*** empty log message ***
[emacs.git] / lib-src / ebrowse.c
blobe0f2f5ce9ef990605ca6e13b8ef2c1d85c1c24e8
1 /* ebrowse.c --- parsing files for the ebrowse C++ browser
3 Copyright (C) 1992-1999, 2000 Free Software Foundation Inc.
5 Author: Gerd Moellmann <gerd@gnu.org>
6 Maintainer: FSF
8 This file is part of GNU Emacs.
10 GNU Emacs is free software; you can redistribute it and/or modify
11 it under the terms of the GNU General Public License as published by
12 the Free Software Foundation; either version 2, or (at your option)
13 any later version.
15 GNU Emacs is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with GNU Emacs; see the file COPYING. If not, write to
22 the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
24 #ifdef HAVE_CONFIG_H
25 #include <config.h>
26 #endif
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <ctype.h>
32 #include <assert.h>
33 #include "getopt.h"
35 /* Conditionalize function prototypes. */
37 #ifdef PROTOTYPES /* From config.h. */
38 #define P_(x) x
39 #else
40 #define P_(x) ()
41 #endif
43 /* Value is non-zero if strings X and Y compare equal. */
45 #define streq(X, Y) (*(X) == *(Y) && strcmp ((X) + 1, (Y) + 1) == 0)
47 /* The ubiquitous `max' and `min' macros. */
49 #ifndef max
50 #define max(X, Y) ((X) > (Y) ? (X) : (Y))
51 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
52 #endif
54 /* Files are read in chunks of this number of bytes. */
56 #define READ_CHUNK_SIZE (100 * 1024)
58 /* The character used as a separator in path lists (like $PATH). */
60 #if defined(__MSDOS__)
61 #define PATH_LIST_SEPARATOR ';'
62 #define FILENAME_EQ(X,Y) (strcasecmp(X,Y) == 0)
63 #else
64 #if defined(WINDOWSNT)
65 #define PATH_LIST_SEPARATOR ';'
66 #define FILENAME_EQ(X,Y) (stricmp(X,Y) == 0)
67 #else
68 #define PATH_LIST_SEPARATOR ':'
69 #define FILENAME_EQ(X,Y) (streq(X,Y))
70 #endif
71 #endif
72 /* The default output file name. */
74 #define DEFAULT_OUTFILE "BROWSE"
76 /* A version string written to the output file. Change this whenever
77 the structure of the output file changes. */
79 #define EBROWSE_FILE_VERSION "ebrowse 5.0"
81 /* The output file consists of a tree of Lisp objects, with major
82 nodes built out of Lisp structures. These are the heads of the
83 Lisp structs with symbols identifying their type. */
85 #define TREE_HEADER_STRUCT "[ebrowse-hs "
86 #define TREE_STRUCT "[ebrowse-ts "
87 #define MEMBER_STRUCT "[ebrowse-ms "
88 #define BROWSE_STRUCT "[ebrowse-bs "
89 #define CLASS_STRUCT "[ebrowse-cs "
91 /* The name of the symbol table entry for global functions, variables,
92 defines etc. This name also appears in the browser display. */
94 #define GLOBALS_NAME "*Globals*"
96 /* Token definitions. */
98 enum token
100 YYEOF = 0, /* end of file */
101 CSTRING = 256, /* string constant */
102 CCHAR, /* character constant */
103 CINT, /* integral constant */
104 CFLOAT, /* real constant */
106 ELLIPSIS, /* ... */
107 LSHIFTASGN, /* <<= */
108 RSHIFTASGN, /* >>= */
109 ARROWSTAR, /* ->* */
110 IDENT, /* identifier */
111 DIVASGN, /* /= */
112 INC, /* ++ */
113 ADDASGN, /* += */
114 DEC, /* -- */
115 ARROW, /* -> */
116 SUBASGN, /* -= */
117 MULASGN, /* *= */
118 MODASGN, /* %= */
119 LOR, /* || */
120 ORASGN, /* |= */
121 LAND, /* && */
122 ANDASGN, /* &= */
123 XORASGN, /* ^= */
124 POINTSTAR, /* .* */
125 DCOLON, /* :: */
126 EQ, /* == */
127 NE, /* != */
128 LE, /* <= */
129 LSHIFT, /* << */
130 GE, /* >= */
131 RSHIFT, /* >> */
133 /* Keywords. The undef's are there because these
134 three symbols are very likely to be defined somewhere. */
135 #undef BOOL
136 #undef TRUE
137 #undef FALSE
139 ASM, /* asm */
140 AUTO, /* auto */
141 BREAK, /* break */
142 CASE, /* case */
143 CATCH, /* catch */
144 CHAR, /* char */
145 CLASS, /* class */
146 CONST, /* const */
147 CONTINUE, /* continue */
148 DEFAULT, /* default */
149 DELETE, /* delete */
150 DO, /* do */
151 DOUBLE, /* double */
152 ELSE, /* else */
153 ENUM, /* enum */
154 EXTERN, /* extern */
155 FLOAT, /* float */
156 FOR, /* for */
157 FRIEND, /* friend */
158 GOTO, /* goto */
159 IF, /* if */
160 T_INLINE, /* inline */
161 INT, /* int */
162 LONG, /* long */
163 NEW, /* new */
164 OPERATOR, /* operator */
165 PRIVATE, /* private */
166 PROTECTED, /* protected */
167 PUBLIC, /* public */
168 REGISTER, /* register */
169 RETURN, /* return */
170 SHORT, /* short */
171 SIGNED, /* signed */
172 SIZEOF, /* sizeof */
173 STATIC, /* static */
174 STRUCT, /* struct */
175 SWITCH, /* switch */
176 TEMPLATE, /* template */
177 THIS, /* this */
178 THROW, /* throw */
179 TRY, /* try */
180 TYPEDEF, /* typedef */
181 UNION, /* union */
182 UNSIGNED, /* unsigned */
183 VIRTUAL, /* virtual */
184 VOID, /* void */
185 VOLATILE, /* volatile */
186 WHILE, /* while */
187 MUTABLE, /* mutable */
188 BOOL, /* bool */
189 TRUE, /* true */
190 FALSE, /* false */
191 SIGNATURE, /* signature (GNU extension) */
192 NAMESPACE, /* namespace */
193 EXPLICIT, /* explicit */
194 TYPENAME, /* typename */
195 CONST_CAST, /* const_cast */
196 DYNAMIC_CAST, /* dynamic_cast */
197 REINTERPRET_CAST, /* reinterpret_cast */
198 STATIC_CAST, /* static_cast */
199 TYPEID, /* typeid */
200 USING, /* using */
201 WCHAR /* wchar_t */
204 /* Storage classes, in a wider sense. */
206 enum sc
208 SC_UNKNOWN,
209 SC_MEMBER, /* Is an instance member. */
210 SC_STATIC, /* Is static member. */
211 SC_FRIEND, /* Is friend function. */
212 SC_TYPE /* Is a type definition. */
215 /* Member visibility. */
217 enum visibility
219 V_PUBLIC,
220 V_PROTECTED,
221 V_PRIVATE
224 /* Member flags. */
226 #define F_VIRTUAL 1 /* Is virtual function. */
227 #define F_INLINE 2 /* Is inline function. */
228 #define F_CONST 4 /* Is const. */
229 #define F_PURE 8 /* Is pure virtual function. */
230 #define F_MUTABLE 16 /* Is mutable. */
231 #define F_TEMPLATE 32 /* Is a template. */
232 #define F_EXPLICIT 64 /* Is explicit constructor. */
233 #define F_THROW 128 /* Has a throw specification. */
234 #define F_EXTERNC 256 /* Is declared extern "C". */
235 #define F_DEFINE 512 /* Is a #define. */
237 /* Two macros to set and test a bit in an int. */
239 #define SET_FLAG(F, FLAG) ((F) |= (FLAG))
240 #define HAS_FLAG(F, FLAG) (((F) & (FLAG)) != 0)
242 /* Structure describing a class member. */
244 struct member
246 struct member *next; /* Next in list of members. */
247 struct member *anext; /* Collision chain in member_table. */
248 struct member **list; /* Pointer to list in class. */
249 unsigned param_hash; /* Hash value for parameter types. */
250 int vis; /* Visibility (public, ...). */
251 int flags; /* See F_* above. */
252 char *regexp; /* Matching regular expression. */
253 char *filename; /* Don't free this shared string. */
254 int pos; /* Buffer position of occurrence. */
255 char *def_regexp; /* Regular expression matching definition. */
256 char *def_filename; /* File name of definition. */
257 int def_pos; /* Buffer position of definition. */
258 char name[1]; /* Member name. */
261 /* Structures of this type are used to connect class structures with
262 their super and subclasses. */
264 struct link
266 struct sym *sym; /* The super or subclass. */
267 struct link *next; /* Next in list or NULL. */
270 /* Structure used to record namespace aliases. */
272 struct alias
274 struct alias *next; /* Next in list. */
275 char name[1]; /* Alias name. */
278 /* The structure used to describe a class in the symbol table,
279 or a namespace in all_namespaces. */
281 struct sym
283 int flags; /* Is class a template class?. */
284 unsigned char visited; /* Used to find circles. */
285 struct sym *next; /* Hash collision list. */
286 struct link *subs; /* List of subclasses. */
287 struct link *supers; /* List of superclasses. */
288 struct member *vars; /* List of instance variables. */
289 struct member *fns; /* List of instance functions. */
290 struct member *static_vars; /* List of static variables. */
291 struct member *static_fns; /* List of static functions. */
292 struct member *friends; /* List of friend functions. */
293 struct member *types; /* List of local types. */
294 char *regexp; /* Matching regular expression. */
295 int pos; /* Buffer position. */
296 char *filename; /* File in which it can be found. */
297 char *sfilename; /* File in which members can be found. */
298 struct sym *namesp; /* Namespace in which defined. . */
299 struct alias *namesp_aliases; /* List of aliases for namespaces. */
300 char name[1]; /* Name of the class. */
303 /* Experimental: Print info for `--position-info'. We print
304 '(CLASS-NAME SCOPE MEMBER-NAME). */
306 #define P_DEFN 1
307 #define P_DECL 2
309 int info_where;
310 struct sym *info_cls = NULL;
311 struct member *info_member = NULL;
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. */
317 int info_position = -1;
319 /* Command line options structure for getopt_long. */
321 struct option options[] =
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}
340 /* Semantic values of tokens. Set by yylex.. */
342 unsigned yyival; /* Set for token CINT. */
343 char *yytext; /* Set for token IDENT. */
344 char *yytext_end;
346 /* Output file. */
348 FILE *yyout;
350 /* Current line number. */
352 int yyline;
354 /* The name of the current input file. */
356 char *filename;
358 /* Three character class vectors, and macros to test membership
359 of characters. */
361 char is_ident[255];
362 char is_digit[255];
363 char is_white[255];
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)]
369 /* Command line flags. */
371 int f_append;
372 int f_verbose;
373 int f_very_verbose;
374 int f_structs = 1;
375 int f_regexps = 1;
376 int f_nested_classes = 1;
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. */
382 int min_regexp = 5;
383 int max_regexp = 50;
385 /* Input buffer. */
387 char *inbuffer;
388 char *in;
389 int inbuffer_size;
391 /* Return the current buffer position in the input file. */
393 #define BUFFER_POS() (in - inbuffer)
395 /* If current lookahead is CSTRING, the following points to the
396 first character in the string constant. Used for recognizing
397 extern "C". */
399 char *string_start;
401 /* The size of the hash tables for classes.and members. Should be
402 prime. */
404 #define TABLE_SIZE 1001
406 /* The hash table for class symbols. */
408 struct sym *class_table[TABLE_SIZE];
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'. */
414 struct member *member_table[TABLE_SIZE];
416 /* The special class symbol used to hold global functions,
417 variables etc. */
419 struct sym *global_symbols;
421 /* The current namespace. */
423 struct sym *current_namespace;
425 /* The list of all known namespaces. */
427 struct sym *all_namespaces;
429 /* Stack of namespaces we're currently nested in, during the parse. */
431 struct sym **namespace_stack;
