 MATCH ();
3230 break;
3231
3232 case '~':
3233 MATCH ();
3234 if (LOOKING_AT (IDENT))
3235 {
3236 id = (char *) xrealloc (id, strlen (yytext) + 2);
3237 *id = '~';
3238 strcpy (id + 1, yytext);
3239 MATCH ();
3240 }
3241 break;
3242
3243 case IDENT:
3244 cls = parse_qualified_ident_or_type (&id);
3245 break;
3246
3247 case '(':
3248 /* Most probably the beginning of a parameter list. */
3249 if (cls)
3250 {
3251 MATCH ();
3252
3253 if (id && cls)
3254 {
3255 if (!(is_constructor = streq (id, cls->name)))
3256 regexp = matching_regexp ();
3257 }
3258 else
3259 is_constructor = 0;
3260
3261 pos = BUFFER_POS ();
3262 hash = parm_list (&flags);
3263
3264 if (is_constructor)
3265 regexp = matching_regexp ();
3266
3267 if (id && cls)
3268 add_member_defn (cls, id, regexp, pos, hash, 0,
3269 SC_UNKNOWN, flags);
3270 }
3271 else
3272 {
3273 /* This may be a C functions, but also a macro
3274 call of the form `declare (A, B)' --- such macros
3275 can be found in some class libraries. */
3276 MATCH ();
3277
3278 if (id)
3279 {
3280 regexp = matching_regexp ();
3281 pos = BUFFER_POS ();
3282 hash = parm_list (&flags);
3283 add_global_decl (id, regexp, pos, hash, 0, sc, flags);
3284 }
3285
3286 /* This is for the case that the function really is
3287 a macro with no `;' following it. If a CLASS directly
3288 follows, we would miss it otherwise. */
3289 if (LOOKING_AT3 (CLASS, STRUCT, UNION))
3290 return;
3291 }
3292
3293 while (!LOOKING_AT3 (';', '{', YYEOF))
3294 MATCH ();
3295
3296 if (!cls && id && LOOKING_AT ('{'))
3297 add_global_defn (id, regexp, pos, hash, 0, sc, flags);
3298
3299 free (id);
3300 id = NULL;
3301 break;
3302 }
3303 }
3304
3305 add_declarator (&cls, &id, flags, sc);
3306 }
3307
3308
3309 /* Parse a list of top-level declarations/definitions. START_FLAGS
3310 says in which context we are parsing. If it is F_EXTERNC, we are
3311 parsing in an `extern "C"' block. Value is 1 if EOF is reached, 0
3312 otherwise. */
3313
3314 static int
globals(int start_flags)3315 globals (int start_flags)
3316 {
3317 int class_tk;
3318 char *class_name;
3319 int flags = start_flags;
3320
3321 for (;;)
3322 {
3323 char *prev_in = in;
3324
3325 switch (LA1)
3326 {
3327 case NAMESPACE:
3328 {
3329 MATCH ();
3330
3331 if (LOOKING_AT (IDENT))
3332 {
3333 char *namespace_name = xstrdup (yytext);
3334 MATCH ();
3335
3336 if (LOOKING_AT ('='))
3337 {
3338 struct link *qna = match_qualified_namespace_alias ();
3339 if (qna)
3340 register_namespace_alias (namespace_name, qna);
3341
3342 if (skip_to (';') == ';')
3343 MATCH ();
3344 }
3345 else if (LOOKING_AT ('{'))
3346 {
3347 MATCH ();
3348 enter_namespace (namespace_name);
3349 globals (0);
3350 leave_namespace ();
3351 MATCH_IF ('}');
3352 }
3353
3354 free (namespace_name);
3355 }
3356 }
3357 break;
3358
3359 case EXTERN:
3360 MATCH ();
3361 if (LOOKING_AT (CSTRING) && *string_start == 'C'
3362 && *(string_start + 1) == '"')
3363 {
3364 /* This is `extern "C"'. */
3365 MATCH ();
3366
3367 if (LOOKING_AT ('{'))
3368 {
3369 MATCH ();
3370 globals (F_EXTERNC);
3371 MATCH_IF ('}');
3372 }
3373 else
3374 set_flag (&flags, F_EXTERNC);
3375 }
3376 break;
3377
3378 case TEMPLATE:
3379 MATCH ();
3380 SKIP_MATCHING_IF ('<');
3381 set_flag (&flags, F_TEMPLATE);
3382 break;
3383
3384 case CLASS: case STRUCT: case UNION:
3385 class_tk = LA1;
3386 MATCH ();
3387 class_name = NULL;
3388
3389 /* More than one ident here to allow for MS-DOS and OS/2
3390 specialties like `far', `_Export' etc. Some C++ libs
3391 have constructs like `_OS_DLLIMPORT(_OS_CLIENT)' in front
3392 of the class name. */
3393 while (!LOOKING_AT4 (YYEOF, ';', ':', '{'))
3394 {
3395 if (LOOKING_AT (IDENT))
3396 {
3397 if (class_name)
3398 {
3399 int size = strlen (yytext);
3400
