1 /* Tags file maker to go with GNU Emacs
2 Copyright (C) 1984, 87, 88, 89, 93, 94, 95
3 Free Software Foundation, Inc. and Ken Arnold
5 This file is not considered part of SXEmacs.
7 This program 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
10 (at your option) any later version.
12 This program 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.
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>. */
22 * Ctags originally by Ken Arnold.
23 * Fortran added by Jim Kleckner.
24 * Ed Pelegri-Llopart added C typedefs.
25 * Gnu Emacs TAGS format and modifications by RMS?
26 * Sam Kendall added C++.
27 * Francesco Potorti` reorganised C and C++ based on work by Joe Wells.
28 * Regexp tags by Tom Tromey.
30 * Francesco Potorti` (F.Potorti@cnuce.cnr.it) is the current maintainer.
33 char pot_etags_version[] = "@(#) pot revision number is 12.28";
35 /* Prototyping magic snarfed from gmalloc.c */
36 #if defined (__cplusplus) || defined (__STDC__)
40 #define __ptr_t void *
41 #else /* Not C++ or ANSI C. */
47 #define __ptr_t char *
48 #endif /* C++ or ANSI C. */
52 /* On some systems, Emacs defines static as nothing for the sake
53 of unexec. We don't want that here since we don't use unexec. */
55 # define ETAGS_REGEXPS /* use the regexp features */
56 #endif /* HAVE_CONFIG_H */
65 #if defined (STDC_HEADERS)
74 extern char *getcwd();
76 #endif /* HAVE_UNISTD_H */
81 #include <sys/types.h>
84 #if !defined (S_ISREG) && defined (S_IFREG)
85 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
94 # define getopt_long(argc,argv,optstr,lopts,lind) getopt (argc, argv, optstr)
96 extern int optind, opterr;
98 # error "ootags cannot be built without getopt, preferably getopt_long"
99 #endif /* LONG_OPTIONS */
103 #endif /* ETAGS_REGEXPS */
105 /* Define CTAGS to make the program "ctags" compatible with the usual one.
106 Leave it undefined to make the program "etags", which makes emacs-style
107 tag tables and tags typedefs, #defines and struct/union/enum by default. */
115 /* Exit codes for success and failure. */
125 #define C_PLPL 0x00001 /* C++ */
126 #define C_STAR 0x00003 /* C* */
127 #define C_JAVA 0x00005 /* JAVA */
128 #define YACC 0x10000 /* yacc file */
130 #define streq(s,t) ((DEBUG && (s) == NULL && (t) == NULL \
131 && (abort (), 1)) || !strcmp (s, t))
132 #define strneq(s,t,n) ((DEBUG && (s) == NULL && (t) == NULL \
133 && (abort (), 1)) || !strncmp (s, t, n))
135 #define lowcase(c) tolower ((char)c)
137 #define CHARS 256 /* 2^sizeof(char) */
138 #define CHAR(x) ((unsigned int)x & (CHARS - 1))
139 #define iswhite(c) (_wht[CHAR(c)]) /* c is white */
140 #define notinname(c) (_nin[CHAR(c)]) /* c is not in a name */
141 #define begtoken(c) (_btk[CHAR(c)]) /* c can start token */
142 #define intoken(c) (_itk[CHAR(c)]) /* c can be in token */
143 #define endtoken(c) (_etk[CHAR(c)]) /* c ends tokens */
146 /*#undef OO_BROWSER*/
147 /* Due to the way this file is constructed, this unfortunately doesn't */
148 /* work except for documentation purposes. -slb */
153 #define set_construct(construct) \
154 if (!oo_browser_construct) oo_browser_construct = construct
155 void oo_browser_clear_all_globals(void);
156 void oo_browser_clear_some_globals(void);
157 void oo_browser_check_and_clear_structtype(void);
161 * xnew, xrnew -- allocate, reallocate storage
163 * SYNOPSIS: Type *xnew (int n, Type);
164 * Type *xrnew (OldPointer, int n, Type);
167 # include "chkmalloc.h"
168 # define xnew(n,Type) ((Type *) trace_malloc (__FILE__, __LINE__, \
169 (n) * sizeof (Type)))
170 # define xrnew(op,n,Type) ((Type *) trace_realloc (__FILE__, __LINE__, \
171 (op), (n) * sizeof (Type)))
173 # define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type)))
174 # define xrnew(op,n,Type) ((Type *) xrealloc ((op), (n) * sizeof (Type)))
179 typedef void Lang_function(FILE *);
183 Lang_function *function;
188 typedef struct node_st { /* sorting structure */
189 char *name; /* function or type name */
191 short int construct; /* Construct type for the OO-Browser */
193 char *file; /* file name */
194 bool is_func; /* use pattern or line no */
195 bool been_warned; /* set if noticed dup */
196 int lno; /* line number tag is on */
197 long cno; /* character number line starts on */
198 char *pat; /* search pattern */
199 struct node_st *left, *right; /* left and right sons */
203 /* If you add to this array, you must add a corresponding entry to the
205 static char *oo_browser_default_classes[] =
206 /* Lack of square brackets around some of these entries are intentional. */
207 { "null", "class", "method", "[constant]", "[enumeration]", "[enum_label]",
208 "extern", "[function]", "[macro]", "objc", "[structure]", "[type]",
209 "[union]", "[variable]"
212 /* If you add to this enum, you must add a corresponding entry to the
214 enum oo_browser_constructs { C_NULL, C_CLASS, C_METHOD, C_CONSTANT,
216 C_ENUM_LABEL, C_EXTERN, C_FUNCTION, C_MACRO,
217 C_OBJC, C_STRUCTURE, C_TYPE, C_UNION, C_VARIABLE
220 enum oo_browser_constructs oo_browser_construct = C_NULL;
224 * A `linebuffer' is a structure which holds a line of text.
225 * `readline_internal' reads a line from a stream into a linebuffer
226 * and works regardless of the length of the line.
227 * SIZE is the size of BUFFER, LEN is the length of the string in
228 * BUFFER after readline reads it.
236 extern char *getenv PP((const char *envvar));
238 /* Many compilers barf on this:
239 Lang_function Asm_labels;
240 so let's write it this way */
241 void Asm_labels PP((FILE * inf));
242 void C_entries PP((int c_ext, FILE * inf));
243 void default_C_entries PP((FILE * inf));
244 void plain_C_entries PP((FILE * inf));
245 void Cjava_entries PP((FILE * inf));
246 void Cplusplus_entries PP((FILE * inf));
247 void Yacc_entries PP((FILE * inf));
248 void Cobol_paragraphs PP((FILE * inf));
249 void Cstar_entries PP((FILE * inf));
250 void Erlang_functions PP((FILE * inf));
251 void Fortran_functions PP((FILE * inf));
252 void Lisp_functions PP((FILE * inf));
253 void Pascal_functions PP((FILE * inf));
254 void Perl_functions PP((FILE * inf));
255 void Postscript_functions PP((FILE * inf));
256 void Prolog_functions PP((FILE * inf));
257 void Python_functions PP((FILE * inf));
258 void Scheme_functions PP((FILE * inf));
259 void TeX_functions PP((FILE * inf));
260 void just_read_file PP((FILE * inf));
262 void print_language_names PP((void));
263 void print_version PP((void));
264 void print_help PP((void));
266 language *get_language_from_name PP((char *name));
267 language *get_language_from_interpreter PP((char *interpreter));
268 language *get_language_from_suffix PP((char *suffix));
269 int total_size_of_entries PP((node * np));
270 long readline PP((linebuffer * lbp, FILE * stream));
271 long readline_internal PP((linebuffer * lbp, FILE * stream));
273 void analyse_regex PP((char *regex_arg));
274 void add_regex PP((char *regexp_pattern, language * lang));
275 void free_patterns PP((void));
276 #endif /* ETAGS_REGEXPS */
277 void error PP((const char *s1, const char *s2));
278 void suggest_asking_for_help PP((void));