432 int namespace_stack_size;
433 int namespace_sp;
435 /* The current lookahead token. */
437 int tk = -1;
439 /* Structure describing a keyword. */
441 struct kw
443 char *name; /* Spelling. */
444 int tk; /* Token value. */
445 struct kw *next; /* Next in collision chain. */
448 /* Keywords are lookup up in a hash table of their own. */
450 #define KEYWORD_TABLE_SIZE 1001
451 struct kw *keyword_table[KEYWORD_TABLE_SIZE];
453 /* Search path. */
455 struct search_path
457 char *path;
458 struct search_path *next;
461 struct search_path *search_path;
462 struct search_path *search_path_tail;
464 /* Function prototypes. */
466 int yylex P_ ((void));
467 void yyparse P_ ((void));
468 void re_init_parser P_ ((void));
469 char *token_string P_ ((int));
470 char *matching_regexp P_ ((void));
471 void init_sym P_ ((void));
472 struct sym *add_sym P_ ((char *, struct sym *));
473 void add_link P_ ((struct sym *, struct sym *));
474 void add_member_defn P_ ((struct sym *, char *, char *,
475 int, unsigned, int, int, int));
476 void add_member_decl P_ ((struct sym *, char *, char *, int,
477 unsigned, int, int, int, int));
478 void dump_roots P_ ((FILE *));
479 void *xmalloc P_ ((int));
480 void add_global_defn P_ ((char *, char *, int, unsigned, int, int, int));
481 void add_global_decl P_ ((char *, char *, int, unsigned, int, int, int));
482 void add_define P_ ((char *, char *, int));
483 void mark_inherited_virtual P_ ((void));
484 void leave_namespace P_ ((void));
485 void enter_namespace P_ ((char *));
486 void register_namespace_alias P_ ((char *, char *));
487 void insert_keyword P_ ((char *, int));
488 void re_init_scanner P_ ((void));
489 void init_scanner P_ ((void));
490 void usage P_ ((int));
491 void version P_ ((void));
492 void process_file P_ ((char *));
493 void add_search_path P_ ((char *));
494 FILE *open_file P_ ((char *));
495 int process_pp_line P_ ((void));
496 int dump_members P_ ((FILE *, struct member *));
497 void dump_sym P_ ((FILE *, struct sym *));
498 int dump_tree P_ ((FILE *, struct sym *));
499 struct member *find_member P_ ((struct sym *, char *, int, int, unsigned));
500 struct member *add_member P_ ((struct sym *, char *, int, int, unsigned));
501 void mark_virtual P_ ((struct sym *));
502 void mark_virtual P_ ((struct sym *));
503 struct sym *make_namespace P_ ((char *));
504 char *sym_scope P_ ((struct sym *));
505 char *sym_scope_1 P_ ((struct sym *));
506 int skip_to P_ ((int));
507 void skip_matching P_ ((void));
508 void member P_ ((struct sym *, int));
509 void class_body P_ ((struct sym *, int));
510 void class_definition P_ ((struct sym *, int, int, int));
511 void declaration P_ ((int));
512 unsigned parm_list P_ ((int *));
513 char *operator_name P_ ((int *));
514 struct sym *parse_classname P_ ((void));
515 struct sym *parse_qualified_ident_or_type P_ ((char **));
516 void parse_qualified_param_ident_or_type P_ ((char **));
517 int globals P_ ((int));
521 /***********************************************************************
522 Utilities
523 ***********************************************************************/
525 /* Print an error in a printf-like style with the current input file
526 name and line number. */
528 void
529 yyerror (format, a1, a2, a3, a4, a5)
530 char *format;
531 int a1, a2, a3, a4, a5;
533 fprintf (stderr, "%s:%d: ", filename, yyline);
534 fprintf (stderr, format, a1, a2, a3, a4, a5);
535 putc ('\n', stderr);
539 /* Like malloc but print an error and exit if not enough memory is
540 available. This isn't called `xmalloc' because src/m/alpha.h,
541 and maybe others, contain an incompatible prototype for xmalloc
542 and xrealloc. */
544 void *
545 xmalloc (nbytes)
546 int nbytes;
548 void *p = malloc (nbytes);
549 if (p == NULL)
551 yyerror ("out of memory");
552 exit (1);
554 return p;
558 /* Like realloc but print an error and exit if out of memory. */
560 void *
561 xrealloc (p, sz)
562 void *p;
563 int sz;
565 p = realloc (p, sz);
566 if (p == NULL)
568 yyerror ("out of memory");
569 exit (1);
571 return p;
575 /* Like strdup, but print an error and exit if not enough memory is
576 available.. If S is null, return null. */
578 char *
579 xstrdup (s)
580 char *s;
582 if (s)
583 s = strcpy (xmalloc (strlen (s) + 1), s);
584 return s;
589 /***********************************************************************
590 Symbols
591 ***********************************************************************/
593 /* Initialize the symbol table. This currently only sets up the
594 special symbol for globals (`*Globals*'). */
596 void
597 init_sym ()
599 global_symbols = add_sym (GLOBALS_NAME, NULL);
603 /* Add a symbol for class NAME to the symbol table. NESTED_IN_CLASS
604 is the class in which class NAME was found. If it is null,
605 this means the scope of NAME is the current namespace.
607 If a symbol for NAME already exists, return that. Otherwise
608 create a new symbol and set it to default values. */
610 struct sym *
611 add_sym (name, nested_in_class)
612 char *name;
613 struct sym *nested_in_class;
615 struct sym *sym;
616 unsigned h;
617 char *s;
618 struct sym *scope = nested_in_class ? nested_in_class : current_namespace;
620 for (s = name, h = 0; *s; ++s)
621 h = (h << 1) ^ *s;
622 h %= TABLE_SIZE;
624 for (sym = class_table[h]; sym; sym = sym->next)
625 if (streq (name, sym->name) && sym->namesp == scope)
626 break;
628 if (sym == NULL)
630 if (f_very_verbose)
632 putchar ('\t');
633 puts (name);
636 sym = (struct sym *) xmalloc (sizeof *sym + strlen (name));
637 bzero (sym, sizeof *sym);
638 strcpy (sym->name, name);
639 sym->namesp = scope;
640 sym->next = class_table[h];
641 class_table[h] = sym;
644 return sym;
648 /* Add links between superclass SUPER and subclass SUB. */
650 void
651 add_link (super, sub)
652 struct sym *super, *sub;
654 struct link *lnk, *lnk2, *p, *prev;
656 /* See if a link already exists. */
657 for (p = super->subs, prev = NULL;
658 p && strcmp (sub->name, p->sym->name) > 0;
659 prev = p, p = p->next)
662 /* Avoid duplicates. */
663 if (p == NULL || p->sym != sub)
665 lnk = (struct link *) xmalloc (sizeof *lnk);
666 lnk2 = (struct link *) xmalloc (sizeof *lnk2);
668 lnk->sym = sub;
669 lnk->next = p;
671 if (prev)
672 prev->next = lnk;
673 else
674 super->subs = lnk;
676 lnk2->sym = super;
677 lnk2->next = sub->supers;
678 sub->supers = lnk2;
683 /* Find in class CLS member NAME.
685 VAR non-zero means look for a member variable; otherwise a function
686 is searched. SC specifies what kind of member is searched---a
687 static, or per-instance member etc. HASH is a hash code for the
688 parameter types of functions. Value is a pointer to the member
689 found or null if not found. */
691 struct member *
692 find_member (cls, name, var, sc, hash)
693 struct sym *cls;
694 char *name;
695 int var, sc;
696 unsigned hash;
698 struct member **list;
699 struct member *p;
700 unsigned name_hash = 0;
701 char *s;
702 int i;
704 switch (sc)
706 case SC_FRIEND:
707 list = &cls->friends;
708 break;
710 case SC_TYPE:
711 list = &cls->types;
712 break;
714 case SC_STATIC:
715 list = var ? &cls->static_vars : &cls->static_fns;
716 break;
718 default:
719 list = var ? &cls->vars : &cls->fns;
720 break;
723 for (s = name; *s; ++s)
724 name_hash = (name_hash << 1) ^ *s;
725 i = name_hash % TABLE_SIZE;
727 for (p = member_table[i]; p; p = p->anext)
728 if (p->list == list && p->param_hash == hash && streq (name, p->name))
729 break;
731 return p;
735 /* Add to class CLS information for the declaration of member NAME.
736 REGEXP is a regexp matching the declaration, if non-null. POS is
737 the position in the source where the declaration is found. HASH is
738 a hash code for the parameter list of the member, if it's a
739 function. VAR non-zero means member is a variable or type. SC
740 specifies the type of member (instance member, static, ...). VIS
741 is the member's visibility (public, protected, private). FLAGS is
742 a bit set giving additional information about the member (see the
743 F_* defines). */
745 void
746 add_member_decl (cls, name, regexp, pos, hash, var, sc, vis, flags)
747 struct sym *cls;
748 char *name;
749 char *regexp;
750 int pos;
751 unsigned hash;
752 int var;
753 int sc;
754 int vis;
755 int flags;
757 struct member *m;
759 m = find_member (cls, name, var, sc, hash);
760 if (m == NULL)
761 m = add_member (cls, name, var, sc, hash);
763 /* Have we seen a new filename? If so record that. */
764 if (!cls->filename || !FILENAME_EQ (cls->filename, filename))
765 m->filename = filename;
767 m->regexp = regexp;
768 m->pos = pos;
769 m->flags = flags;
771 switch (vis)
773 case PRIVATE:
774 m->vis = V_PRIVATE;
775 break;
777 case PROTECTED:
778 m->vis = V_PROTECTED;
779 break;
781 case PUBLIC:
782 m->vis = V_PUBLIC;
783 break;
786 info_where = P_DECL;
787 info_cls = cls;
788 info_member = m;
792 /* Add to class CLS information for the definition of member NAME.
793 REGEXP is a regexp matching the declaration, if non-null. POS is
794 the position in the source where the declaration is found. HASH is
795 a hash code for the parameter list of the member, if it's a
796 function. VAR non-zero means member is a variable or type. SC
797 specifies the type of member (instance member, static, ...). VIS
798 is the member's visibility (public, protected, private). FLAGS is
799 a bit set giving additional information about the member (see the
800 F_* defines). */
802 void
803 add_member_defn (cls, name, regexp, pos, hash, var, sc, flags)
804 struct sym *cls;
805 char *name;
806 char *regexp;
807 int pos;
808 unsigned hash;
809 int var;
810 int sc;
811 int flags;
813 struct member *m;
815 if (sc == SC_UNKNOWN)
817 m = find_member (cls, name, var, SC_MEMBER, hash);
818 if (m == NULL)
820 m = find_member (cls, name, var, SC_STATIC, hash);
821 if (m == NULL)
822 m = add_member (cls, name, var, sc, hash);
825 else
827 m = find_member (cls, name, var, sc, hash);
828 if (m == NULL)
829 m = add_member (cls, name, var, sc, hash);
832 if (!cls->sfilename)
833 cls->sfilename = filename;
835 if (!FILENAME_EQ (cls->sfilename, filename))
836 m->def_filename = filename;
838 m->def_regexp = regexp;
839 m->def_pos = pos;
840 m->flags |= flags;
842 info_where = P_DEFN;
843 info_cls = cls;
844 info_member = m;
848 /* Add a symbol for a define named NAME to the symbol table.
849 REGEXP is a regular expression matching the define in the source,
850 if it is non-null. POS is the position in the file. */
852 void
853 add_define (name, regexp, pos)
854 char *name, *regexp;
855 int pos;
857 add_global_defn (name, regexp, pos, 0, 1, SC_FRIEND, F_DEFINE);
858 add_global_decl (name, regexp, pos, 0, 1, SC_FRIEND, F_DEFINE);
862 /* Add information for the global definition of NAME.