3401 if(strlen (class_name) < size)
3402 {
3403 class_name = (char *) xrealloc(class_name, size + 1);
3404 }
3405
3406 memcpy(class_name, yytext, size + 1);
3407 }
3408 else
3409 {
3410 class_name = xstrdup(yytext);
3411 }
3412 }
3413
3414 MATCH ();
3415 }
3416
3417 /* Don't add anonymous unions. */
3418 if (LOOKING_AT2 (':', '{') && class_name)
3419 class_definition (NULL, class_name, class_tk, flags, 0);
3420 else
3421 {
3422 if (skip_to (';') == ';')
3423 MATCH ();
3424 }
3425
3426 free(class_name);
3427 flags = start_flags;
3428 break;
3429
3430 case YYEOF:
3431 return 1;
3432
3433 case '}':
3434 return 0;
3435
3436 default:
3437 declaration (flags);
3438 flags = start_flags;
3439 break;
3440 }
3441
3442 if (prev_in == in)
3443 yyerror ("parse error", NULL);
3444 }
3445 }
3446
3447
3448 /* Parse the current input file. */
3449
3450 static void
yyparse(void)3451 yyparse (void)
3452 {
3453 while (globals (0) == 0)
3454 MATCH_IF ('}');
3455 }
3456
3457
3458
3459 /***********************************************************************
3460 Main Program
3461 ***********************************************************************/
3462
3463 /* Add the list of paths PATH_LIST to the current search path for
3464 input files. */
3465
3466 static void
add_search_path(char * path_list)3467 add_search_path (char *path_list)
3468 {
3469 while (*path_list)
3470 {
3471 char *start = path_list;
3472 struct search_path *p;
3473
3474 while (*path_list && *path_list != SEPCHAR)
3475 ++path_list;
3476
3477 p = (struct search_path *) xmalloc (sizeof *p);
3478 p->path = (char *) xmalloc (path_list - start + 1);
3479 memcpy (p->path, start, path_list - start);
3480 p->path[path_list - start] = '\0';
3481 p->next = NULL;
3482
3483 if (search_path_tail)
3484 {
3485 search_path_tail->next = p;
3486 search_path_tail = p;
3487 }
3488 else
3489 search_path = search_path_tail = p;
3490
3491 while (*path_list == SEPCHAR)
3492 ++path_list;
3493 }
3494 }
3495
3496
3497 /* Open FILE and return a file handle for it, or -1 if FILE cannot be
3498 opened. Try to find FILE in search_path first, then try the
3499 unchanged file name. */
3500
3501 static FILE *
open_file(char * file)3502 open_file (char *file)
3503 {
3504 FILE *fp = NULL;
3505 static char *buffer;
3506 static int buffer_size;
3507 struct search_path *path;
3508 int flen = strlen (file) + 1; /* +1 for the slash */
3509
3510 filename = xstrdup (file);
3511
3512 for (path = search_path; path && fp == NULL; path = path->next)
3513 {
3514 int len = strlen (path->path) + flen;
3515
3516 if (len + 1 >= buffer_size)
3517 {
3518 buffer_size = max (len + 1, 2 * buffer_size);
3519 buffer = (char *) xrealloc (buffer, buffer_size);
3520 }
3521
3522 char *z = stpcpy (buffer, path->path);
3523 *z++ = '/';
3524 strcpy (z, file);
3525 fp = fopen (buffer, "r");
3526 }
3527
3528 /* Try the original file name. */
3529 if (fp == NULL)
3530 fp = fopen (file, "r");
3531
3532 if (fp == NULL)
3533 yyerror ("cannot open", NULL);
3534
3535 return fp;
3536 }
3537
3538
3539 /* Display usage information and exit program. */
3540
3541 static char const *const usage_message[] =
3542 {
3543 "\
3544 Usage: ebrowse [options] {files}\n\
3545 \n\
3546 -a, --append append output to existing file\n\
3547 -f, --files=FILES read input file names from FILE\n\
3548 -I, --search-path=LIST set search path for input files\n\
3549 -m, --min-regexp-length=N set minimum regexp length to N\n\
3550 -M, --max-regexp-length=N set maximum regexp length to N\n\
3551 ",
3552 "\
3553 -n, --no-nested-classes exclude nested classes\n\
3554 -o, --output-file=FILE set output file name to FILE\n\
3555 -p, --position-info print info about position in file\n\
3556 -s, --no-structs-or-unions don't record structs or unions\n\