279 void fatal PP((char *s1, char *s2));
280 void pfatal PP((char *s1));
281 void add_node PP((node * np, node ** cur_node_p));
283 void init PP((void));
284 void initbuffer PP((linebuffer * lbp));
285 void find_entries PP((char *file, FILE * inf));
286 void free_tree PP((node * np));
288 PP((char *name, bool is_func, char *linestart, int linelen, int lno, long cno));
290 PP((char *name, int namelen, bool is_func, char *linestart, int linelen,
292 void process_file PP((char *file));
293 void put_entries PP((node * np));
294 void takeprec PP((void));
296 char *concat PP((char *s1, char *s2, char *s3));
297 char *skip_spaces PP((char *cp));
298 char *skip_non_spaces PP((char *cp));
299 char *savenstr PP((char *cp, int len));
300 char *savestr PP((char *cp));
301 char *etags_strchr PP((char *sp, int c));
302 char *etags_strrchr PP((char *sp, int c));
303 char *etags_getcwd PP((void));
304 char *relative_filename PP((char *file, char *dir));
305 char *absolute_filename PP((char *file, char *dir));
306 char *absolute_dirname PP((char *file, char *dir));
307 bool filename_is_absolute PP((char *fn));
308 void canonicalize_filename PP((char *fn));
309 void grow_linebuffer PP((linebuffer * lbp, int toksize));
310 long *xmalloc PP((unsigned int size));
311 long *xrealloc PP((char *ptr, unsigned int size));
313 char searchar = '/'; /* use /.../ searches */
315 char *tagfile; /* output file */
316 char *progname; /* name this program was invoked with */
317 char *cwd; /* current working directory */
318 char *tagfiledir; /* directory of tagfile */
319 FILE *tagf; /* ioptr for tags file */
321 char *curfile; /* current input file name */
322 language *curlang; /* current language */
324 int lineno; /* line number of current line */
325 long charno; /* current character number */
326 long linecharno; /* charno of start of current line */
327 char *dbp; /* pointer to start of current tag */
328 node *head; /* the head of the binary tree of tags */
330 linebuffer lb; /* the current line */
331 linebuffer token_name; /* used by C_entries as a temporary area */
334 linebuffer lb; /* used by C_entries instead of lb */
337 /* boolean "functions" (see init) */
338 bool _wht[CHARS], _nin[CHARS], _itk[CHARS], _btk[CHARS], _etk[CHARS];
341 *white = " \f\t\n\r",
343 *nonam = " \f\t\n\r(=,[;",
344 /* token ending chars */
345 *endtk = " \t\n\r\"'#()[]{}=-+%*/&|^~!<>;,.:?",
346 /* token starting chars */
347 *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~@",
348 /* valid in-token chars */
349 *midtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789";
351 bool append_to_tagfile; /* -a: append to tags */
352 /* The following four default to TRUE for etags, but to FALSE for ctags. */
353 bool typedefs; /* -t: create tags for C typedefs */
354 bool typedefs_and_cplusplus; /* -T: create tags for C typedefs, level */
355 /* 0 struct/enum/union decls, and C++ */
356 /* member functions. */
357 bool constantypedefs; /* -d: create tags for C #define, enum */
358 /* constants and variables. */
359 /* -D: opposite of -d. Default under ctags. */
360 bool globals; /* create tags for global variables */
361 bool members; /* create tags for C member variables */
362 bool update; /* -u: update tags */
363 bool vgrind_style; /* -v: create vgrind style index output */
364 bool no_warnings; /* -w: suppress warnings */
365 bool cxref_style; /* -x: create cxref style output */
366 bool cplusplus; /* .[hc] means C++, not C */
367 bool noindentypedefs; /* -I: ignore indentation in C */
369 bool oo_browser_format; /* -O: OO-Browser tags format */
373 struct option longopts[] = {
374 {"append", no_argument, NULL, 'a'},
375 {"backward-search", no_argument, NULL, 'B'},
376 {"c++", no_argument, NULL, 'C'},
377 {"cxref", no_argument, NULL, 'x'},
378 {"defines", no_argument, NULL, 'd'},
379 {"no-defines", no_argument, NULL, 'D'},
380 {"globals", no_argument, &globals, TRUE},
381 {"no-globals", no_argument, &globals, FALSE},
382 {"help", no_argument, NULL, 'h'},
383 {"help", no_argument, NULL, 'H'},
384 {"ignore-indentation", no_argument, NULL, 'I'},
385 {"include", required_argument, NULL, 'i'},
386 {"language", required_argument, NULL, 'l'},
387 {"members", no_argument, &members, TRUE},
388 {"no-members", no_argument, &members, FALSE},
389 {"no-warn", no_argument, NULL, 'w'},
390 {"output", required_argument, NULL, 'o'},
392 {"oo-browser", no_argument, NULL, 'O'},
395 {"regex", required_argument, NULL, 'r'},
396 {"no-regex", no_argument, NULL, 'R'},
397 #endif /* ETAGS_REGEXPS */
398 {"typedefs", no_argument, NULL, 't'},
399 {"typedefs-and-c++", no_argument, NULL, 'T'},
400 {"update", no_argument, NULL, 'u'},
401 {"version", no_argument, NULL, 'V'},
402 {"vgrind", no_argument, NULL, 'v'},
405 #endif /* LONG_OPTIONS */
408 /* Structure defining a regular expression. Elements are
409 the compiled pattern, and the name string. */
410 typedef struct pattern {
411 struct pattern *p_next;
414 struct re_pattern_buffer *pattern;
415 struct re_registers regs;
420 /* Array of all regexps. */
421 pattern *p_head = NULL;
422 #endif /* ETAGS_REGEXPS */
428 /* Non-NULL if language fixed. */
429 language *forced_lang = NULL;
432 char *Asm_suffixes[] = { "a", /* Unix assembler */
433 "asm", /* Microcontroller assembly */
434 "def", /* BSO/Tasking definition includes */
435 "inc", /* Microcontroller include files */
436 "ins", /* Microcontroller include files */
437 "s", "sa", /* Unix assembler */
438 "src", /* BSO/Tasking C compiler output */
442 /* Note that .c and .h can be considered C++, if the --c++ flag was
443 given. That is why default_C_entries is called here. */
444 char *default_C_suffixes[] = { "c", "h", NULL };
446 char *Cplusplus_suffixes[] =
447 { "C", "H", "c++", "cc", "cpp", "cxx", "h++", "hh", "hpp", "hxx",
448 "M", /* Objective C++ */
449 "pdb", /* Postscript with C syntax */
453 char *Cjava_suffixes[] = { "java", NULL };
455 char *Cobol_suffixes[] = { "COB", "cob", NULL };
457 char *Cstar_suffixes[] = { "cs", "hs", NULL };
459 char *Erlang_suffixes[] = { "erl", "hrl", NULL };
461 char *Fortran_suffixes[] = { "F", "f", "f90", "for", NULL };
463 char *Lisp_suffixes[] = { "cl", "clisp", "el", "l", "lisp", "lsp", "ml", NULL };
465 char *Pascal_suffixes[] = { "p", "pas", NULL };
467 char *Perl_suffixes[] = { "pl", "pm", NULL };
468 char *Perl_interpreters[] = { "perl", "@PERL@", NULL };
470 char *plain_C_suffixes[] = { "pc", /* Pro*C file */
471 "m", /* Objective C file */
472 "lm", /* Objective lex file */
476 char *Postscript_suffixes[] = { "ps", NULL };
478 char *Prolog_suffixes[] = { "prolog", NULL };
480 char *Python_suffixes[] = { "py", NULL };
482 /* Can't do the `SCM' or `scm' prefix with a version number. */
483 char *Scheme_suffixes[] =
484 { "SCM", "SM", "oak", "sch", "scheme", "scm", "sm", "ss", "t", NULL };
486 char *TeX_suffixes[] =
487 { "TeX", "bib", "clo", "cls", "ltx", "sty", "tex", NULL };
489 char *Yacc_suffixes[] = { "y", "ym", NULL }; /* .ym is Objective yacc file */
492 * Table of languages.
494 * It is ok for a given function to be listed under more than one
495 * name. I just didn't.