863 REGEXP is a regexp matching the declaration, if non-null. POS is
864 the position in the source where the declaration is found. HASH is
865 a hash code for the parameter list of the member, if it's a
866 function. VAR non-zero means member is a variable or type. SC
867 specifies the type of member (instance member, static, ...). VIS
868 is the member's visibility (public, protected, private). FLAGS is
869 a bit set giving additional information about the member (see the
870 F_* defines). */
872 void
873 add_global_defn (name, regexp, pos, hash, var, sc, flags)
874 char *name, *regexp;
875 int pos;
876 unsigned hash;
877 int var;
878 int sc;
879 int flags;
881 int i;
882 struct sym *sym;
884 /* Try to find out for which classes a function is a friend, and add
885 what we know about it to them. */
886 if (!var)
887 for (i = 0; i < TABLE_SIZE; ++i)
888 for (sym = class_table[i]; sym; sym = sym->next)
889 if (sym != global_symbols && sym->friends)
890 if (find_member (sym, name, 0, SC_FRIEND, hash))
891 add_member_defn (sym, name, regexp, pos, hash, 0,
892 SC_FRIEND, flags);
894 /* Add to global symbols. */
895 add_member_defn (global_symbols, name, regexp, pos, hash, var, sc, flags);
899 /* Add information for the global declaration of NAME.
900 REGEXP is a regexp matching the declaration, if non-null. POS is
901 the position in the source where the declaration is found. HASH is
902 a hash code for the parameter list of the member, if it's a
903 function. VAR non-zero means member is a variable or type. SC
904 specifies the type of member (instance member, static, ...). VIS
905 is the member's visibility (public, protected, private). FLAGS is
906 a bit set giving additional information about the member (see the
907 F_* defines). */
909 void
910 add_global_decl (name, regexp, pos, hash, var, sc, flags)
911 char *name, *regexp;
912 int pos;
913 unsigned hash;
914 int var;
915 int sc;
916 int flags;
918 /* Add declaration only if not already declared. Header files must
919 be processed before source files for this to have the right effect.
920 I do not want to handle implicit declarations at the moment. */
921 struct member *m;
922 struct member *found;
924 m = found = find_member (global_symbols, name, var, sc, hash);
925 if (m == NULL)
926 m = add_member (global_symbols, name, var, sc, hash);
928 /* Definition already seen => probably last declaration implicit.
929 Override. This means that declarations must always be added to
930 the symbol table before definitions. */
931 if (!found)
933 if (!global_symbols->filename
934 || !FILENAME_EQ (global_symbols->filename, filename))
935 m->filename = filename;
937 m->regexp = regexp;
938 m->pos = pos;
939 m->vis = V_PUBLIC;
940 m->flags = flags;
942 info_where = P_DECL;
943 info_cls = global_symbols;
944 info_member = m;
949 /* Add a symbol for member NAME to class CLS.
950 VAR non-zero means it's a variable. SC specifies the kind of
951 member. HASH is a hash code for the parameter types of a function.
952 Value is a pointer to the member's structure. */
954 struct member *
955 add_member (cls, name, var, sc, hash)
956 struct sym *cls;
957 char *name;
958 int var;
959 int sc;
960 unsigned hash;
962 struct member *m = (struct member *) xmalloc (sizeof *m + strlen (name));
963 struct member **list;
964 struct member *p;
965 struct member *prev;
966 unsigned name_hash = 0;
967 int i;
968 char *s;
970 strcpy (m->name, name);
971 m->param_hash = hash;
973 m->vis = 0;
974 m->flags = 0;
975 m->regexp = NULL;
976 m->filename = NULL;
977 m->pos = 0;
978 m->def_regexp = NULL;
979 m->def_filename = NULL;
980 m->def_pos = 0;
982 assert (cls != NULL);
984 switch (sc)
986 case SC_FRIEND:
987 list = &cls->friends;
988 break;
990 case SC_TYPE:
991 list = &cls->types;
992 break;
994 case SC_STATIC:
995 list = var ? &cls->static_vars : &cls->static_fns;
996 break;
998 default:
999 list = var ? &cls->vars : &cls->fns;
1000 break;
1003 for (s = name; *s; ++s)
1004 name_hash = (name_hash << 1) ^ *s;
1005 i = name_hash % TABLE_SIZE;
1006 m->anext = member_table[i];
1007 member_table[i] = m;
1008 m->list = list;
1010 /* Keep the member list sorted. It's cheaper to do it here than to
1011 sort them in Lisp. */
1012 for (prev = NULL, p = *list;
1013 p && strcmp (name, p->name) > 0;
1014 prev = p, p = p->next)
1017 m->next = p;
1018 if (prev)
1019 prev->next = m;
1020 else
1021 *list = m;
1022 return m;
1026 /* Given the root R of a class tree, step through all subclasses
1027 recursively, marking functions as virtual that are declared virtual
1028 in base classes. */
1030 void
1031 mark_virtual (r)
1032 struct sym *r;
1034 struct link *p;
1035 struct member *m, *m2;
1037 for (p = r->subs; p; p = p->next)
1039 for (m = r->fns; m; m = m->next)
1040 if (HAS_FLAG (m->flags, F_VIRTUAL))
1042 for (m2 = p->sym->fns; m2; m2 = m2->next)
1043 if (m->param_hash == m2->param_hash && streq (m->name, m2->name))
1044 SET_FLAG (m2->flags, F_VIRTUAL);
1047 mark_virtual (p->sym);
1052 /* For all roots of the class tree, mark functions as virtual that
1053 are virtual because of a virtual declaration in a base class. */
1055 void
1056 mark_inherited_virtual ()
1058 struct sym *r;
1059 int i;
1061 for (i = 0; i < TABLE_SIZE; ++i)
1062 for (r = class_table[i]; r; r = r->next)
1063 if (r->supers == NULL)
1064 mark_virtual (r);
1068 /* Create and return a symbol for a namespace with name NAME. */
1070 struct sym *
1071 make_namespace (name)
1072 char *name;
1074 struct sym *s = (struct sym *) xmalloc (sizeof *s + strlen (name));
1075 bzero (s, sizeof *s);
1076 strcpy (s->name, name);
1077 s->next = all_namespaces;
1078 s->namesp = current_namespace;
1079 all_namespaces = s;
1080 return s;
1084 /* Find the symbol for namespace NAME. If not found, add a new symbol
1085 for NAME to all_namespaces. */
1087 struct sym *
1088 find_namespace (name)
1089 char *name;
1091 struct sym *p;
1093 for (p = all_namespaces; p; p = p->next)
1095 if (streq (p->name, name))
1096 break;
1097 else
1099 struct alias *p2;
1100 for (p2 = p->namesp_aliases; p2; p2 = p2->next)
1101 if (streq (p2->name, name))
1102 break;
1103 if (p2)
1104 break;
1108 if (p == NULL)
1109 p = make_namespace (name);
1111 return p;
1115 /* Register the name NEW_NAME as an alias for namespace OLD_NAME. */
1117 void
1118 register_namespace_alias (new_name, old_name)
1119 char *new_name, *old_name;
1121 struct sym *p = find_namespace (old_name);
1122 struct alias *al;
1124 /* Is it already in the list of aliases? */
1125 for (al = p->namesp_aliases; al; al = al->next)
1126 if (streq (new_name, p->name))
1127 return;
1129 al = (struct alias *) xmalloc (sizeof *al + strlen (new_name));
1130 strcpy (al->name, new_name);
1131 al->next = p->namesp_aliases;
1132 p->namesp_aliases = al;
1136 /* Enter namespace with name NAME. */
1138 void
1139 enter_namespace (name)
1140 char *name;
1142 struct sym *p = find_namespace (name);
1144 if (namespace_sp == namespace_stack_size)
1146 int size = max (10, 2 * namespace_stack_size);
1147 namespace_stack = (struct sym **) xrealloc (namespace_stack, size);
1148 namespace_stack_size = size;
1151 namespace_stack[namespace_sp++] = current_namespace;
1152 current_namespace = p;
1156 /* Leave the current namespace. */
1158 void
1159 leave_namespace ()
1161 assert (namespace_sp > 0);
1162 current_namespace = namespace_stack[--namespace_sp];
1167 /***********************************************************************
1168 Writing the Output File
1169 ***********************************************************************/
1171 /* Write string S to the output file FP in a Lisp-readable form.
1172 If S is null, write out `()'. */
1174 #define PUTSTR(s, fp) \
1175 do { \
1176 if (!s) \
1178 putc ('(', fp); \
1179 putc (')', fp); \
1180 putc (' ', fp); \
1182 else \
1184 putc ('"', fp); \
1185 fputs (s, fp); \
1186 putc ('"', fp); \
1187 putc (' ', fp); \
1189 } while (0)
1191 /* A dynamically allocated buffer for constructing a scope name. */
1193 char *scope_buffer;
1194 int scope_buffer_size;
1195 int scope_buffer_len;
1198 /* Make sure scope_buffer has enough room to add LEN chars to it. */
1200 void
1201 ensure_scope_buffer_room (len)
1202 int len;
1204 if (scope_buffer_len + len >= scope_buffer_size)
1206 int new_size = max (2 * scope_buffer_size, scope_buffer_len + len);
1207 scope_buffer = (char *) xrealloc (new_size);
1208 scope_buffer_size = new_size;
1213 /* Recursively add the scope names of symbol P and the scopes of its
1214 namespaces to scope_buffer. Value is a pointer to the complete
1215 scope name constructed. */
1217 char *
1218 sym_scope_1 (p)
1219 struct sym *p;
1221 int len;
1223 if (p->namesp)
1224 sym_scope_1 (p->namesp);
1226 if (*scope_buffer)
1228 ensure_scope_buffer_room (3);
1229 strcat (scope_buffer, "::");
1230 scope_buffer_len += 2;
1233 len = strlen (p->name);
1234 ensure_scope_buffer_room (len + 1);
1235 strcat (scope_buffer, p->name);
1236 scope_buffer_len += len;
1238 if (HAS_FLAG (p->flags, F_TEMPLATE))
1240 ensure_scope_buffer_room (3);
1241 strcat (scope_buffer, "<>");
1242 scope_buffer_len += 2;
1245 return scope_buffer;
1249 /* Return the scope of symbol P in printed representation, i.e.