3557 -v, --verbose be verbose\n\
3558 -V, --very-verbose be very verbose\n\
3559 -x, --no-regexps don't record regular expressions\n\
3560 --help display this help\n\
3561 --version display version info\n\
3562 \n\
3563 "
3564 };
3565
3566 static _Noreturn void
usage(int error)3567 usage (int error)
3568 {
3569 int i;
3570 for (i = 0; i < sizeof usage_message / sizeof *usage_message; i++)
3571 fputs (usage_message[i], stdout);
3572 exit (error ? EXIT_FAILURE : EXIT_SUCCESS);
3573 }
3574
3575
3576 /* Display version and copyright info. */
3577
3578 static _Noreturn void
version(void)3579 version (void)
3580 {
3581 fputs (("ebrowse " PACKAGE_VERSION "\n"
3582 COPYRIGHT "\n"
3583 "This program is distributed under the same terms as Emacs.\n"),
3584 stdout);
3585 exit (EXIT_SUCCESS);
3586 }
3587
3588
3589 /* Parse one input file FILE, adding classes and members to the symbol
3590 table. */
3591
3592 static void
process_file(char * file)3593 process_file (char *file)
3594 {
3595 FILE *fp;
3596
3597 fp = open_file (file);
3598 if (fp)
3599 {
3600 size_t nread, nbytes;
3601
3602 /* Give a progress indication if needed. */
3603 if (f_very_verbose)
3604 {
3605 puts (filename);
3606 fflush (stdout);
3607 }
3608 else if (f_verbose)
3609 {
3610 putchar ('.');
3611 fflush (stdout);
3612 }
3613
3614 /* Read file to inbuffer. */
3615 for (nread = 0;;)
3616 {
3617 if (nread + READ_CHUNK_SIZE >= inbuffer_size)
3618 {
3619 inbuffer_size = nread + READ_CHUNK_SIZE + 1;
3620 inbuffer = (char *) xrealloc (inbuffer, inbuffer_size);
3621 }
3622
3623 nbytes = fread (inbuffer + nread, 1, READ_CHUNK_SIZE, fp);
3624 if (nbytes == 0)
3625 break;
3626 nread += nbytes;
3627 }
3628 inbuffer[nread] = '\0';
3629
3630 /* Reinitialize scanner and parser for the new input file. */
3631 re_init_scanner ();
3632 re_init_parser ();
3633
3634 /* Parse it and close the file. */
3635 yyparse ();
3636 fclose (fp);
3637 }
3638 }
3639
3640
3641 /* Read a line from stream FP and return a pointer to a static buffer
3642 containing its contents without the terminating newline. Value
3643 is null when EOF is reached. */
3644
3645 static char *
read_line(FILE * fp)3646 read_line (FILE *fp)
3647 {
3648 static char *buffer;
3649 static int buffer_size;
3650 int i = 0, c;
3651
3652 while ((c = getc (fp)) != EOF && c != '\n')
3653 {
3654 if (i >= buffer_size)
3655 {
3656 buffer_size = max (100, buffer_size * 2);
3657 buffer = (char *) xrealloc (buffer, buffer_size);
3658 }
3659
3660 buffer[i++] = c;
3661 }
3662
3663 if (c == EOF && i == 0)
3664 return NULL;
3665
3666 if (i == buffer_size)
3667 {
3668 buffer_size = max (100, buffer_size * 2);
3669 buffer = (char *) xrealloc (buffer, buffer_size);
3670 }
3671
3672 buffer[i] = '\0';
3673 if (i > 0 && buffer[i - 1] == '\r')
3674 buffer[i - 1] = '\0';
3675 return buffer;
3676 }
3677
3678
3679 /* Main entry point. */
3680
3681 int
main(int argc,char ** argv)3682 main (int argc, char **argv)
3683 {
3684 int i;
3685 int any_inputfiles = 0;
3686 static const char *out_filename = DEFAULT_OUTFILE;
3687 static char **input_filenames = NULL;
3688 static int input_filenames_size = 0;
3689 static int n_input_files;
3690
3691 filename = "command line";
3692 yyout = stdout;
3693
3694 while ((i = getopt_long (argc, argv, "af:I:m:M:no:p:svVx",
3695 options, NULL)) != EOF)
3696 {
3697 switch (i)
3698 {
3699 /* Experimental. */
3700 case 'p':
3701 info_position = atoi (optarg);
3702 break;
3703
3704 case 'n':
3705 f_nested_classes = 0;
3706 break;
3707
3708 case 'x':
3709 f_regexps = 0;
3710 break;
3711
3712 /* Add the name of a file containing more input files. */
3713 case 'f':
3714 if (n_input_files == input_filenames_size)
3715 {
3716 input_filenames_size = max (10, 2 * input_filenames_size);