498 language lang_names[] = {
499 {"asm", Asm_labels, Asm_suffixes, NULL},
500 {"c", default_C_entries, default_C_suffixes, NULL},
501 {"c++", Cplusplus_entries, Cplusplus_suffixes, NULL},
502 {"c*", Cstar_entries, Cstar_suffixes, NULL},
503 {"cobol", Cobol_paragraphs, Cobol_suffixes, NULL},
504 {"erlang", Erlang_functions, Erlang_suffixes, NULL},
505 {"fortran", Fortran_functions, Fortran_suffixes, NULL},
506 {"java", Cjava_entries, Cjava_suffixes, NULL},
507 {"lisp", Lisp_functions, Lisp_suffixes, NULL},
508 {"pascal", Pascal_functions, Pascal_suffixes, NULL},
509 {"perl", Perl_functions, Perl_suffixes, Perl_interpreters},
510 {"postscript", Postscript_functions, Postscript_suffixes, NULL},
511 {"proc", plain_C_entries, plain_C_suffixes, NULL},
512 {"prolog", Prolog_functions, Prolog_suffixes, NULL},
513 {"python", Python_functions, Python_suffixes, NULL},
514 {"scheme", Scheme_functions, Scheme_suffixes, NULL},
515 {"tex", TeX_functions, TeX_suffixes, NULL},
516 {"yacc", Yacc_entries, Yacc_suffixes, NULL},
517 {"auto", NULL}, /* default guessing scheme */
518 {"none", just_read_file}, /* regexp matching only */
519 {NULL, NULL} /* end of list */
522 void print_language_names()
527 puts("\nThese are the currently supported languages, along with the\n\
528 default file name suffixes:");
529 for (lang = lang_names; lang->name != NULL; lang++) {
530 printf("\t%s\t", lang->name);
531 if (lang->suffixes != NULL)
532 for (ext = lang->suffixes; *ext != NULL; ext++)
533 printf(" .%s", *ext);
536 puts("Where `auto' means use default language for files based on file\n\
537 name suffix, and `none' means only do regexp processing on files.\n\
538 If no language is specified and no matching suffix is found,\n\
539 the first line of the file is read for a sharp-bang (#!) sequence\n\
540 followed by the name of an interpreter. If no such sequence is found,\n\
541 Fortran is tried first; if no tags are found, C is tried next.");
545 # define VERSION "20"
549 printf("%s (GNU Emacs %s)\n", (CTAGS) ? "ctags" : "etags", VERSION);
550 puts("Copyright (C) 1996 Free Software Foundation, Inc. and Ken Arnold");
551 puts("This program is distributed under the same terms as Emacs");
558 printf("Usage: %s [options] [[regex-option ...] file-name] ...\n\
560 These are the options accepted by %s.\n", progname, progname);
562 puts("You may use unambiguous abbreviations for the long option names.");
564 puts("Long option names do not work with this executable, as it is not\n\
565 linked with GNU getopt.");
566 #endif /* LONG_OPTIONS */
567 puts("A - as file name means read names from stdin (one per line).");
570 (" Absolute names are stored in the output file as they are.\n\
571 Relative ones are stored relative to the output file's directory.");
574 puts("-a, --append\n\
575 Append tag entries to existing tags file.");
578 puts("-B, --backward-search\n\
579 Write the search commands for the tag entries using '?', the\n\
580 backward-search command instead of '/', the forward-search command.");
583 Treat files whose name suffix defaults to C language as C++ files.");
586 puts("-d, --defines\n\
587 Create tag entries for C #define constants and enum constants, too.");
589 puts("-D, --no-defines\n\
590 Don't create tag entries for C #define constants and enum constants.\n\
591 This makes the tags file smaller.");
594 puts("-i FILE, --include=FILE\n\
595 Include a note in tag file indicating that, when searching for\n\
596 a tag, one should also consult the tags file FILE after\n\
597 checking the current file.");
598 puts("-l LANG, --language=LANG\n\
599 Force the following files to be considered as written in the\n\
600 named language up to the next --language=LANG option.");
605 Create tag entries for global variables in some languages.");
607 puts("--no-globals\n\
608 Do not create tag entries for global variables in some\n\
609 languages. This makes the tags file smaller.");
611 Create tag entries for member variables in C and derived languages.");
614 puts("-r /REGEXP/, --regex=/REGEXP/ or --regex=@regexfile\n\
615 Make a tag for each line matching pattern REGEXP in the\n\
616 following files. regexfile is a file containing one REGEXP\n\
617 per line. REGEXP is anchored (as if preceded by ^).\n\
618 The form /REGEXP/NAME/ creates a named tag. For example Tcl\n\
619 named tags can be created with:\n\
620 --regex=/proc[ \\t]+\\([^ \\t]+\\)/\\1/.");
621 puts("-R, --no-regex\n\
622 Don't create tags from regexps for the following files.");
623 #endif /* ETAGS_REGEXPS */
624 puts("-o FILE, --output=FILE\n\
625 Write the tags to FILE.");
627 puts("-O, --oo-browser\n\
628 Generate a specialized tags format used only by the Altrasoft OO-Browser.");
630 puts("-I, --ignore-indentation\n\
631 Don't rely on indentation quite as much as normal. Currently,\n\
632 this means not to assume that a closing brace in the first\n\
633 column is the final brace of a function or structure\n\
634 definition in C and C++.");
637 puts("-t, --typedefs\n\
638 Generate tag entries for C typedefs.");
639 puts("-T, --typedefs-and-c++\n\
640 Generate tag entries for C typedefs, C struct/enum/union tags,\n\
641 and C++ member functions.");
642 puts("-u, --update\n\
643 Update the tag entries for the given files, leaving tag\n\
644 entries for other files in place. Currently, this is\n\
645 implemented by deleting the existing entries for the given\n\
646 files and then rewriting the new entries at the end of the\n\
647 tags file. It is often faster to simply rebuild the entire\n\
648 tag file than to use this.");
649 puts("-v, --vgrind\n\
650 Generates an index of items intended for human consumption,\n\
651 similar to the output of vgrind. The index is sorted, and\n\
652 gives the page number of each item.");
653 puts("-w, --no-warn\n\
654 Suppress warning messages about entries defined in multiple\n\
657 Like --vgrind, but in the style of cxref, rather than vgrind.\n\
658 The output uses line numbers instead of page numbers, but\n\
659 beyond that the differences are cosmetic; try both to see\n\
663 puts("-V, --version\n\
664 Print the version of the program.\n\
666 Print this help message.");
668 print_language_names();
671 puts("Report bugs to bug-gnu-emacs@prep.ai.mit.edu");
682 /* This structure helps us allow mixing of --lang and file names. */
684 enum argument_type arg_type;
689 #ifdef VMS /* VMS specific functions */
693 /* This is a BUG! ANY arbitrary limit is a BUG!
694 Won't someone please fix this? */
695 #define MAX_FILE_SPEC_LEN 255
698 char body[MAX_FILE_SPEC_LEN + 1];
702 v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names
703 returning in each successive call the next file name matching the input
704 spec. The function expects that each in_spec passed
705 to it will be processed to completion; in particular, up to and
706 including the call following that in which the last matching name
707 is returned, the function ignores the value of in_spec, and will
708 only start processing a new spec with the following call.
709 If an error occurs, on return out_spec contains the value
710 of in_spec when the error occurred.
712 With each successive file name returned in out_spec, the
713 function's return value is one. When there are no more matching
714 names the function returns zero. If on the first call no file
715 matches in_spec, or there is any other error, -1 is returned.
720 #define OUTSIZE MAX_FILE_SPEC_LEN
721 short fn_exp(out, in)
725 static long context = 0;
726 static struct dsc$descriptor_s o;
727 static struct dsc$descriptor_s i;
728 static bool pass1 = TRUE;
734 o.dsc$a_pointer = (char *)out;
735 o.dsc$w_length = (short)OUTSIZE;
736 i.dsc$a_pointer = in;
737 i.dsc$w_length = (short)strlen(in);
738 i.dsc$b_dtype = DSC$K_DTYPE_T;
739 i.dsc$b_class = DSC$K_CLASS_S;
740 o.dsc$b_dtype = DSC$K_DTYPE_VT;
741 o.dsc$b_class = DSC$K_CLASS_VS;
743 if ((status = lib$find_file(&i, &o, &context, 0, 0)) == RMS$_NORMAL) {
744 out->body[out->curlen] = EOS;
746 } else if (status == RMS$_NMF)
749 strcpy(out->body, in);
752 lib$find_file_end(&context);
758 v1.01 nmm 19-Aug-85 gfnames - return in successive calls the
759 name of each file specified by the provided arg expanding wildcards.
761 char *gfnames(arg, p_error)
765 static vspec filename = { MAX_FILE_SPEC_LEN, "\0" };
767 switch (fn_exp(&filename, arg)) {
770 return filename.body;
776 return filename.body;
780 #ifndef OLD /* Newer versions of VMS do provide `system'. */
784 error("%s", "system() function not implemented under VMS");
788 #define VERSION_DELIM ';'
789 char *massage_name(s)
795 if (*s == VERSION_DELIM) {
804 int main(int argc, char *argv[])
807 unsigned int nincluded_files;
808 char **included_files;
811 int current_arg, file_count;
812 linebuffer filename_lb;
819 included_files = xnew(argc, char *);
823 /* Allocate enough no matter what happens. Overkill, but each one
825 argbuffer = xnew(argc, argument);
828 /* Set syntax for regular expression routines. */
829 re_set_syntax(RE_SYNTAX_EMACS | RE_INTERVALS);
830 #endif /* ETAGS_REGEXPS */
833 * If etags, always find typedefs and structure tags. Why not?