1250 as it would appear in a C*+ source file. */
1252 char *
1253 sym_scope (p)
1254 struct sym *p;
1256 if (!scope_buffer)
1258 scope_buffer_size = 1024;
1259 scope_buffer = (char *) xmalloc (scope_buffer_size);
1262 *scope_buffer = '\0';
1263 scope_buffer_len = 0;
1265 if (p->namesp)
1266 sym_scope_1 (p->namesp);
1268 return scope_buffer;
1272 /* Dump the list of members M to file FP. Value is the length of the
1273 list. */
1276 dump_members (fp, m)
1277 FILE *fp;
1278 struct member *m;
1280 int n;
1282 putc ('(', fp);
1284 for (n = 0; m; m = m->next, ++n)
1286 fputs (MEMBER_STRUCT, fp);
1287 PUTSTR (m->name, fp);
1288 PUTSTR (NULL, fp); /* FIXME? scope for globals */
1289 fprintf (fp, "%u ", (unsigned) m->flags);
1290 PUTSTR (m->filename, fp);
1291 PUTSTR (m->regexp, fp);
1292 fprintf (fp, "%u ", (unsigned) m->pos);
1293 fprintf (fp, "%u ", (unsigned) m->vis);
1294 putc (' ', fp);
1295 PUTSTR (m->def_filename, fp);
1296 PUTSTR (m->def_regexp, fp);
1297 fprintf (fp, "%u", (unsigned) m->def_pos);
1298 putc (']', fp);
1299 putc ('\n', fp);
1302 putc (')', fp);
1303 putc ('\n', fp);
1304 return n;
1308 /* Dump class ROOT to stream FP. */
1310 void
1311 dump_sym (fp, root)
1312 FILE *fp;
1313 struct sym *root;
1315 fputs (CLASS_STRUCT, fp);
1316 PUTSTR (root->name, fp);
1318 /* Print scope, if any. */
1319 if (root->namesp)
1320 PUTSTR (sym_scope (root), fp);
1321 else
1322 PUTSTR (NULL, fp);
1324 /* Print flags. */
1325 fprintf (fp, "%u", root->flags);
1326 PUTSTR (root->filename, fp);
1327 PUTSTR (root->regexp, fp);
1328 fprintf (fp, "%u", (unsigned) root->pos);
1329 PUTSTR (root->sfilename, fp);
1330 putc (']', fp);
1331 putc ('\n', fp);
1335 /* Dump class ROOT and its subclasses to file FP. Value is the
1336 number of classes written. */
1339 dump_tree (fp, root)
1340 FILE *fp;
1341 struct sym *root;
1343 struct link *lk;
1344 unsigned n = 0;
1346 dump_sym (fp, root);
1348 if (f_verbose)
1350 putchar ('+');
1351 fflush (stdout);
1354 putc ('(', fp);
1356 for (lk = root->subs; lk; lk = lk->next)
1358 fputs (TREE_STRUCT, fp);
1359 n += dump_tree (fp, lk->sym);
1360 putc (']', fp);
1363 putc (')', fp);
1365 dump_members (fp, root->vars);
1366 n += dump_members (fp, root->fns);
1367 dump_members (fp, root->static_vars);
1368 n += dump_members (fp, root->static_fns);
1369 n += dump_members (fp, root->friends);
1370 dump_members (fp, root->types);
1372 /* Superclasses. */
1373 putc ('(', fp);
1374 putc (')', fp);
1376 /* Mark slot. */
1377 putc ('(', fp);
1378 putc (')', fp);
1380 putc ('\n', fp);
1381 return n;
1385 /* Dump the entire class tree to file FP. */
1387 void
1388 dump_roots (fp)
1389 FILE *fp;
1391 int i, n = 0;
1392 struct sym *r;
1394 /* Output file header containing version string, command line
1395 options etc. */
1396 if (!f_append)
1398 fputs (TREE_HEADER_STRUCT, fp);
1399 PUTSTR (EBROWSE_FILE_VERSION, fp);
1401 putc ('\"', fp);
1402 if (!f_structs)
1403 fputs (" -s", fp);
1404 if (f_regexps)
1405 fputs (" -x", fp);
1406 putc ('\"', fp);
1407 fputs (" ()", fp);
1408 fputs (" ()", fp);
1409 putc (']', fp);
1412 /* Mark functions as virtual that are so because of functions
1413 declared virtual in base classes. */
1414 mark_inherited_virtual ();
1416 /* Dump the roots of the graph. */
1417 for (i = 0; i < TABLE_SIZE; ++i)
1418 for (r = class_table[i]; r; r = r->next)
1419 if (!r->supers)
1421 fputs (TREE_STRUCT, fp);
1422 n += dump_tree (fp, r);
1423 putc (']', fp);
1426 if (f_verbose)
1427 putchar ('\n');
1432 /***********************************************************************
1433 Scanner
1434 ***********************************************************************/
1436 #ifdef DEBUG
1437 #define INCREMENT_LINENO \
1438 do { \
1439 if (f_very_verbose) \
1441 ++yyline; \
1442 printf ("%d:\n", yyline); \
1444 else \
1445 ++yyline; \
1446 } while (0)
1447 #else
1448 #define INCREMENT_LINENO ++yyline
1449 #endif
1451 /* Define two macros for accessing the input buffer (current input
1452 file). GET(C) sets C to the next input character and advances the
1453 input pointer. UNGET retracts the input pointer. */
1455 #define GET(C) ((C) = *in++)
1456 #define UNGET() (--in)
1459 /* Process a preprocessor line. Value is the next character from the
1460 input buffer not consumed. */
1463 process_pp_line ()
1465 int in_comment = 0, in_string = 0;
1466 int c;
1467 char *p = yytext;
1469 /* Skip over white space. The `#' has been consumed already. */
1470 while (WHITEP (GET (c)))
1473 /* Read the preprocessor command (if any). */
1474 while (IDENTP (c))
1476 *p++ = c;
1477 GET (c);
1480 /* Is it a `define'? */
1481 *p = '\0';
1483 if (*yytext && streq (yytext, "define"))
1485 p = yytext;
1486 while (WHITEP (c))
1487 GET (c);
1488 while (IDENTP (c))
1490 *p++ = c;
1491 GET (c);
1494 *p = '\0';
1496 if (*yytext)
1498 char *regexp = matching_regexp ();
1499 int pos = BUFFER_POS ();
1500 add_define (yytext, regexp, pos);
1504 while (c && (c != '\n' || in_comment || in_string))
1506 if (c == '\\')
1507 GET (c);
1508 else if (c == '/' && !in_comment)
1510 if (GET (c) == '*')
1511 in_comment = 1;
1513 else if (c == '*' && in_comment)
1515 if (GET (c) == '/')
1516 in_comment = 0;
1518 else if (c == '"')
1519 in_string = !in_string;
1521 if (c == '\n')
1522 INCREMENT_LINENO;
1524 GET (c);
1527 return c;
1531 /* Value is the next token from the input buffer. */
1534 yylex ()
1536 int c;
1537 char end_char;
1538 char *p;
1540 for (;;)
1542 while (WHITEP (GET (c)))
1545 switch (c)
1547 case '\n':
1548 INCREMENT_LINENO;
1549 break;
1551 case '\r':
1552 break;
1554 case 0:
1555 /* End of file. */
1556 return YYEOF;
1558 case '\\':
1559 GET (c);
1560 break;
1562 case '"':
1563 case '\'':
1564 /* String and character constants. */
1565 end_char = c;
1566 string_start = in;
1567 while (GET (c) && c != end_char)
1569 switch (c)
1571 case '\\':
1572 /* Escape sequences. */
1573 if (!GET (c))
1575 if (end_char == '\'')
1576 yyerror ("EOF in character constant");
1577 else
1578 yyerror ("EOF in string constant");
1579 goto end_string;
1581 else switch (c)
1583 case '\n':
1584 INCREMENT_LINENO;
1585 case 'a':
1586 case 'b':
1587 case 'f':
1588 case 'n':
1589 case 'r':
1590 case 't':
1591 case 'v':
1592 break;
1594 case 'x':
1596 /* Hexadecimal escape sequence. */
1597 int i;
1598 for (i = 0; i < 2; ++i)
1600 GET (c);
1602 if (c >= '0' && c <= '7')
1604 else if (c >= 'a' && c <= 'f')
1606 else if (c >= 'A' && c <= 'F')
1608 else
1610 UNGET ();
1611 break;
1615 break;
1617 case '0':
1619 /* Octal escape sequence. */
1620 int i;
1621 for (i = 0; i < 3; ++i)
1623 GET (c);
1625 if (c >= '0' && c <= '7')
1627 else
1629 UNGET ();
1630 break;
1634 break;
1636 default:
1637 break;
1639 break;
1641 case '\n':
1642 if (end_char == '\'')
1643 yyerror ("newline in character constant");
1644 else
1645 yyerror ("newline in string constant");
1646 INCREMENT_LINENO;
1647 break;
1649 default:
1650 break;
1654 end_string:
1655 return end_char == '\'' ? CCHAR : CSTRING;
1657 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': case 'g':
1658 case 'h': case 'i': case 'j': case 'k': case 'l': case 'm': case 'n':
1659 case 'o': case 'p': case 'q': case 'r': case 's': case 't': case 'u':
1660 case 'v': case 'w': case 'x': case 'y': case 'z':
1661 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': case 'G':
1662 case 'H': case 'I': case 'J': case 'K': case 'L': case 'M': case 'N':
1663 case 'O': case 'P': case 'Q': case 'R': case 'S': case 'T': case 'U':
1664 case 'V': case 'W': case 'X': case 'Y': case 'Z': case '_':
1666 /* Identifier and keywords. */
1667 unsigned hash;
1668 struct kw *k;
1670 p = yytext;
1671 *p++ = hash = c;
1673 while (IDENTP (GET (*p)))
1675 hash = (hash << 1) ^ *p++;
1676 if (p == yytext_end - 1)
1678 int size = yytext_end - yytext;
1679 yytext = (char *) xrealloc (yytext, 2 * size);
1680 yytext_end = yytext + 2 * size;
1681 p = yytext + size - 1;
1685 UNGET ();
1686 *p = 0;
1688 for (k = keyword_table[hash % KEYWORD_TABLE_SIZE]; k; k = k->next)
1689 if (streq (k->name, yytext))
1690 return k->tk;
1692 return IDENT;
1695 case '/':
1696 /* C and C++ comments, '/' and '/='. */
1697 switch (GET (c))
1699 case '*':
1700 while (GET (c))
1702 switch (c)
1704 case '*':
1705 if (GET (c) == '/')
1706 goto comment_end;
1707 UNGET ();
1708 break;
1709 case '\\':
1710 GET (c);
1711 break;
1712 case '\n':
1713 INCREMENT_LINENO;
1714 break;
1717 comment_end:;
1718 break;
1720 case '=':
1721 return DIVASGN;
1723 case '/':
1724 while (GET (c) && c != '\n')
1726 INCREMENT_LINENO;
1727 break;
1729 default:
1730 UNGET ();
1731 return '/';
1733 break;
1735 case '+':
1736 if (GET (c) == '+')
1737 return INC;
1738 else if (c == '=')
1739 return ADDASGN;
1740 UNGET ();
1741 return '+';
1743 case '-':
1744 switch (GET (c))
1746 case '-':
1747 return DEC;
1748 case '>':
1749 if (GET (c) == '*')
1750 return ARROWSTAR;
1751 UNGET ();
1752 return ARROW;
1753 case '=':
1754 return SUBASGN;
1756 UNGET ();
1757 return '-';
1759 case '*':
1760 if (GET (c) == '=')
1761 return MULASGN;
1762 UNGET ();
1763 return '*';
1765 case '%':
1766 if (GET (c) == '=')
1767 return MODASGN;
1768 UNGET ();
1769 return '%';
1771 case '|':
1772 if (GET (c) == '|')
1773 return LOR;
1774 else if (c == '=')
1775 return ORASGN;
1776 UNGET ();
1777 return '|';
1779 case '&':
1780 if (GET (c) == '&')
1781 return LAND;
1782 else if (c == '=')
1783 return ANDASGN;
1784 UNGET ();
1785 return '&';
1787 case '^':
1788 if (GET (c) == '=')
1789 return XORASGN;
1790 UNGET ();
1791 return '^';
1793 case '.':
1794 if (GET (c) == '*')
1795 return POINTSTAR;
1796 else if (c == '.')
1798 if (GET (c) != '.')
1799 yyerror ("invalid token '..' ('...' assumed)");
1800 UNGET ();
1801 return ELLIPSIS;
1803 else if (!DIGITP (c))
1805 UNGET ();
1806 return '.';
1808 goto mantissa;
1810 case ':':
1811 if (GET (c) == ':')
1812 return DCOLON;
1813 UNGET ();
1814 return ':';
1816 case '=':
1817 if (GET (c) == '=')
1818 return EQ;
1819 UNGET ();
1820 return '=';
1822 case '!':
1823 if (GET (c) == '=')
1824 return NE;
1825 UNGET ();
1826 return '!';
1828 case '<':
1829 switch (GET (c))
1831 case '=':
1832 return LE;
1833 case '<':
1834 if (GET (c) == '=')
1835 return LSHIFTASGN;
1836 UNGET ();
1837 return LSHIFT;
1839 UNGET ();
1840 return '<';
1842 case '>':
1843 switch (GET (c))
1845 case '=':
1846 return GE;
1847 case '>':
1848 if (GET (c) == '=')
1849 return RSHIFTASGN;
1850 UNGET ();
1851 return RSHIFT;
1853 UNGET ();
1854 return '>';
1856 case '#':
1857 c = process_pp_line ();
1858 if (c == 0)
1859 return YYEOF;
1860 break;
1862 case '(': case ')': case '[': case ']': case '{': case '}':
1863 case ';': case ',': case '?': case '~':
1864 return c;
1866 case '0':
1867 yyival = 0;
1869 if (GET (c) == 'x' || c == 'X')
1871 while (GET (c))
1873 if (DIGITP (c))
1874 yyival = yyival * 16 + c - '0';
1875 else if (c >= 'a' && c <= 'f')
1876 yyival = yyival * 16 + c - 'a' + 10;
1877 else if (c >= 'A' && c <= 'F')
1878 yyival = yyival * 16 + c - 'A' + 10;
1879 else
1880 break;
1883 goto int_suffixes;
1885 else if (c == '.')