3717 input_filenames = (char **) xrealloc ((void *)input_filenames,
3718 input_filenames_size);
3719 }
3720 input_filenames[n_input_files++] = xstrdup (optarg);
3721 break;
3722
3723 /* Append new output to output file instead of truncating it. */
3724 case 'a':
3725 f_append = 1;
3726 break;
3727
3728 /* Include structs in the output */
3729 case 's':
3730 f_structs = 0;
3731 break;
3732
3733 /* Be verbose (give a progress indication). */
3734 case 'v':
3735 f_verbose = 1;
3736 break;
3737
3738 /* Be very verbose (print file names as they are processed). */
3739 case 'V':
3740 f_verbose = 1;
3741 f_very_verbose = 1;
3742 break;
3743
3744 /* Change the name of the output file. */
3745 case 'o':
3746 out_filename = optarg;
3747 break;
3748
3749 /* Set minimum length for regular expression strings
3750 when recorded in the output file. */
3751 case 'm':
3752 min_regexp = atoi (optarg);
3753 break;
3754
3755 /* Set maximum length for regular expression strings
3756 when recorded in the output file. */
3757 case 'M':
3758 max_regexp = atoi (optarg);
3759 break;
3760
3761 /* Add to search path. */
3762 case 'I':
3763 add_search_path (optarg);
3764 break;
3765
3766 /* Display help */
3767 case -2:
3768 usage (0);
3769 break;
3770
3771 case -3:
3772 version ();
3773 break;
3774 }
3775 }
3776
3777 /* Call init_scanner after command line flags have been processed to be
3778 able to add keywords depending on command line (not yet
3779 implemented). */
3780 init_scanner ();
3781 init_sym ();
3782
3783 /* Open output file */
3784 if (*out_filename)
3785 {
3786 if (f_append)
3787 {
3788 /* Check that the file to append to exists, and is not
3789 empty. More specifically, it should be a valid file
3790 produced by a previous run of ebrowse, but that's too
3791 difficult to check. */
3792 FILE *fp;
3793 int rc;
3794
3795 fp = fopen (out_filename, "r");
3796 if (fp == NULL)
3797 {
3798 yyerror ("file '%s' must exist for --append", out_filename);
3799 exit (EXIT_FAILURE);
3800 }
3801
3802 rc = fseek (fp, 0, SEEK_END);
3803 if (rc == -1)
3804 {
3805 yyerror ("error seeking in file '%s'", out_filename);
3806 exit (EXIT_FAILURE);
3807 }
3808
3809 rc = ftell (fp);
3810 if (rc == -1)
3811 {
3812 yyerror ("error getting size of file '%s'", out_filename);
3813 exit (EXIT_FAILURE);
3814 }
3815
3816 else if (rc == 0)
3817 {
3818 yyerror ("file '%s' is empty", out_filename);
3819 /* It may be ok to use an empty file for appending.
3820 exit (EXIT_FAILURE); */
3821 }
3822
3823 fclose (fp);
3824 }
3825
3826 yyout = fopen (out_filename, f_append ? "a" : "w");
3827 if (yyout == NULL)
3828 {
3829 yyerror ("cannot open output file '%s'", out_filename);
3830 exit (EXIT_FAILURE);
3831 }
3832 }
3833
3834 /* Process input files specified on the command line. */
3835 while (optind < argc)
3836 {
3837 process_file (argv[optind++]);
3838 any_inputfiles = 1;
3839 }
3840
3841 /* Process files given on stdin if no files specified. */
3842 if (!any_inputfiles && n_input_files == 0)
3843 {
3844 char *file;
3845 while ((file = read_line (stdin)) != NULL)
3846 process_file (file);
3847 }
3848 else
3849 {
3850 /* Process files from `--files=FILE'. Every line in FILE names
3851 one input file to process. */
3852 for (i = 0; i < n_input_files; ++i)
3853 {
3854 FILE *fp = fopen (input_filenames[i], "r");
3855
3856 if (fp == NULL)
3857 yyerror ("cannot open input file '%s'", input_filenames[i]);
3858 else
3859 {
3860 char *file;
3861 while ((file = read_line (fp)) != NULL)
3862 process_file (file);
3863 fclose (fp);
3864 }
3865 }
3866 }
3867
3868 /* Write output file. */
3869 dump_roots (yyout);
3870
3871 /* Close output file. */
3872 if (yyout != stdout)
3873 fclose (yyout);
3874
3875 return EXIT_SUCCESS;
3876 }
3877
3878 /* ebrowse.c ends here */
3879