834 * Also default is to find macro constants, enum constants and
838 typedefs = typedefs_and_cplusplus = constantypedefs = TRUE;
849 optstring = "-aCdDf:Il:o:r:RStTi:BuvxwVhH";
851 optstring = "-aCdDf:Il:o:r:RStTi:BOuvxwVhH";
855 optstring = "-aCdDf:Il:o:StTi:BuvxwVhH";
857 optstring = "-aCdDf:Il:o:StTi:BOuvxwVhH";
859 #endif /* ETAGS_REGEXPS */
862 optstring = optstring + 1;
863 #endif /* LONG_OPTIONS */
865 opt = getopt_long(argc, argv, optstring, longopts, 0);
871 /* If getopt returns 0, then it has already processed a
872 long-named option. We should do nothing. */
876 /* This means that a file name has been seen. Record it. */
877 argbuffer[current_arg].arg_type = at_filename;
878 argbuffer[current_arg].what = optarg;
883 /* Common options. */
885 append_to_tagfile = TRUE;
891 constantypedefs = TRUE;
894 constantypedefs = FALSE;
896 case 'f': /* for compatibility with old makefiles */
899 /* convert char to string, to call error with */
902 error("-%s option may only be given once.",
904 suggest_asking_for_help();
910 oo_browser_format = TRUE;
914 case 'S': /* for backward compatibility */
915 noindentypedefs = TRUE;
919 language *lang = get_language_from_name(optarg);
921 argbuffer[current_arg].lang = lang;
922 argbuffer[current_arg].arg_type =
930 argbuffer[current_arg].arg_type = at_regexp;
931 argbuffer[current_arg].what = optarg;
935 argbuffer[current_arg].arg_type = at_regexp;
936 argbuffer[current_arg].what = NULL;
939 #endif /* ETAGS_REGEXPS */
951 typedefs = typedefs_and_cplusplus = TRUE;
956 included_files[nincluded_files++] = optarg;
968 /*FALLTHRU*/ case 'x':
976 suggest_asking_for_help();
980 for (; optind < argc; ++optind) {
981 argbuffer[current_arg].arg_type = at_filename;
982 argbuffer[current_arg].what = argv[optind];
987 if (nincluded_files == 0 && file_count == 0) {
988 error("no input files specified.", 0);
989 suggest_asking_for_help();
993 tagfile = CTAGS ? "tags" : "TAGS";
994 cwd = etags_getcwd(); /* the current working directory */
995 if (cwd[strlen(cwd) - 1] != '/') {
997 cwd = concat(oldcwd, "/", "");
1000 if (streq(tagfile, "-"))
1003 tagfiledir = absolute_dirname(tagfile, cwd);
1005 init(); /* set up boolean "functions" */
1008 initbuffer(&token_name);
1009 initbuffer(&lbs[0].lb);
1010 initbuffer(&lbs[1].lb);
1011 initbuffer(&filename_lb);
1014 if (streq(tagfile, "-")) {
1017 tagf = fopen(tagfile, append_to_tagfile ? "a" : "w");
1023 * Loop through files finding functions.
1025 for (i = 0; i < current_arg; ++i) {
1026 switch (argbuffer[i].arg_type) {
1028 forced_lang = argbuffer[i].lang;
1030 #ifdef ETAGS_REGEXPS
1032 analyse_regex(argbuffer[i].what);
1038 gfnames(argbuffer[i].what, &got_err)) != NULL) {
1040 error("can't find file %s\n",
1044 this_file = massage_name(this_file);
1047 this_file = argbuffer[i].what;
1050 oo_browser_clear_all_globals();
1052 /* Input file named "-" means read file names from stdin
1053 (one per line) and use them. */
1054 if (streq(this_file, "-"))
1055 while (readline_internal(&filename_lb, stdin) >
1059 oo_browser_clear_some_globals();
1061 process_file(filename_lb.buffer);
1066 process_file(this_file);
1076 #ifdef ETAGS_REGEXPS
1078 #endif /* ETAGS_REGEXPS */
1081 while (nincluded_files-- > 0)
1082 fprintf(tagf, "\f\n%s,include\n", *included_files++);
1088 /* If CTAGS, we are here. process_file did not write the tags yet,
1089 because we want them ordered. Let's do it now. */
1098 for (i = 0; i < current_arg; ++i) {
1099 if (argbuffer[i].arg_type != at_filename)
1101 sz = snprintf(cmd, sizeof(cmd),
1102 "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS",
1103 tagfile, argbuffer[i].what, tagfile);
1104 if(sz >= 0 && (size_t)sz < sizeof(cmd))
1105 fatal("failed to build shell command line", (char *)NULL);
1106 if (system(cmd) != GOOD)
1107 fatal("failed to execute shell command", (char *)NULL);
1109 append_to_tagfile = TRUE;
1112 tagf = fopen(tagfile, append_to_tagfile ? "a" : "w");
1120 int sz = snprintf(cmd, sizeof(cmd), "sort %s -o %s", tagfile, tagfile);
1121 if(sz >= 0 && (size_t)sz < sizeof(cmd))
1122 fatal("failed to build sort command line", (char *)NULL);
1129 * Return a language given the name.
1131 language *get_language_from_name(name)
1137 error("empty language name", (char *)NULL);
1139 for (lang = lang_names; lang->name != NULL; lang++)
1140 if (streq(name, lang->name))
1142 error("unknown language \"%s\"", name);
1149 * Return a language given the interpreter name.
1151 language *get_language_from_interpreter(interpreter)
1157 if (interpreter == NULL)
1159 for (lang = lang_names; lang->name != NULL; lang++)
1160 if (lang->interpreters != NULL)
1161 for (iname = lang->interpreters; *iname != NULL;
1163 if (streq(*iname, interpreter))
1170 * Return a language given the file suffix.
1172 language *get_language_from_suffix(suffix)
1180 for (lang = lang_names; lang->name != NULL; lang++)
1181 if (lang->suffixes != NULL)
1182 for (ext = lang->suffixes; *ext != NULL; ext++)
1183 if (streq(*ext, suffix))
1190 * This routine is called on each file argument.
1192 void process_file(file)
1195 struct stat stat_buf;
1198 canonicalize_filename(file);
1199 if (streq(file, tagfile) && !streq(tagfile, "-")) {
1200 error("skipping inclusion of %s in self.", file);
1203 inf = fopen(file, "r");
1204 if (stat(file, &stat_buf) == 0 && !S_ISREG(stat_buf.st_mode)) {
1205 error("skipping %s: it is not a regular file.", file);
1215 find_entries(file, inf);
1220 if (filename_is_absolute(file)) {
1221 /* file is an absolute file name. Canonicalise it. */
1222 filename = absolute_filename(file, cwd);
1224 /* file is a file name relative to cwd. Make it relative
1225 to the directory of the tags file. */
1226 filename = relative_filename(file, tagfiledir);
1229 if (oo_browser_format)
1230 fprintf(tagf, "\f\n%s\n", filename);
1233 fprintf(tagf, "\f\n%s,%d\n", filename,
1234 total_size_of_entries(head));
1243 * This routine sets up the boolean pseudo-functions which work
1244 * by setting boolean flags dependent upon the corresponding character.
1245 * Every char which is NOT in that string is not a white char. Therefore,
1246 * all of the array "_wht" is set to FALSE, and then the elements
1247 * subscripted by the chars in "white" are set to TRUE. Thus "_wht"
1248 * of a char is TRUE if it is the string "white", else FALSE.
1255 for (i = 0; i < CHARS; i++)
1256 iswhite(i) = notinname(i) = begtoken(i) = intoken(i) =
1257 endtoken(i) = FALSE;
1258 for (sp = white; *sp != '\0'; sp++)
1259 iswhite(*sp) = TRUE;
1260 for (sp = nonam; *sp != '\0'; sp++)
1261 notinname(*sp) = TRUE;
1262 for (sp = begtk; *sp != '\0'; sp++)
1263 begtoken(*sp) = TRUE;
1264 for (sp = midtk; *sp != '\0'; sp++)
1265 intoken(*sp) = TRUE;
1266 for (sp = endtk; *sp != '\0'; sp++)
1267 endtoken(*sp) = TRUE;
1268 iswhite('\0') = iswhite('\n');
1269 notinname('\0') = notinname('\n');
1270 begtoken('\0') = begtoken('\n');
1271 intoken('\0') = intoken('\n');
1272 endtoken('\0') = endtoken('\n');
1276 * This routine opens the specified file and calls the function
1277 * which finds the function and type definitions.