1886 goto mantissa;
1888 while (c >= '0' && c <= '7')
1890 yyival = (yyival << 3) + c - '0';
1891 GET (c);
1894 int_suffixes:
1895 /* Integer suffixes. */
1896 while (isalpha (c))
1897 GET (c);
1898 UNGET ();
1899 return CINT;
1901 case '1': case '2': case '3': case '4': case '5': case '6':
1902 case '7': case '8': case '9':
1903 /* Integer or floating constant, part before '.'. */
1904 yyival = c - '0';
1906 while (GET (c) && DIGITP (c))
1907 yyival = 10 * yyival + c - '0';
1909 if (c != '.')
1910 goto int_suffixes;
1912 mantissa:
1913 /* Digits following '.'. */
1914 while (DIGITP (c))
1915 GET (c);
1917 /* Optional exponent. */
1918 if (c == 'E' || c == 'e')
1920 if (GET (c) == '-' || c == '+')
1921 GET (c);
1923 while (DIGITP (c))
1924 GET (c);
1927 /* Optional type suffixes. */
1928 while (isalpha (c))
1929 GET (c);
1930 UNGET ();
1931 return CFLOAT;
1933 default:
1934 break;
1940 /* Value is the string from the start of the line to the current
1941 position in the input buffer, or maybe a bit more if that string is
1942 shorter than min_regexp. */
1944 char *
1945 matching_regexp ()
1947 char *p;
1948 char *s;
1949 char *t;
1950 static char *buffer, *end_buf;
1952 if (!f_regexps)
1953 return NULL;
1955 if (buffer == NULL)
1957 buffer = (char *) xmalloc (max_regexp);
1958 end_buf = &buffer[max_regexp] - 1;
1961 /* Scan back to previous newline of buffer start. */
1962 for (p = in - 1; p > inbuffer && *p != '\n'; --p)
1965 if (*p == '\n')
1967 while (in - p < min_regexp && p > inbuffer)
1969 /* Line probably not significant enough */
1970 for (--p; p >= inbuffer && *p != '\n'; --p)
1973 if (*p == '\n')
1974 ++p;
1977 /* Copy from end to make sure significant portions are included.
1978 This implies that in the browser a regular expressing of the form
1979 `^.*{regexp}' has to be used. */
1980 for (s = end_buf - 1, t = in; s > buffer && t > p;)
1982 *--s = *--t;
1984 if (*s == '"')
1985 *--s = '\\';
1988 *(end_buf - 1) = '\0';
1989 return xstrdup (s);
1993 /* Return a printable representation of token T. */
1995 char *
1996 token_string (t)
1997 int t;
1999 static char b[3];
2001 switch (t)
2003 case CSTRING: return "string constant";
2004 case CCHAR: return "char constant";
2005 case CINT: return "int constant";
2006 case CFLOAT: return "floating constant";
2007 case ELLIPSIS: return "...";
2008 case LSHIFTASGN: return "<<=";
2009 case RSHIFTASGN: return ">>=";
2010 case ARROWSTAR: return "->*";
2011 case IDENT: return "identifier";
2012 case DIVASGN: return "/=";
2013 case INC: return "++";
2014 case ADDASGN: return "+=";
2015 case DEC: return "--";
2016 case ARROW: return "->";
2017 case SUBASGN: return "-=";
2018 case MULASGN: return "*=";
2019 case MODASGN: return "%=";
2020 case LOR: return "||";
2021 case ORASGN: return "|=";
2022 case LAND: return "&&";
2023 case ANDASGN: return "&=";
2024 case XORASGN: return "^=";
2025 case POINTSTAR: return ".*";
2026 case DCOLON: return "::";
2027 case EQ: return "==";
2028 case NE: return "!=";
2029 case LE: return "<=";
2030 case LSHIFT: return "<<";
2031 case GE: return ">=";
2032 case RSHIFT: return ">>";
2033 case ASM: return "asm";
2034 case AUTO: return "auto";
2035 case BREAK: return "break";
2036 case CASE: return "case";
2037 case CATCH: return "catch";
2038 case CHAR: return "char";
2039 case CLASS: return "class";
2040 case CONST: return "const";
2041 case CONTINUE: return "continue";
2042 case DEFAULT: return "default";
2043 case DELETE: return "delete";
2044 case DO: return "do";
2045 case DOUBLE: return "double";
2046 case ELSE: return "else";
2047 case ENUM: return "enum";
2048 case EXTERN: return "extern";
2049 case FLOAT: return "float";
2050 case FOR: return "for";
2051 case FRIEND: return "friend";
2052 case GOTO: return "goto";
2053 case IF: return "if";
2054 case T_INLINE: return "inline";
2055 case INT: return "int";
2056 case LONG: return "long";
2057 case NEW: return "new";
2058 case OPERATOR: return "operator";
2059 case PRIVATE: return "private";
2060 case PROTECTED: return "protected";
2061 case PUBLIC: return "public";
2062 case REGISTER: return "register";
2063 case RETURN: return "return";
2064 case SHORT: return "short";
2065 case SIGNED: return "signed";
2066 case SIZEOF: return "sizeof";
2067 case STATIC: return "static";
2068 case STRUCT: return "struct";
2069 case SWITCH: return "switch";
2070 case TEMPLATE: return "template";
2071 case THIS: return "this";
2072 case THROW: return "throw";
2073 case TRY: return "try";
2074 case TYPEDEF: return "typedef";
2075 case UNION: return "union";
2076 case UNSIGNED: return "unsigned";
2077 case VIRTUAL: return "virtual";
2078 case VOID: return "void";
2079 case VOLATILE: return "volatile";
2080 case WHILE: return "while";
2081 case MUTABLE: return "mutable";
2082 case BOOL: return "bool";
2083 case TRUE: return "true";
2084 case FALSE: return "false";
2085 case SIGNATURE: return "signature";
2086 case NAMESPACE: return "namespace";
2087 case EXPLICIT: return "explicit";
2088 case TYPENAME: return "typename";
2089 case CONST_CAST: return "const_cast";
2090 case DYNAMIC_CAST: return "dynamic_cast";
2091 case REINTERPRET_CAST: return "reinterpret_cast";
2092 case STATIC_CAST: return "static_cast";
2093 case TYPEID: return "typeid";
2094 case USING: return "using";
2095 case WCHAR: return "wchar_t";
2096 case YYEOF: return "EOF";
2098 default:
2099 if (t < 255)
2101 b[0] = t;
2102 b[1] = '\0';
2103 return b;
2105 else
2106 return "???";
2111 /* Reinitialize the scanner for a new input file. */
2113 void
2114 re_init_scanner ()
2116 in = inbuffer;
2117 yyline = 1;
2119 if (yytext == NULL)
2121 int size = 256;
2122 yytext = (char *) xmalloc (size * sizeof *yytext);
2123 yytext_end = yytext + size;
2128 /* Insert a keyword NAME with token value TK into the keyword hash
2129 table. */
2131 void
2132 insert_keyword (name, tk)
2133 char *name;
2134 int tk;
2136 char *s;
2137 unsigned h = 0;
2138 struct kw *k = (struct kw *) xmalloc (sizeof *k);
2140 for (s = name; *s; ++s)
2141 h = (h << 1) ^ *s;
2143 h %= KEYWORD_TABLE_SIZE;
2144 k->name = name;
2145 k->tk = tk;
2146 k->next = keyword_table[h];
2147 keyword_table[h] = k;
2151 /* Initialize the scanner for the first file. This sets up the
2152 character class vectors and fills the keyword hash table. */
2154 void
2155 init_scanner ()
2157 int i;
2159 /* Allocate the input buffer */
2160 inbuffer_size = READ_CHUNK_SIZE + 1;
2161 inbuffer = in = (char *) xmalloc (inbuffer_size);
2162 yyline = 1;
2164 /* Set up character class vectors. */
2165 for (i = 0; i < sizeof is_ident; ++i)
2167 if (i == '_' || isalnum (i))
2168 is_ident[i] = 1;
2170 if (i >= '0' && i <= '9')
2171 is_digit[i] = 1;
2173 if (i == ' ' || i == '\t' || i == '\f' || i == '\v')
2174 is_white[i] = 1;
2177 /* Fill keyword hash table. */
2178 insert_keyword ("and", LAND);
2179 insert_keyword ("and_eq", ANDASGN);
2180 insert_keyword ("asm", ASM);
2181 insert_keyword ("auto", AUTO);
2182 insert_keyword ("bitand", '&');
2183 insert_keyword ("bitor", '|');
2184 insert_keyword ("bool", BOOL);
2185 insert_keyword ("break", BREAK);
2186 insert_keyword ("case", CASE);
2187 insert_keyword ("catch", CATCH);
2188 insert_keyword ("char", CHAR);
2189 insert_keyword ("class", CLASS);
2190 insert_keyword ("compl", '~');
2191 insert_keyword ("const", CONST);
2192 insert_keyword ("const_cast", CONST_CAST);
2193 insert_keyword ("continue", CONTINUE);
2194 insert_keyword ("default", DEFAULT);
2195 insert_keyword ("delete", DELETE);
2196 insert_keyword ("do", DO);
2197 insert_keyword ("double", DOUBLE);
2198 insert_keyword ("dynamic_cast", DYNAMIC_CAST);
2199 insert_keyword ("else", ELSE);
2200 insert_keyword ("enum", ENUM);
2201 insert_keyword ("explicit", EXPLICIT);
2202 insert_keyword ("extern", EXTERN);
2203 insert_keyword ("false", FALSE);
2204 insert_keyword ("float", FLOAT);
2205 insert_keyword ("for", FOR);
2206 insert_keyword ("friend", FRIEND);
2207 insert_keyword ("goto", GOTO);
2208 insert_keyword ("if", IF);
2209 insert_keyword ("inline", T_INLINE);
2210 insert_keyword ("int", INT);
2211 insert_keyword ("long", LONG);
2212 insert_keyword ("mutable", MUTABLE);
2213 insert_keyword ("namespace", NAMESPACE);
2214 insert_keyword ("new", NEW);
2215 insert_keyword ("not", '!');
2216 insert_keyword ("not_eq", NE);
2217 insert_keyword ("operator", OPERATOR);
2218 insert_keyword ("or", LOR);
2219 insert_keyword ("or_eq", ORASGN);
2220 insert_keyword ("private", PRIVATE);
2221 insert_keyword ("protected", PROTECTED);
2222 insert_keyword ("public", PUBLIC);
2223 insert_keyword ("register", REGISTER);
2224 insert_keyword ("reinterpret_cast", REINTERPRET_CAST);
2225 insert_keyword ("return", RETURN);
2226 insert_keyword ("short", SHORT);
2227 insert_keyword ("signed", SIGNED);
2228 insert_keyword ("sizeof", SIZEOF);
2229 insert_keyword ("static", STATIC);
2230 insert_keyword ("static_cast", STATIC_CAST);
2231 insert_keyword ("struct", STRUCT);
2232 insert_keyword ("switch", SWITCH);
2233 insert_keyword ("template", TEMPLATE);
2234 insert_keyword ("this", THIS);
2235 insert_keyword ("throw", THROW);
2236 insert_keyword ("true", TRUE);
2237 insert_keyword ("try", TRY);
2238 insert_keyword ("typedef", TYPEDEF);
2239 insert_keyword ("typeid", TYPEID);
2240 insert_keyword ("typename", TYPENAME);
2241 insert_keyword ("union", UNION);
2242 insert_keyword ("unsigned", UNSIGNED);
2243 insert_keyword ("using", USING);
2244 insert_keyword ("virtual", VIRTUAL);
2245 insert_keyword ("void", VOID);
2246 insert_keyword ("volatile", VOLATILE);
2247 insert_keyword ("wchar_t", WCHAR);
2248 insert_keyword ("while", WHILE);
2249 insert_keyword ("xor", '^');
2250 insert_keyword ("xor_eq", XORASGN);
2255 /***********************************************************************
2256 Parser
2257 ***********************************************************************/
2259 /* Match the current lookahead token and set it to the next token. */
2261 #define MATCH() (tk = yylex ())
2263 /* Return the lookahead token. If current lookahead token is cleared,
2264 read a new token. */
2266 #define LA1 (tk == -1 ? (tk = yylex ()) : tk)
2268 /* Is the current lookahead equal to the token T? */
2270 #define LOOKING_AT(T) (tk == (T))
2272 /* Is the current lookahead one of T1 or T2? */
2274 #define LOOKING_AT2(T1, T2) (tk == (T1) || tk == (T2))
2276 /* Is the current lookahead one of T1, T2 or T3? */
2278 #define LOOKING_AT3(T1, T2, T3) (tk == (T1) || tk == (T2) || tk == (T3))
2280 /* Is the current lookahead one of T1...T4? */
2282 #define LOOKING_AT4(T1, T2, T3, T4) \
2283 (tk == (T1) || tk == (T2) || tk == (T3) || tk == (T4))
2285 /* Match token T if current lookahead is T. */
2287 #define MATCH_IF(T) if (LOOKING_AT (T)) MATCH (); else ((void) 0)
2289 /* Skip to matching token if current token is T. */
2291 #define SKIP_MATCHING_IF(T) \
2292 if (LOOKING_AT (T)) skip_matching (); else ((void) 0)
2295 /* Skip forward until a given token TOKEN or YYEOF is seen and return
2296 the current lookahead token after skipping. */
2299 skip_to (token)
2300 int token;
2302 while (!LOOKING_AT2 (YYEOF, token))
2303 MATCH ();
2304 return tk;
2308 /* Skip over pairs of tokens (parentheses, square brackets,
2309 angle brackets, curly brackets) matching the current lookahead. */
2311 void
2312 skip_matching ()
2314 int open, close, n;
2316 switch (open = LA1)
2318 case '{':
2319 close = '}';
2320 break;
2322 case '(':
2323 close = ')';
2324 break;
2326 case '<':
2327 close = '>';
2328 break;
2330 case '[':
2331 close = ']';
2332 break;
2334 default:
2335 abort ();
2338 for (n = 0;;)
2340 if (LOOKING_AT (open))
2341 ++n;
2342 else if (LOOKING_AT (close))
2343 --n;
2344 else if (LOOKING_AT (YYEOF))
2345 break;
2347 MATCH ();
2349 if (n == 0)
2350 break;
2355 /* Re-initialize the parser by resetting the lookahead token. */
2357 void
2358 re_init_parser ()
2360 tk = -1;
2364 /* Parse a parameter list, including the const-specifier,
2365 pure-specifier, and throw-list that may follow a parameter list.
2366 Return in FLAGS what was seen following the parameter list.
2367 Returns a hash code for the parameter types. This value is used to
2368 distinguish between overloaded functions. */
2370 unsigned
2371 parm_list (flags)
2372 int *flags;
2374 unsigned hash = 0;
2375 int type_seen = 0;
2377 while (!LOOKING_AT2 (YYEOF, ')'))
2379 switch (LA1)
2381 /* Skip over grouping parens or parameter lists in parameter
2382 declarations. */
2383 case '(':
2384 skip_matching ();
2385 break;
2387 /* Next parameter. */
2388 case ',':
2389 MATCH ();
2390 type_seen = 0;
2391 break;
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)
2401 char *last_id;
2402 unsigned ident_type_hash = 0;
2404 parse_qualified_param_ident_or_type (&last_id);
2405 if (last_id)
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;
2414 else
2415 MATCH ();
2416 break;
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;
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;
2437 case '*': case '&': case '[': case ']':
2438 hash = (hash << 1) ^ LA1;
2439 MATCH ();
2440 break;
2442 default:
2443 MATCH ();
2444 break;
2448 if (LOOKING_AT (')'))
2450 MATCH ();
2452 if (LOOKING_AT (CONST))
2454 /* We can overload the same function on `const' */
2455 hash = (hash << 1) ^ CONST;
2456 SET_FLAG (*flags, F_CONST);
2457 MATCH ();
2460 if (LOOKING_AT (THROW))
2462 MATCH ();
2463 SKIP_MATCHING_IF ('(');
2464 SET_FLAG (*flags, F_THROW);
2467 if (LOOKING_AT ('='))
2469 MATCH ();
2470 if (LOOKING_AT (CINT) && yyival == 0)
2472 MATCH ();
2473 SET_FLAG (*flags, F_PURE);
2478 return hash;
2482 /* Print position info to stdout. */
2484 void
2485 print_info ()
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);
2495 /* Parse a member declaration within the class body of CLS. VIS is
2496 the access specifier for the member (private, protected,
2497 public). */
2499 void
2500 member (cls, vis)
2501 struct sym *cls;
2502 int vis;
2504 char *id = NULL;
2505 int sc = SC_MEMBER;
2506 char *regexp = NULL;
2507 int pos;
2508 int is_constructor;
2509 int anonymous = 0;
2510 int flags = 0;
2511 int class_tag;
2512 int type_seen = 0;
2513 int paren_seen = 0;
2514 unsigned hash = 0;
2515 int tilde = 0;
2517 while (!LOOKING_AT4 (';', '{', '}', YYEOF))
2519 switch (LA1)
2521 default:
2522 MATCH ();
2523 break;
2525 /* A function or class may follow. */
2526 case TEMPLATE:
2527 MATCH();
2528 SET_FLAG (flags, F_TEMPLATE);
2529 /* Skip over template argument list */
2530 SKIP_MATCHING_IF ('<');
2531 break;
2533 case EXPLICIT:
2534 SET_FLAG (flags, F_EXPLICIT);
2535 goto typeseen;
2537 case MUTABLE:
2538 SET_FLAG (flags, F_MUTABLE);
2539 goto typeseen;
2541 case T_INLINE:
2542 SET_FLAG (flags, F_INLINE);
2543 goto typeseen;
2545 case VIRTUAL:
2546 SET_FLAG (flags, F_VIRTUAL);
2547 goto typeseen;
2549 case '[':
2550 skip_matching ();
2551 break;
2553 case ENUM:
2554 sc = SC_TYPE;
2555 goto typeseen;
2557 case TYPEDEF:
2558 sc = SC_TYPE;
2559 goto typeseen;
2561 case FRIEND:
2562 sc = SC_FRIEND;
2563 goto typeseen;
2565 case STATIC:
2566 sc = SC_STATIC;
2567 goto typeseen;
2569 case '~':
2570 tilde = 1;
2571 MATCH ();
2572 break;
2574 case IDENT:
2575 /* Remember IDENTS seen so far. Among these will be the member
2576 name. */
2577 id = (char *) alloca (strlen (yytext) + 2);
2578 if (tilde)
2580 *id = '~';
2581 strcpy (id + 1, yytext);
2583 else
2584 strcpy (id, yytext);
2585 MATCH ();
2586 break;
2588 case OPERATOR:
2589 id = operator_name (&sc);
2590 break;
2592 case '(':
2593 /* Most probably the beginning of a parameter list. */
2594 MATCH ();
2595 paren_seen = 1;
2597 if (id && cls)
2599 if (!(is_constructor = streq (id, cls->name)))
2600 regexp = matching_regexp ();
2602 else
2603 is_constructor = 0;
2605 pos = BUFFER_POS ();
2606 hash = parm_list (&flags);
2608 if (is_constructor)
2609 regexp = matching_regexp ();
2611 if (id && cls != NULL)
2612 add_member_decl (cls, id, regexp, pos, hash, 0, sc, vis, flags);
2614 while (!LOOKING_AT3 (';', '{', YYEOF))
2615 MATCH ();
2617 if (LOOKING_AT ('{') && id && cls)
2618 add_member_defn (cls, id, regexp, pos, hash, 0, sc, flags);
2620 id = NULL;
2621 sc = SC_MEMBER;
2622 break;
2624 case STRUCT: case UNION: case CLASS:
2625 /* Nested class */
2626 class_tag = LA1;
2627 type_seen = 1;
2628 MATCH ();
2629 anonymous = 1;
2631 /* More than one ident here to allow for MS-DOS specialties
2632 like `_export class' etc. The last IDENT seen counts
2633 as the class name. */
2634 while (!LOOKING_AT4 (YYEOF, ';', ':', '{'))
2636 if (LOOKING_AT (IDENT))
2637 anonymous = 0;
2638 MATCH ();
2641 if (LOOKING_AT2 (':', '{'))
2642 class_definition (anonymous ? NULL : cls, class_tag, flags, 1);
2643 else
2644 skip_to (';');
2645 break;
2647 case INT: case CHAR: case LONG: case UNSIGNED:
2648 case SIGNED: case CONST: case DOUBLE: case VOID:
2649 case SHORT: case VOLATILE: case BOOL: case WCHAR:
2650 case TYPENAME:
2651 typeseen:
2652 type_seen = 1;
2653 MATCH ();
2654 break;
2658 if (LOOKING_AT (';'))
2660 /* The end of a member variable, a friend declaration or an access
2661 declaration. We don't want to add friend classes as members. */
2662 if (id && sc != SC_FRIEND && cls)
2664 regexp = matching_regexp ();
2665 pos = BUFFER_POS ();
2667 if (cls != NULL)
2669 if (type_seen || !paren_seen)
2670 add_member_decl (cls, id, regexp, pos, 0, 1, sc, vis, 0);
2671 else
2672 add_member_decl (cls, id, regexp, pos, hash, 0, sc, vis, 0);
2676 MATCH ();
2677 print_info ();
2679 else if (LOOKING_AT ('{'))
2681 /* A named enum. */
2682 if (sc == SC_TYPE && id && cls)
2684 regexp = matching_regexp ();
2685 pos = BUFFER_POS ();
2687 if (cls != NULL)
2689 add_member_decl (cls, id, regexp, pos, 0, 1, sc, vis, 0);
2690 add_member_defn (cls, id, regexp, pos, 0, 1, sc, 0);
2694 skip_matching ();
2695 print_info ();
2700 /* Parse the body of class CLS. TAG is the tag of the class (struct,
2701 union, class). */
2703 void
2704 class_body (cls, tag)
2705 struct sym *cls;
2706 int tag;
2708 int vis = tag == CLASS ? PRIVATE : PUBLIC;
2709 int temp;
2711 while (!LOOKING_AT2 (YYEOF, '}'))
2713 switch (LA1)
2715 case PRIVATE: case PROTECTED: case PUBLIC:
2716 temp = LA1;
2717 MATCH ();
2719 if (LOOKING_AT (':'))
2721 vis = temp;
2722 MATCH ();
2724 else
2726 /* Probably conditional compilation for inheritance list.