1279 node *last_node = NULL;
1281 void find_entries(file, inf)
1287 node *old_last_node;
1289 curfile = savestr(file);
1291 /* If user specified a language, use it. */
1293 if (lang != NULL && lang->function != NULL) {
1295 lang->function(inf);
1301 cp = etags_strrchr(file, '.');
1304 lang = get_language_from_suffix(cp);
1305 if (lang != NULL && lang->function != NULL) {
1307 lang->function(inf);
1314 /* Look for sharp-bang as the first two characters. */
1315 if (readline_internal(&lb, inf) > 0
1316 && lb.len >= 2 && lb.buffer[0] == '#' && lb.buffer[1] == '!') {
1319 /* Set lp to point at the first char after the last slash in the
1320 line or, if no slashes, at the first nonblank. Then set cp to
1321 the first successive blank and terminate the string. */
1322 lp = etags_strrchr(lb.buffer + 2, '/');
1326 lp = skip_spaces(lb.buffer + 2);
1327 cp = skip_non_spaces(lp);
1330 if (strlen(lp) > 0) {
1331 lang = get_language_from_interpreter(lp);
1332 if (lang != NULL && lang->function != NULL) {
1334 lang->function(inf);
1344 old_last_node = last_node;
1345 curlang = get_language_from_name("fortran");
1346 Fortran_functions(inf);
1348 /* No Fortran entries found. Try C. */
1349 if (old_last_node == last_node) {
1351 curlang = get_language_from_name(cplusplus ? "c++" : "c");
1352 default_C_entries(inf);
1360 void pfnote(name, is_func, linestart, linelen, lno, cno)
1361 char *name; /* tag name, or NULL if unnamed */
1362 bool is_func; /* tag is a function */
1363 char *linestart; /* start of the line where tag is */
1364 int linelen; /* length of the line where tag is */
1365 int lno; /* line number */
1366 long cno; /* character number */
1370 if (CTAGS && name == NULL)
1375 /* If ctags mode, change name "main" to M<thisfilename>. */
1376 if (CTAGS && !cxref_style && streq(name, "main")) {
1377 register char *fp = etags_strrchr(curfile, '/');
1378 np->name = concat("M", fp == 0 ? curfile : fp + 1, "");
1379 fp = etags_strrchr(np->name, '.');
1380 if (fp && fp[1] != '\0' && fp[2] == '\0')
1384 np->been_warned = FALSE;
1386 np->is_func = is_func;
1388 /* Our char numbers are 0-base, because of C language tradition?
1389 ctags compatibility? old versions compatibility? I don't know.
1390 Anyway, since emacs's are 1-base we expect etags.el to take care
1391 of the difference. If we wanted to have 1-based numbers, we would
1392 uncomment the +1 below. */
1393 np->cno = cno /* + 1 */ ;
1394 np->left = np->right = NULL;
1395 if (CTAGS && !cxref_style) {
1396 if (strlen(linestart) < 50)
1397 np->pat = concat(linestart, "$", "");
1399 np->pat = savenstr(linestart, 50);
1401 np->pat = savenstr(linestart, linelen);
1404 if (oo_browser_format)
1405 np->construct = oo_browser_construct;
1406 oo_browser_construct = C_NULL;
1407 oo_browser_check_and_clear_structtype();
1410 add_node(np, &head);
1413 /* Date: Wed, 22 Jan 1997 02:56:31 -0500 [last amended 18 Sep 1997]
1414 * From: Sam Kendall <kendall@mv.mv.com>
1415 * Subject: Proposal for firming up the TAGS format specification
1416 * To: F.Potorti@cnuce.cnr.it
1418 * pfnote should emit the optimized form [unnamed tag] only if:
1419 * 1. name does not contain any of the characters " \t\r\n(),;";
1420 * 2. linestart contains name as either a rightmost, or rightmost but
1421 * one character, substring;
1422 * 3. the character, if any, immediately before name in linestart must
1423 * be one of the characters " \t(),;";
1424 * 4. the character, if any, immediately after name in linestart must
1425 * also be one of the characters " \t(),;".
1427 * The real implementation uses the notinname() macro, which recognises
1428 * characters slightly different form " \t\r\n(),;". See the variable
1431 #define traditional_tag_style TRUE
1432 void new_pfnote(name, namelen, is_func, linestart, linelen, lno, cno)
1433 char *name; /* tag name, or NULL if unnamed */
1434 int namelen; /* tag length */
1435 bool is_func; /* tag is a function */
1436 char *linestart; /* start of the line where tag is */
1437 int linelen; /* length of the line where tag is */
1438 int lno; /* line number */
1439 long cno; /* character number */
1446 for (cp = name; !notinname(*cp); cp++)
1448 if (*cp == '\0') { /* rule #1 */
1449 cp = linestart + linelen - namelen;
1450 if (notinname(linestart[linelen - 1]))
1451 cp -= 1; /* rule #4 */
1453 if (!oo_browser_format && cp >= linestart /* rule #2 */
1455 if (cp >= linestart /* rule #2 */
1457 && (cp == linestart || notinname(cp[-1])) /* rule #3 */
1458 &&strneq(name, cp, namelen)) /* rule #2 */
1459 named = FALSE; /* use unnamed tag */
1464 name = savenstr(name, namelen);
1467 pfnote(name, is_func, linestart, linelen, lno, cno);
1472 * recurse on left children, iterate on right children.
1478 register node *node_right = np->right;
1479 free_tree(np->left);
1480 if (np->name != NULL)
1490 * Adds a node to the tree of nodes. In etags mode, we don't keep
1491 * it sorted; we just keep a linear list. In ctags mode, maintain
1492 * an ordered tree, with no attempt at balancing.
1494 * add_node is the only function allowed to add nodes, so it can
1497 void add_node(np, cur_node_p)
1498 node *np, **cur_node_p;
1501 register node *cur_node = *cur_node_p;
1503 if (cur_node == NULL) {
1511 if (last_node == NULL)
1512 fatal("internal error in add_node", (char *)NULL);
1513 last_node->right = np;
1517 dif = strcmp(np->name, cur_node->name);
1520 * If this tag name matches an existing one, then
1521 * do not add the node, but maybe print a warning.
1524 if (streq(np->file, cur_node->file)) {
1527 "Duplicate entry in file %s, line %d: %s\n",
1528 np->file, lineno, np->name);
1530 "Second entry ignored\n");
1532 } else if (!cur_node->been_warned && !no_warnings) {
1535 "Duplicate entry in files %s and %s: %s (Warning only)\n",
1536 np->file, cur_node->file, np->name);
1537 cur_node->been_warned = TRUE;
1542 /* Actually add the node */
1543 add_node(np, dif < 0 ? &cur_node->left : &cur_node->right);
1548 /* Default class name for the current OO-Browser tag. */
1549 static char *oo_browser_class;
1550 /* Prefix character to use in OO-Browser listings for the current tag. */
1551 static char oo_browser_prefix;
1554 void put_entries(node * np)
1561 /* Output subentries that precede this one */
1562 put_entries(np->left);
1564 /* Output this entry */
1568 if (oo_browser_format) {
1569 /* Omit C++ `class' and `method' entries as well as Objective-C
1570 entries from this OO-Browser tags file since the browser handles
1571 them independently of this file. Omit `extern' variable declarations
1572 as they are unused by the OO-Browser. */
1573 if (np->construct != C_CLASS
1574 && np->construct != C_METHOD
1575 && np->construct != C_EXTERN
1576 && np->construct != C_OBJC) {
1578 oo_browser_default_classes[np->construct];
1579 switch (np->construct) {
1587 oo_browser_prefix = '=';
1591 oo_browser_prefix = '-';
1596 if (np->name != NULL)
1597 fprintf(tagf, "%s@%c %s@%s\n",
1599 oo_browser_prefix, np->name,
1602 fprintf(tagf, "%s@%c ???@%s\n",
1604 oo_browser_prefix, np->pat);
1608 if (np->name != NULL)
1609 fprintf(tagf, "%s\177%s\001%d,%ld\n",
1610 np->pat, np->name, np->lno, np->cno);
1612 fprintf(tagf, "%s\177%d,%ld\n",
1613 np->pat, np->lno, np->cno);
1618 if (np->name == NULL)
1619 error("internal error: NULL name in ctags mode.",
1624 fprintf(stdout, "%s %s %d\n",
1626 (np->lno + 63) / 64);
1628 fprintf(stdout, "%-16s %3d %-16s %s\n",
1629 np->name, np->lno, np->file, np->pat);
1631 fprintf(tagf, "%s\t%s\t", np->name, np->file);
1633 if (np->is_func) { /* a function */
1634 putc(searchar, tagf);
1637 for (sp = np->pat; *sp; sp++) {
1638 if (*sp == '\\' || *sp == searchar)
1642 putc(searchar, tagf);
1643 } else { /* a typedef; text pattern inadequate */
1644 fprintf(tagf, "%d", np->lno);
1650 /* Output subentries that follow this one */
1651 put_entries(np->right);
1654 /* Length of a number's decimal representation. */
1655 int number_len PP((long num));
1660 while ((num /= 10) > 0)
1666 * Return total number of characters that put_entries will output for
1667 * the nodes in the subtree of the specified node. Works only if
1668 * we are not ctags, but called only in that case. This count
1669 * is irrelevant with the new tags.el, but is still supplied for
1670 * backward compatibility.