2727 We don't known whether there comes more of this.
2728 This is only a crude fix that works most of the time. */
2731 MATCH ();
2733 while (LOOKING_AT2 (IDENT, ',')
2734 || LOOKING_AT3 (PUBLIC, PROTECTED, PRIVATE));
2736 break;
2738 case TYPENAME:
2739 case USING:
2740 skip_to (';');
2741 break;
2743 /* Try to synchronize */
2744 case CHAR: case CLASS: case CONST:
2745 case DOUBLE: case ENUM: case FLOAT: case INT:
2746 case LONG: case SHORT: case SIGNED: case STRUCT:
2747 case UNION: case UNSIGNED: case VOID: case VOLATILE:
2748 case TYPEDEF: case STATIC: case T_INLINE: case FRIEND:
2749 case VIRTUAL: case TEMPLATE: case IDENT: case '~':
2750 case BOOL: case WCHAR: case EXPLICIT: case MUTABLE:
2751 member (cls, vis);
2752 break;
2754 default:
2755 MATCH ();
2756 break;
2762 /* Parse a qualified identifier. Current lookahead is IDENT. A
2763 qualified ident has the form `X<..>::Y<...>::T<...>. Returns a
2764 symbol for that class. */
2766 struct sym *
2767 parse_classname ()
2769 struct sym *last_class = NULL;
2771 while (LOOKING_AT (IDENT))
2773 last_class = add_sym (yytext, last_class);
2774 MATCH ();
2776 if (LOOKING_AT ('<'))
2778 skip_matching ();
2779 SET_FLAG (last_class->flags, F_TEMPLATE);
2782 if (!LOOKING_AT (DCOLON))
2783 break;
2785 MATCH ();
2788 return last_class;
2792 /* Parse an operator name. Add the `static' flag to *SC if an
2793 implicitly static operator has been parsed. Value is a pointer to
2794 a static buffer holding the constructed operator name string. */
2796 char *
2797 operator_name (sc)
2798 int *sc;
2800 static int id_size = 0;
2801 static char *id = NULL;
2802 char *s;
2803 int len;
2805 MATCH ();
2807 if (LOOKING_AT2 (NEW, DELETE))
2809 /* `new' and `delete' are implicitly static. */
2810 if (*sc != SC_FRIEND)
2811 *sc = SC_STATIC;
2813 s = token_string (LA1);
2814 MATCH ();
2816 len = strlen (s) + 10;
2817 if (len > id_size)
2819 int new_size = max (len, 2 * id_size);
2820 id = (char *) xrealloc (id, new_size);
2821 id_size = new_size;
2823 strcpy (id, s);
2825 /* Vector new or delete? */
2826 if (LOOKING_AT ('['))
2828 strcat (id, "[");
2829 MATCH ();
2831 if (LOOKING_AT (']'))
2833 strcat (id, "]");
2834 MATCH ();
2838 else
2840 int tokens_matched = 0;
2842 len = 20;
2843 if (len > id_size)
2845 int new_size = max (len, 2 * id_size);
2846 id = (char *) xrealloc (id, new_size);
2847 id_size = new_size;
2849 strcpy (id, "operator");
2851 /* Beware access declarations of the form "X::f;" Beware of
2852 `operator () ()'. Yet another difficulty is found in
2853 GCC 2.95's STL: `operator == __STL_NULL_TMPL_ARGS (...'. */
2854 while (!(LOOKING_AT ('(') && tokens_matched)
2855 && !LOOKING_AT2 (';', YYEOF))
2857 s = token_string (LA1);
2858 len += strlen (s) + 2;
2859 if (len > id_size)
2861 int new_size = max (len, 2 * id_size);
2862 id = (char *) xrealloc (id, new_size);
2863 id_size = new_size;
2866 if (*s != ')' && *s != ']')
2867 strcat (id, " ");
2868 strcat (id, s);
2869 MATCH ();
2871 /* If this is a simple operator like `+', stop now. */
2872 if (!isalpha (*s) && *s != '(' && *s != '[')
2873 break;
2875 ++tokens_matched;
2879 return id;
2883 /* This one consumes the last IDENT of a qualified member name like
2884 `X::Y::z'. This IDENT is returned in LAST_ID. Value if the
2885 symbol structure for the ident. */
2887 struct sym *
2888 parse_qualified_ident_or_type (last_id)
2889 char **last_id;
2891 struct sym *cls = NULL;
2892 static char *id = NULL;
2893 static int id_size = 0;
2895 while (LOOKING_AT (IDENT))
2897 int len = strlen (yytext) + 1;
2898 if (len > id_size)
2900 id = (char *) xrealloc (id, len);
2901 id_size = len;
2903 strcpy (id, yytext);
2904 *last_id = id;
2905 MATCH ();
2907 SKIP_MATCHING_IF ('<');
2909 if (LOOKING_AT (DCOLON))
2911 cls = add_sym (id, cls);
2912 *last_id = NULL;
2913 MATCH ();
2915 else
2916 break;
2919 return cls;
2923 /* This one consumes the last IDENT of a qualified member name like
2924 `X::Y::z'. This IDENT is returned in LAST_ID. Value if the
2925 symbol structure for the ident. */
2927 void
2928 parse_qualified_param_ident_or_type (last_id)
2929 char **last_id;
2931 struct sym *cls = NULL;
2932 static char *id = NULL;
2933 static int id_size = 0;
2935 while (LOOKING_AT (IDENT))
2937 int len = strlen (yytext) + 1;
2938 if (len > id_size)
2940 id = (char *) xrealloc (id, len);
2941 id_size = len;
2943 strcpy (id, yytext);
2944 *last_id = id;
2945 MATCH ();
2947 SKIP_MATCHING_IF ('<');
2949 if (LOOKING_AT (DCOLON))
2951 cls = add_sym (id, cls);
2952 *last_id = NULL;
2953 MATCH ();
2955 else
2956 break;
2961 /* Parse a class definition.
2963 CONTAINING is the class containing the class being parsed or null.
2964 This may also be null if NESTED != 0 if the containing class is
2965 anonymous. TAG is the tag of the class (struct, union, class).
2966 NESTED is non-zero if we are parsing a nested class.
2968 Current lookahead is the class name. */
2970 void
2971 class_definition (containing, tag, flags, nested)
2972 struct sym *containing;
2973 int tag;
2974 int flags;
2975 int nested;
2977 struct sym *current;
2978 struct sym *base_class;
2980 /* Set CURRENT to null if no entry has to be made for the class
2981 parsed. This is the case for certain command line flag
2982 settings. */
2983 if ((tag != CLASS && !f_structs) || (nested && !f_nested_classes))
2984 current = NULL;
2985 else
2987 current = add_sym (yytext, containing);
2988 current->pos = BUFFER_POS ();
2989 current->regexp = matching_regexp ();
2990 current->filename = filename;
2991 current->flags = flags;
2994 /* If at ':', base class list follows. */
2995 if (LOOKING_AT (':'))
2997 int done = 0;
2998 MATCH ();
3000 while (!done)
3002 switch (LA1)
3004 case VIRTUAL: case PUBLIC: case PROTECTED: case PRIVATE:
3005 MATCH ();
3006 break;
3008 case IDENT:
3009 base_class = parse_classname ();
3010 if (base_class && current && base_class != current)
3011 add_link (base_class, current);
3012 break;
3014 /* The `,' between base classes or the end of the base
3015 class list. Add the previously found base class.
3016 It's done this way to skip over sequences of
3017 `A::B::C' until we reach the end.
3019 FIXME: it is now possible to handle `class X : public B::X'
3020 because we have enough information. */
3021 case ',':
3022 MATCH ();
3023 break;
3025 default:
3026 /* A syntax error, possibly due to preprocessor constructs
3027 like
3029 #ifdef SOMETHING
3030 class A : public B
3031 #else
3032 class A : private B.
3034 MATCH until we see something like `;' or `{'. */
3035 while (!LOOKING_AT3 (';', YYEOF, '{'))
3036 MATCH ();
3037 done = 1;
3039 case '{':
3040 done = 1;
3041 break;
3046 /* Parse the class body if there is one. */
3047 if (LOOKING_AT ('{'))
3049 if (tag != CLASS && !f_structs)
3050 skip_matching ();
3051 else
3053 MATCH ();
3054 class_body (current, tag);
3056 if (LOOKING_AT ('}'))
3058 MATCH ();
3059 if (LOOKING_AT (';') && !nested)
3060 MATCH ();
3067 /* Parse a declaration. */
3069 void
3070 declaration (flags)
3071 int flags;
3073 char *id = NULL;
3074 struct sym *cls = NULL;
3075 char *regexp = NULL;
3076 int pos = 0;
3077 unsigned hash = 0;
3078 int is_constructor;
3079 int sc = 0;
3081 while (!LOOKING_AT3 (';', '{', YYEOF))
3083 switch (LA1)
3085 default:
3086 MATCH ();
3087 break;
3089 case '[':
3090 skip_matching ();
3091 break;
3093 case ENUM:
3094 case TYPEDEF:
3095 sc = SC_TYPE;
3096 MATCH ();
3097 break;
3099 case STATIC:
3100 sc = SC_STATIC;
3101 MATCH ();
3102 break;
3104 case INT: case CHAR: case LONG: case UNSIGNED:
3105 case SIGNED: case CONST: case DOUBLE: case VOID:
3106 case SHORT: case VOLATILE: case BOOL: case WCHAR:
3107 MATCH ();
3108 break;
3110 case CLASS: case STRUCT: case UNION:
3111 /* This is for the case `STARTWRAP class X : ...' or
3112 `declare (X, Y)\n class A : ...'. */
3113 if (id)
3114 return;
3116 case '=':
3117 /* Assumed to be the start of an initialization in this context.
3118 Skip over everything up to ';'. */
3119 skip_to (';');
3120 break;
3122 case OPERATOR:
3123 id = operator_name (&sc);
3124 break;
3126 case T_INLINE:
3127 SET_FLAG (flags, F_INLINE);
3128 MATCH ();
3129 break;
3131 case '~':
3132 MATCH ();
3133 if (LOOKING_AT (IDENT))
3135 id = (char *) alloca (strlen (yytext) + 2);
3136 *id = '~';
3137 strcpy (id + 1, yytext);
3138 MATCH ();
3140 break;
3142 case IDENT:
3143 cls = parse_qualified_ident_or_type (&id);
3144 break;
3146 case '(':
3147 /* Most probably the beginning of a parameter list. */
3148 if (cls)
3150 MATCH ();
3152 if (id && cls)
3154 if (!(is_constructor = streq (id, cls->name)))
3155 regexp = matching_regexp ();
3157 else
3158 is_constructor = 0;
3160 pos = BUFFER_POS ();
3161 hash = parm_list (&flags);
3163 if (is_constructor)
3164 regexp = matching_regexp ();
3166 if (id && cls)
3167 add_member_defn (cls, id, regexp, pos, hash, 0,
3168 SC_UNKNOWN, flags);
3170 else
3172 /* This may be a C functions, but also a macro
3173 call of the form `declare (A, B)' --- such macros
3174 can be found in some class libraries. */
3175 MATCH ();
3177 if (id)
3179 regexp = matching_regexp ();
3180 pos = BUFFER_POS ();
3181 hash = parm_list (&flags);
3182 add_global_decl (id, regexp, pos, hash, 0, sc, flags);
3185 /* This is for the case that the function really is
3186 a macro with no `;' following it. If a CLASS directly
3187 follows, we would miss it otherwise. */
3188 if (LOOKING_AT3 (CLASS, STRUCT, UNION))
3189 return;
3192 while (!LOOKING_AT3 (';', '{', YYEOF))
3193 MATCH ();
3195 if (!cls && id && LOOKING_AT ('{'))
3196 add_global_defn (id, regexp, pos, hash, 0, sc, flags);
3197 id = NULL;
3198 break;
3202 if (LOOKING_AT (';'))
3204 /* The end of a member variable or of an access declaration
3205 `X::f'. To distinguish between them we have to know whether
3206 type information has been seen. */
3207 if (id)
3209 char *regexp = matching_regexp ();
3210 int pos = BUFFER_POS ();
3212 if (cls)
3213 add_member_defn (cls, id, regexp, pos, 0, 1, SC_UNKNOWN, flags);
3214 else
3215 add_global_defn (id, regexp, pos, 0, 1, sc, flags);
3218 MATCH ();
3219 print_info ();
3221 else if (LOOKING_AT ('{'))
3223 if (sc == SC_TYPE && id)
3225 /* A named enumeration. */
3226 regexp = matching_regexp ();
3227 pos = BUFFER_POS ();
3228 add_global_defn (id, regexp, pos, 0, 1, sc, flags);
3231 skip_matching ();
3232 print_info ();
3237 /* Parse a list of top-level declarations/definitions. START_FLAGS
3238 says in which context we are parsing. If it is F_EXTERNC, we are
3239 parsing in an `extern "C"' block. Value is 1 if EOF is reached, 0
3240 otherwise. */
3243 globals (start_flags)
3244 int start_flags;
3246 int anonymous;
3247 int class_tk;
3248 int flags = start_flags;
3250 for (;;)
3252 char *prev_in = in;
3254 switch (LA1)
3256 case NAMESPACE:
3258 MATCH ();
3260 if (LOOKING_AT (IDENT))
3262 char *namespace_name
3263 = (char *) alloca (strlen (yytext) + 1);
3264 strcpy (namespace_name, yytext);
3265 MATCH ();
3267 if (LOOKING_AT ('='))
3269 if (skip_to (';') == ';')
3270 MATCH ();
3271 register_namespace_alias (namespace_name, yytext);
3273 else if (LOOKING_AT ('{'))
3275 MATCH ();
3276 enter_namespace (namespace_name);
3277 globals (0);
3278 leave_namespace ();
3279 MATCH_IF ('}');
3283 break;
3285 case EXTERN:
3286 MATCH ();
3287 if (LOOKING_AT (CSTRING) && *string_start == 'C'
3288 && *(string_start + 1) == '"')
3290 /* This is `extern "C"'. */
3291 MATCH ();
3293 if (LOOKING_AT ('{'))
3295 MATCH ();
3296 globals (F_EXTERNC);
3297 MATCH_IF ('}');
3299 else
3300 SET_FLAG (flags, F_EXTERNC);
3302 break;
3304 case TEMPLATE:
3305 MATCH ();
3306 SKIP_MATCHING_IF ('<');
3307 SET_FLAG (flags, F_TEMPLATE);
3308 break;
3310 case CLASS: case STRUCT: case UNION:
3311 class_tk = LA1;
3312 MATCH ();
3313 anonymous = 1;
3315 /* More than one ident here to allow for MS-DOS and OS/2
3316 specialties like `far', `_Export' etc. Some C++ libs
3317 have constructs like `_OS_DLLIMPORT(_OS_CLIENT)' in front
3318 of the class name. */
3319 while (!LOOKING_AT4 (YYEOF, ';', ':', '{'))
3321 if (LOOKING_AT (IDENT))
3322 anonymous = 0;
3323 MATCH ();
3326 /* Don't add anonymous unions. */
3327 if (LOOKING_AT2 (':', '{') && !anonymous)
3328 class_definition (NULL, class_tk, flags, 0);
3329 else
3331 if (skip_to (';') == ';')
3332 MATCH ();
3335 flags = start_flags;
3336 break;
3338 case YYEOF:
3339 return 1;
3341 case '}':
3342 return 0;
3344 default:
3345 declaration (flags);
3346 flags = start_flags;
3347 break;
3350 if (prev_in == in)
3351 yyerror ("parse error");
3356 /* Parse the current input file. */
3358 void
3359 yyparse ()
3361 while (globals (0) == 0)
3362 MATCH_IF ('}');
3367 /***********************************************************************
3368 Main Program
3369 ***********************************************************************/
3371 /* Add the list of paths PATH_LIST to the current search path for
3372 input files. */
3374 void
3375 add_search_path (path_list)
3376 char *path_list;
3378 while (*path_list)
3380 char *start = path_list;
3381 struct search_path *p;
3383 while (*path_list && *path_list != PATH_LIST_SEPARATOR)
3384 ++path_list;
3386 p = (struct search_path *) xmalloc (sizeof *p);
3387 p->path = (char *) xmalloc (path_list - start + 1);
3388 memcpy (p->path, start, path_list - start);
3389 p->path[path_list - start] = '\0';
3390 p->next = NULL;
3392 if (search_path_tail)
3394 search_path_tail->next = p;
3395 search_path_tail = p;
3397 else
3398 search_path = search_path_tail = p;
3400 while (*path_list == PATH_LIST_SEPARATOR)
3401 ++path_list;
3406 /* Open FILE and return a file handle for it, or -1 if FILE cannot be
3407 opened. Try to find FILE in search_path first, then try the
3408 unchanged file name. */
3410 FILE *
3411 open_file (file)
3412 char *file;
3414 FILE *fp = NULL;
3415 static char *buffer;
3416 static int buffer_size;
3417 struct search_path *path;
3418 int flen = strlen (file) + 1; /* +1 for the slash */
3420 filename = xstrdup (file);
3422 for (path = search_path; path && fp == NULL; path = path->next)
3424 int len = strlen (path->path) + flen;
3426 if (len + 1 >= buffer_size)
3428 buffer_size = max (len + 1, 2 * buffer_size);
3429 buffer = (char *) xrealloc (buffer, buffer_size);
3432 strcpy (buffer, path->path);
3433 strcat (buffer, "/");
3434 strcat (buffer, file);
3435 fp = fopen (buffer, "r");
3438 /* Try the original file name. */
3439 if (fp == NULL)
3440 fp = fopen (file, "r");
3442 if (fp == NULL)
3443 yyerror ("cannot open");
3445 return fp;
3449 /* Display usage information and exit program. */
3451 #define USAGE "\
3452 Usage: ebrowse [options] {files}\n\
3454 -a, --append append output\n\
3455 -f, --files=FILES read input file names from FILE\n\
3456 -I, --search-path=LIST set search path for input files\n\
3457 -m, --min-regexp-length=N set minimum regexp length to N\n\
3458 -M, --max-regexp-length=N set maximum regexp length to N\n\
3459 -n, --no-nested-classes exclude nested classes\n\
3460 -o, --output-file=FILE set output file name to FILE\n\
3461 -p, --position-info print info about position in file\n\
3462 -s, --no-structs-or-unions don't record structs or unions\n\
3463 -v, --verbose be verbose\n\
3464 -V, --very-verbose be very verbose\n\
3465 -x, --no-regexps don't record regular expressions\n\
3466 --help display this help\n\
3467 --version display version info\n\
3470 void
3471 usage (error)
3472 int error;
3474 puts (USAGE);
3475 exit (error ? 1 : 0);
3479 /* Display version and copyright info. The VERSION macro is set
3480 from the Makefile and contains the Emacs version. */
3482 void
3483 version ()
3485 printf ("ebrowse %s\n", VERSION);
3486 puts ("Copyright (C) 1992-1999, 2000 Free Software Foundation, Inc.");
3487 puts ("This program is distributed under the same terms as Emacs.");
3488 exit (0);
3492 /* Parse one input file FILE, adding classes and members to the symbol
3493 table. */
3495 void
3496 process_file (file)
3497 char *file;
3499 FILE *fp;
3501 fp = open_file (file);
3502 if (fp)
3504 int nread, nbytes;
3506 /* Give a progress indication if needed. */
3507 if (f_very_verbose)
3509 puts (filename);
3510 fflush (stdout);
3512 else if (f_verbose)
3514 putchar ('.');
3515 fflush (stdout);
3518 /* Read file to inbuffer. */
3519 for (nread = 0;;)
3521 if (nread + READ_CHUNK_SIZE >= inbuffer_size)
3523 inbuffer_size = nread + READ_CHUNK_SIZE + 1;
3524 inbuffer = (char *) xrealloc (inbuffer, inbuffer_size);
3527 nbytes = fread (inbuffer + nread, 1, READ_CHUNK_SIZE, fp);
3528 if (nbytes <= 0)
3529 break;
3530 nread += nbytes;
3532 if (nread < 0)
3533 nread = 0;
3534 inbuffer[nread] = '\0';
3536 /* Reinitialize scanner and parser for the new input file. */
3537 re_init_scanner ();
3538 re_init_parser ();
3540 /* Parse it and close the file. */
3541 yyparse ();
3542 fclose (fp);
3547 /* Read a line from stream FP and return a pointer to a static buffer
3548 containing its contents without the terminating newline. Value
3549 is null when EOF is reached. */
3551 char *
3552 read_line (fp)
3553 FILE *fp;
3555 static char *buffer;
3556 static int buffer_size;
3557 int i = 0, c;
3559 while ((c = getc (fp)) != EOF && c != '\n')
3561 if (i >= buffer_size)
3563 buffer_size = max (100, buffer_size * 2);
3564 buffer = (char *) xrealloc (buffer, buffer_size);
3567 buffer[i++] = c;
3570 if (c == EOF && i == 0)
3571 return NULL;
3573 if (i == buffer_size)
3575 buffer_size = max (100, buffer_size * 2);
3576 buffer = (char *) xrealloc (buffer, buffer_size);
3579 buffer[i] = '\0';
3580 return buffer;
3584 /* Main entry point. */
3587 main (argc, argv)
3588 int argc;
3589 char **argv;
3591 int i;
3592 int any_inputfiles = 0;
3593 static char *out_filename = DEFAULT_OUTFILE;
3594 static char **input_filenames = NULL;
3595 static int input_filenames_size = 0;
3596 static int n_input_files;
3598 filename = "command line";
3599 yyout = stdout;
3601 while ((i = getopt_long (argc, argv, "af:I:m:M:no:p:svVx",
3602 options, NULL)) != EOF)
3604 switch (i)
3606 /* Experimental. */
3607 case 'p':
3608 info_position = atoi (optarg);
3609 break;
3611 case 'n':
3612 f_nested_classes = 0;
3613 break;
3615 case 'x':
3616 f_regexps = 0;
3617 break;
3619 /* Add the name of a file containing more input files. */
3620 case 'f':
3621 if (n_input_files == input_filenames_size)
3623 input_filenames_size = max (10, 2 * input_filenames_size);
3624 input_filenames = (char **) xrealloc (input_filenames,
3625 input_filenames_size);
3627 input_filenames[n_input_files++] = xstrdup (optarg);
3628 break;
3630 /* Append new output to output file instead of truncating it. */
3631 case 'a':
3632 f_append = 1;
3633 break;
3635 /* Include structs in the output */
3636 case 's':
3637 f_structs = 0;
3638 break;
3640 /* Be verbose (give a progress indication). */
3641 case 'v':
3642 f_verbose = 1;
3643 break;
3645 /* Be very verbose (print file names as they are processed). */
3646 case 'V':
3647 f_verbose = 1;
3648 f_very_verbose = 1;
3649 break;
3651 /* Change the name of the output file. */
3652 case 'o':
3653 out_filename = optarg;
3654 break;
3656 /* Set minimum length for regular expression strings
3657 when recorded in the output file. */
3658 case 'm':
3659 min_regexp = atoi (optarg);
3660 break;
3662 /* Set maximum length for regular expression strings
3663 when recorded in the output file. */
3664 case 'M':
3665 max_regexp = atoi (optarg);
3666 break;
3668 /* Add to search path. */
3669 case 'I':
3670 add_search_path (optarg);
3671 break;
3673 /* Display help */
3674 case -2:
3675 usage (0);
3676 break;
3678 case -3:
3679 version ();
3680 break;
3684 /* Call init_scanner after command line flags have been processed to be
3685 able to add keywords depending on command line (not yet
3686 implemented). */
3687 init_scanner ();
3688 init_sym ();
3690 /* Open output file */
3691 if (*out_filename)
3693 yyout = fopen (out_filename, f_append ? "a" : "w");
3694 if (yyout == NULL)
3696 yyerror ("cannot open output file `%s'", out_filename);
3697 exit (1);
3701 /* Process input files specified on the command line. */
3702 while (optind < argc)
3704 process_file (argv[optind++]);
3705 any_inputfiles = 1;
3708 /* Process files given on stdin if no files specified. */
3709 if (!any_inputfiles && n_input_files == 0)
3711 char *file;
3712 while ((file = read_line (stdin)) != NULL)
3713 process_file (file);
3715 else
3717 /* Process files from `--files=FILE'. Every line in FILE names
3718 one input file to process. */
3719 for (i = 0; i < n_input_files; ++i)
3721 FILE *fp = fopen (input_filenames[i], "r");
3723 if (fp == NULL)
3724 yyerror ("cannot open input file `%s'", input_filenames[i]);
3725 else
3727 char *file;
3728 while ((file = read_line (fp)) != NULL)
3729 process_file (file);
3730 fclose (fp);
3735 /* Write output file. */
3736 dump_roots (yyout);
3738 /* Close output file. */
3739 if (yyout != stdout)
3740 fclose (yyout);
3742 return 0;
3746 /* ebrowse.c ends here. */