1672 int total_size_of_entries(np)
1680 for (total = 0; np != NULL; np = np->right) {
1681 /* Count left subentries. */
1682 total += total_size_of_entries(np->left);
1684 /* Count this entry */
1685 total += strlen(np->pat) + 1;
1687 number_len((long)np->lno) + 1 + number_len(np->cno) + 1;
1688 if (np->name != NULL)
1689 total += 1 + strlen(np->name); /* \001name */
1696 * The C symbol tables.
1700 st_C_objprot, st_C_objimpl, st_C_objend,
1704 st_C_struct, st_C_enum, st_C_define, st_C_typedef, st_C_typespec,
1707 , st_C_union, st_C_class, st_C_extern, st_C_inline
1711 /* Feed stuff between (but not including) %[ and %] lines to:
1712 gperf -c -k 1,3 -o -p -r -t
1714 struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
1716 @interface, 0, st_C_objprot
1717 @protocol, 0, st_C_objprot
1718 @implementation,0, st_C_objimpl
1719 @end, 0, st_C_objend
1720 import, C_JAVA, st_C_ignore
1721 package, C_JAVA, st_C_ignore
1722 friend, C_PLPL, st_C_ignore
1723 extends, C_JAVA, st_C_javastruct
1724 implements, C_JAVA, st_C_javastruct
1725 interface, C_JAVA, st_C_struct
1726 class, C_PLPL, st_C_class
1727 namespace, C_PLPL, st_C_struct
1728 domain, C_STAR, st_C_struct
1729 union, 0, st_C_union
1730 struct, 0, st_C_struct
1732 typedef, 0, st_C_typedef
1733 define, 0, st_C_define
1734 inline, 0, st_C_inline
1735 bool, C_PLPL, st_C_typespec
1736 long, 0, st_C_typespec
1737 short, 0, st_C_typespec
1738 int, 0, st_C_typespec
1739 char, 0, st_C_typespec
1740 float, 0, st_C_typespec
1741 double, 0, st_C_typespec
1742 signed, 0, st_C_typespec
1743 unsigned, 0, st_C_typespec
1744 auto, 0, st_C_typespec
1745 void, 0, st_C_typespec
1746 extern, 0, st_C_extern
1747 static, 0, st_C_typespec
1748 const, 0, st_C_const
1749 volatile, 0, st_C_typespec
1750 explicit, C_PLPL, st_C_typespec
1751 mutable, C_PLPL, st_C_typespec
1752 typename, C_PLPL, st_C_typespec
1753 # DEFUN used in emacs, the next three used in glibc (SYSCALL only for mach).
1754 DEFUN, 0, st_C_gnumacro
1755 SYSCALL, 0, st_C_gnumacro
1756 ENTRY, 0, st_C_gnumacro
1757 PSEUDO, 0, st_C_gnumacro
1758 # These are defined inside C functions, so currently they are not met.
1759 # EXFUN used in glibc, DEFVAR_* in emacs.
1760 #EXFUN, 0, st_C_gnumacro
1761 #DEFVAR_, 0, st_C_gnumacro
1763 and replace lines between %< and %> with its output. */
1765 /* C code produced by gperf version 2.5 (GNU C++ version) */
1766 /* Command-line: gperf -c -k 1,3 -o -p -r -t */
1767 struct C_stab_entry {
1773 #define TOTAL_KEYWORDS 41
1774 #define MIN_WORD_LENGTH 3
1775 #define MAX_WORD_LENGTH 15
1776 #define MIN_HASH_VALUE 13
1777 #define MAX_HASH_VALUE 129
1778 /* maximum key range = 117, duplicates = 0 */
1780 static unsigned int hash(char *str, unsigned int len)
1782 static unsigned char asso_values[] = {
1783 130, 130, 130, 130, 130, 130, 130, 130, 130, 130,
1784 130, 130, 130, 130, 130, 130, 130, 130, 130, 130,
1785 130, 130, 130, 130, 130, 130, 130, 130, 130, 130,
1786 130, 130, 130, 130, 130, 130, 130, 130, 130, 130,
1787 130, 130, 130, 130, 130, 130, 130, 130, 130, 130,
1788 130, 130, 130, 130, 130, 130, 130, 130, 130, 130,
1789 130, 130, 130, 130, 13, 130, 130, 130, 33, 32,
1790 47, 130, 130, 130, 130, 130, 130, 130, 130, 130,
1791 5, 130, 130, 20, 32, 130, 130, 130, 130, 130,
1792 130, 130, 130, 130, 130, 130, 130, 47, 55, 8,
1793 15, 33, 61, 38, 130, 60, 130, 130, 2, 9,
1794 10, 62, 59, 130, 28, 27, 50, 19, 3, 130,
1795 130, 130, 130, 130, 130, 130, 130, 130,
1798 asso_values[(unsigned char)str[2]] +
1799 asso_values[(unsigned char)str[0]];
1802 static struct C_stab_entry *in_word_set(char *str, unsigned int len)
1804 static struct C_stab_entry wordlist[] = {
1805 {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
1806 {"",}, {"",}, {"",}, {"",},
1807 {"volatile", 0, st_C_typespec},
1809 {"long", 0, st_C_typespec},
1810 {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
1811 {"const", 0, st_C_const},
1812 {"",}, {"",}, {"",},
1813 {"@end", 0, st_C_objend},
1814 {"namespace", C_PLPL, st_C_struct},
1816 {"domain", C_STAR, st_C_struct},
1818 {"@interface", 0, st_C_objprot},
1819 {"",}, {"",}, {"",},
1820 {"@implementation", 0, st_C_objimpl},
1822 {"double", 0, st_C_typespec},
1824 {"PSEUDO", 0, st_C_gnumacro},
1825 {"",}, {"",}, {"",},
1826 {"SYSCALL", 0, st_C_gnumacro},
1828 {"@protocol", 0, st_C_objprot},
1829 {"",}, {"",}, {"",},
1830 {"unsigned", 0, st_C_typespec},
1832 {"enum", 0, st_C_enum},
1834 {"char", 0, st_C_typespec},
1835 {"class", C_PLPL, st_C_class},
1836 {"struct", 0, st_C_struct},
1837 {"",}, {"",}, {"",}, {"",},
1838 {"mutable", C_PLPL, st_C_typespec},
1839 {"void", 0, st_C_typespec},
1840 {"inline", 0, st_C_inline},
1841 {"ENTRY", 0, st_C_gnumacro},
1843 {"signed", 0, st_C_typespec},
1845 {"package", C_JAVA, st_C_ignore},
1846 {"",}, {"",}, {"",}, {"",}, {"",},
1847 {"static", 0, st_C_typespec},
1849 {"define", 0, st_C_define},
1851 {"union", 0, st_C_union},
1852 {"DEFUN", 0, st_C_gnumacro},
1853 {"",}, {"",}, {"",},
1854 {"extern", 0, st_C_extern},
1855 {"extends", C_JAVA, st_C_javastruct},
1856 {"",}, {"",}, {"",},
1857 {"short", 0, st_C_typespec},
1858 {"",}, {"",}, {"",}, {"",}, {"",},
1859 {"explicit", C_PLPL, st_C_typespec},
1860 {"auto", 0, st_C_typespec},
1861 {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
1863 {"int", 0, st_C_typespec},
1865 {"typedef", 0, st_C_typedef},
1866 {"typename", C_PLPL, st_C_typespec},
1868 {"interface", C_JAVA, st_C_struct},
1870 {"bool", C_PLPL, st_C_typespec},
1871 {"",}, {"",}, {"",},
1872 {"import", C_JAVA, st_C_ignore},
1874 {"friend", C_PLPL, st_C_ignore},
1875 {"float", 0, st_C_typespec},
1876 {"implements", C_JAVA, st_C_javastruct},
1879 if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH) {
1880 register int key = hash(str, len);
1882 if (key <= MAX_HASH_VALUE && key >= 0) {
1883 register char *s = wordlist[key].name;
1885 if (*s == *str && !strncmp(str + 1, s + 1, len - 1))
1886 return &wordlist[key];
1894 enum sym_type C_symtype PP((char *str, int len, int c_ext));
1895 enum sym_type C_symtype(str, len, c_ext)
1900 register struct C_stab_entry *se = in_word_set(str, len);
1902 if (se == NULL || (se->c_ext && !(c_ext & se->c_ext)))
1908 * C functions and variables are recognized using a simple
1909 * finite automaton. fvdef is its state variable.
1912 fvnone, /* nothing seen */
1913 fvnameseen, /* function or variable name seen */
1914 fstartlist, /* func: just after open parenthesis */
1915 finlist, /* func: in parameter list */
1916 flistseen, /* func: after parameter list */
1917 fignore, /* func: before open brace */
1918 vignore /* var-like: ignore until ';' */
1922 * typedefs are recognized using a simple finite automaton.
1923 * typdef is its state variable.
1926 tnone, /* nothing seen */
1927 ttypedseen, /* typedef keyword seen */
1928 tinbody, /* inside typedef body */
1929 tend, /* just before typedef tag */
1930 tignore /* junk after typedef tag */
1934 * struct-like structures (enum, struct and union) are recognized
1935 * using another simple finite automaton. `structdef' is its state
1939 snone, /* nothing seen yet */
1940 skeyseen, /* struct-like keyword seen */
1941 stagseen, /* struct-like tag seen */
1942 scolonseen, /* colon seen after struct-like tag */
1943 sinbody /* in struct body: recognize member func defs */
1947 * When structdef is stagseen, scolonseen, or sinbody, structtag is the
1948 * struct tag, and structtype is the type of the preceding struct-like
1951 char *structtag = "<uninited>";
1952 enum sym_type structtype;
1955 void oo_browser_check_and_clear_structtype(void)
1957 /* Allow for multiple enum_label tags. */
1958 if (structtype != st_C_enum)
1959 structtype = st_none;
1964 * When objdef is different from onone, objtag is the name of the class.
1966 char *objtag = "<uninited>";
1969 * Yet another little state machine to deal with preprocessor lines.
1972 dnone, /* nothing seen */
1973 dsharpseen, /* '#' seen as first char on line */
1974 ddefineseen, /* '#' and 'define' seen */
1975 dignorerest /* ignore rest of line */
1979 * State machine for Objective C protocols and implementations.
1980 * Tom R.Hageman <tom@basil.icce.rug.nl>
1983 onone, /* nothing seen */
1984 oprotocol, /* @interface or @protocol seen */
1985 oimplementation, /* @implementations seen */
1986 otagseen, /* class name seen */
1987 oparenseen, /* parenthesis before category seen */
1988 ocatseen, /* category name seen */
1989 oinbody, /* in @implementation body */
1990 omethodsign, /* in @implementation body, after +/- */
1991 omethodtag, /* after method name */
1992 omethodcolon, /* after method colon */
1993 omethodparm, /* after method parameter */
1994 oignore /* wait for @end */
1998 * Use this structure to keep info about the token read, and how it
1999 * should be tagged. Used by the make_C_tag function to build a tag.
2011 token tok; /* latest token read */
2014 * Set this to TRUE, and the next token considered is called a function.
2015 * Used only for GNU emacs's function-defining macros.
2017 bool next_token_is_func;
2020 * TRUE in the rules part of a yacc file, FALSE outside (parse as C).
2025 * methodlen is the length of the method name stored in token_name.
2030 void oo_browser_clear_all_globals(void)
2032 /* Initialize globals so there is no carry over between files. */
2033 oo_browser_construct = C_NULL;
2039 structtype = st_none;
2040 next_token_is_func = yacc_rules = FALSE;
2043 void oo_browser_clear_some_globals(void)
2045 oo_browser_construct = C_NULL;
2046 structtype = st_none;
2052 * checks to see if the current token is at the start of a
2053 * function or variable, or corresponds to a typedef, or
2054 * is a struct/union/enum tag, or #define, or an enum constant.
2056 * *IS_FUNC gets TRUE iff the token is a function or #define macro
2057 * with args. C_EXT is which language we are looking at.
2059 * In the future we will need some way to adjust where the end of
2060 * the token is; for instance, implementing the C++ keyword
2061 * `operator' properly will adjust the end of the token to be after
2062 * whatever follows `operator'.
2070 * next_token_is_func IN OUT
2072 bool consider_token PP((char *str, int len, int c, int c_ext,
2073 int cblev, int parlev, bool * is_func_or_var));
2074 bool consider_token(str, len, c, c_ext, cblev, parlev, is_func_or_var)
2075 register char *str; /* IN: token pointer */
2076 register int len; /* IN: token length */
2077 register int c; /* IN: first char after the token */
2078 int c_ext; /* IN: C extensions mask */
2079 int cblev; /* IN: curly brace level */
2080 int parlev; /* IN: parenthesis level */
2081 bool *is_func_or_var; /* OUT: function or variable found */
2083 enum sym_type toktype = C_symtype(str, len, c_ext);
2086 switch ((unsigned int)toktype) {
2088 set_construct(C_STRUCTURE);
2091 set_construct(C_UNION);
2094 set_construct(C_CLASS);
2097 set_construct(C_ENUMERATION);
2100 set_construct(C_TYPE);
2103 set_construct(C_EXTERN);
2106 set_construct(C_FUNCTION);
2116 * Advance the definedef state machine.
2118 switch (definedef) {
2120 /* We're not on a preprocessor line. */
2123 if (toktype == st_C_define) {
2124 definedef = ddefineseen;
2126 definedef = dignorerest;
2131 * Make a tag for any macro, unless it is a constant
2132 * and constantypedefs is FALSE.
2134 definedef = dignorerest;
2136 *is_func_or_var = (c == '(');
2139 char *p = str + len * sizeof(char);
2142 /* This must be a macro since there is no
2143 whitespace between the opening parenthesis
2144 and the definition name. */
2145 *is_func_or_var = TRUE;
2147 *is_func_or_var = FALSE;
2149 /* Handle possible whitespace between macro tag and opening
2150 parenthesis and ensure this is an actual macro.
2151 -- Bob Weiner, Altrasoft, 11/19/1997 */
2152 while (*p && isspace(*p))
2157 /* Skip over nested parentheses. */
2161 while (*++p && depth > 0 && *p != '\n') {
2174 /* If this is a macro, we have just passed
2175 the arguments and there will be more on
2176 the line before the NULL character that marks
2177 the end of the line token. */
2178 while (*p == ' ' || *p == '\t')
2181 *is_func_or_var = TRUE;
2186 set_construct((*is_func_or_var) ? C_MACRO : C_CONSTANT);
2188 if (!*is_func_or_var && !constantypedefs)
2195 error("internal error: definedef value.", (char *)NULL);
2203 if (toktype == st_C_typedef) {
2205 typdef = ttypedseen;
2211 switch ((unsigned int)toktype) {
2213 set_construct(C_CONSTANT);
2234 /* Do not return here, so the structdef stuff has a chance. */
2237 switch ((unsigned int)toktype) {
2239 set_construct(C_CONSTANT);
2265 * This structdef business is currently only invoked when cblev==0.
2266 * It should be recursively invoked whatever the curly brace level,
2267 * and a stack of states kept, to allow for definitions of structs
2270 * This structdef business is NOT invoked when we are ctags and the
2271 * file is plain C. This is because a struct tag may have the same
2272 * name as another tag, and this loses with ctags.
2274 switch ((unsigned int)toktype) {
2275 case st_C_javastruct:
2276 if (structdef == stagseen)
2277 structdef = scolonseen;
2286 if (typdef == ttypedseen
2287 || (typedefs_and_cplusplus && cblev == 0
2288 && structdef == snone)) {
2289 structdef = skeyseen;
2290 structtype = toktype;
2299 if (structdef == skeyseen) {
2300 /* Save the tag for struct/union/class, for functions and variables
2301 that may be defined inside. */
2303 if (structtype == st_C_struct)
2305 if (structtype == st_C_struct
2306 || structtype == st_C_union || structtype == st_C_class)
2308 structtag = savenstr(str, len);
2310 structtag = "<enum>";
2311 structdef = stagseen;
2315 /* Avoid entering fvdef stuff if typdef is going on. */
2316 if (typdef != tnone) {
2321 /* Detect GNU macros.
2323 DEFUN note for writers of emacs C code:
2324 The DEFUN macro, used in emacs C source code, has a first arg
2325 that is a string (the lisp function name), and a second arg that
2326 is a C function name. Since etags skips strings, the second arg
2327 is tagged. This is unfortunate, as it would be better to tag the
2328 first arg. The simplest way to deal with this problem would be
2329 to name the tag with a name built from the function name, by
2330 removing the initial 'F' character and substituting '-' for '_'.
2331 Anyway, this assumes that the conventions of naming lisp
2332 functions will never change. Currently, this method is not
2333 implemented, so writers of emacs code are recommended to put the
2334 first two args of a DEFUN on the same line. */
2335 if (definedef == dnone && toktype == st_C_gnumacro) {
2336 next_token_is_func = TRUE;
2339 if (next_token_is_func) {
2340 next_token_is_func = FALSE;
2342 *is_func_or_var = TRUE;
2346 /* Detect Objective C constructs. */
2349 switch ((unsigned int)toktype) {
2352 set_construct(C_OBJC);
2358 set_construct(C_OBJC);
2360 objdef = oimplementation;
2368 case oimplementation:
2369 /* Save the class tag for functions or variables defined inside. */
2370 objtag = savenstr(str, len);
2374 /* Save the class tag for categories. */
2375 objtag = savenstr(str, len);
2377 *is_func_or_var = TRUE;
2381 *is_func_or_var = TRUE;
2387 objdef = omethodtag;
2389 grow_linebuffer(&token_name, methodlen + 1);
2390 strncpy(token_name.buffer, str, len);
2391 token_name.buffer[methodlen] = '\0';
2392 token_name.len = methodlen;
2398 objdef = omethodparm;
2402 objdef = omethodtag;
2404 grow_linebuffer(&token_name, methodlen + 1);
2405 strncat(token_name.buffer, str, len);
2406 token_name.len = methodlen;
2411 if (toktype == st_C_objend) {
2412 /* Memory leakage here: the string pointed by objtag is
2413 never released, because many tests would be needed to
2414 avoid breaking on incorrect input code. The amount of
2415 memory leaked here is the sum of the lengths of the
2430 /* A function, variable or enum constant? */
2431 switch ((unsigned int)toktype) {
2433 set_construct(C_CONSTANT);
2439 if (fvdef != finlist && fvdef != fignore && fvdef != vignore)
2440 fvdef = fvnone; /* should be useless */
2446 if (constantypedefs && structdef == sinbody
2447 && structtype == st_C_enum)
2450 oo_browser_construct = C_ENUM_LABEL;
2456 if (fvdef == fvnone) {
2457 fvdef = fvnameseen; /* function or variable */
2458 *is_func_or_var = TRUE;
2472 * This routine finds functions, variables, typedefs,
2473 * #define's, enum constants and struct/union/enum definitions in
2474 * #C syntax and adds them to the list.
2476 #define current_lb_is_new (newndx == curndx)
2477 #define switch_line_buffers() (curndx = 1 - curndx)
2479 #define curlb (lbs[curndx].lb)
2480 #define othlb (lbs[1-curndx].lb)
2481 #define newlb (lbs[newndx].lb)
2482 #define curlinepos (lbs[curndx].linepos)
2483 #define othlinepos (lbs[1-curndx].linepos)
2484 #define newlinepos (lbs[newndx].linepos)
2486 #define CNL_SAVE_DEFINEDEF() \
2488 curlinepos = charno; \
2490 linecharno = charno; \
2491 charno += readline (&curlb, inf); \
2492 lp = curlb.buffer; \
2499 CNL_SAVE_DEFINEDEF(); \
2500 if (savetok.valid) \
2503 savetok.valid = FALSE; \
2505 definedef = dnone; \
2508 void make_C_tag PP((bool isfun));
2509 void make_C_tag(isfun)
2512 /* This function should never be called when tok.valid is FALSE, but
2513 we must protect against invalid input or internal errors. */
2515 if (traditional_tag_style) {
2516 /* This was the original code. Now we call new_pfnote instead,
2517 which uses the new method for naming tags (see new_pfnote). */
2520 if (CTAGS || tok.named)
2521 name = savestr(token_name.buffer);
2523 tok.buffer, tok.linelen, tok.lineno,
2526 new_pfnote(token_name.buffer, token_name.len, isfun,
2527 tok.buffer, tok.linelen, tok.lineno,
2534 void C_entries(c_ext, inf)
2535 int c_ext; /* extension of C */
2536 FILE *inf; /* input file */
2538 register char c; /* latest char read; '\0' for end of line */
2539 register char *lp; /* pointer one beyond the character `c' */
2540 int curndx, newndx; /* indices for current and new lb */
2541 register int tokoff; /* offset in line of start of current token */
2542 register int toklen; /* length of current token */
2543 char *qualifier; /* string used to qualify names */
2544 int qlen; /* length of qualifier */
2545 int cblev; /* current curly brace level */
2546 int parlev; /* current parenthesis level */
2547 bool incomm, inquote, inchar, quotednl, midtoken;
2549 token savetok; /* token saved during preprocessor handling */
2551 /* initialise savetok */
2552 memset(&savetok, 0, sizeof(token));
2554 tokoff = toklen = 0; /* keep compiler quiet */
2555 curndx = newndx = 0;
2566 next_token_is_func = yacc_rules = FALSE;
2567 midtoken = inquote = inchar = incomm = quotednl = FALSE;
2568 tok.valid = savetok.valid = FALSE;
2571 cplpl = (c_ext & C_PLPL) == C_PLPL;
2572 cjava = (c_ext & C_JAVA) == C_JAVA;
2581 while (!feof(inf)) {
2584 /* If we're at the end of the line, the next character is a
2585 '\0'; don't skip it, because it's the thing that tells us
2586 to read the next line. */
2593 } else if (incomm) {
2602 /* Newlines inside comments do not end macro definitions in
2604 CNL_SAVE_DEFINEDEF();
2610 } else if (inquote) {
2616 /* Newlines inside strings do not end macro definitions
2617 in traditional cpp, even though compilers don't
2618 usually accept them. */
2619 CNL_SAVE_DEFINEDEF();
2625 } else if (inchar) {
2628 /* Hmmm, something went wrong. */
2642 if (fvdef != finlist && fvdef != fignore
2643 && fvdef != vignore)
2648 if (fvdef != finlist && fvdef != fignore
2649 && fvdef != vignore)
2657 } else if ( /* cplpl && */ *lp == '/') {
2663 if ((c_ext & YACC) && *lp == '%') {
2664 /* entering or exiting rules section in yacc file */
2670 next_token_is_func = FALSE;
2671 midtoken = inquote = inchar = incomm =
2674 yacc_rules = !yacc_rules;
2679 if (definedef == dnone) {
2681 bool cpptoken = TRUE;
2683 /* Look back on this line. If all blanks, or nonblanks
2684 followed by an end of comment, this is a preprocessor
2686 for (cp = newlb.buffer; cp < lp - 1;
2688 if (!iswhite(*cp)) {
2699 definedef = dsharpseen;
2701 /* if (definedef == dnone) */
2707 /* Consider token only if some complicated conditions are satisfied. */
2708 if ((definedef != dnone
2709 || (cblev == 0 && structdef != scolonseen)
2710 || (cblev == 1 && cplpl && structdef == sinbody)
2711 || (structdef == sinbody && structtype == st_C_enum))
2712 && typdef != tignore
2713 && definedef != dignorerest && fvdef != finlist) {
2716 if (c == ':' && cplpl && *lp == ':'
2717 && begtoken(*(lp + 1))) {
2719 * This handles :: in the middle, but not at the
2720 * beginning of an identifier.
2725 set_construct(C_METHOD);
2728 bool funorvar = FALSE;
2731 || consider_token(newlb.
2741 if (structdef == sinbody
2744 /* function or var defined in C++ class body */
2774 oo_browser_construct
2777 } else if (objdef ==
2779 /* Objective C category */
2812 oo_browser_construct
2815 } else if (objdef ==
2819 /* Objective C method */
2824 oo_browser_construct
2853 /* Also name #define constants,
2854 enumerations and enum_labels.
2855 Conditionalize `funorvar' reference
2856 here or #defines will appear without
2858 -- Bob Weiner, Altrasoft, 4/25/1998 */
2860 ((oo_browser_format || funorvar)
2868 (oo_browser_construct
2872 oo_browser_construct
2885 tok.lineno = lineno;
2887 tokoff + toklen + 1;
2894 if (definedef == dnone
2903 if (current_lb_is_new)
2912 } /* if (endtoken (c)) */
2913 else if (intoken(c)) {
2917 } /* if (midtoken) */
2918 else if (begtoken(c)) {
2919 switch (definedef) {
2921 switch ((unsigned int)fvdef) {
2927 set_construct(C_MACRO);
2929 make_C_tag(TRUE); /* a function */
2940 if (structdef == stagseen && !cjava)
2952 if (!yacc_rules || lp == newlb.buffer + 1) {
2953 tokoff = lp - 1 - newlb.buffer;
2958 } /* if (begtoken) */
2961 /* if must look at token */
2962 /* Detect end of line, colon, comma, semicolon and various braces
2963 after having handled a token. */
2966 if (definedef != dnone)
2968 switch ((unsigned int)objdef) {
2971 make_C_tag(TRUE); /* an Objective C class */
2975 objdef = omethodcolon;
2977 grow_linebuffer(&token_name, methodlen + 1);
2978 strcat(token_name.buffer, ":");
2979 token_name.len = methodlen;
2986 if (structdef == stagseen)
2987 structdef = scolonseen;
2989 switch ((unsigned int)fvdef) {
2992 make_C_tag(FALSE); /* a yacc function */
3006 if (definedef != dnone)
3009 switch ((unsigned int)typdef) {
3012 set_construct(C_TYPE);
3014 make_C_tag(FALSE); /* a typedef */
3019 switch ((unsigned int)fvdef) {
3023 if ((globals && cblev == 0)
3024 || (members && cblev == 1))
3026 make_C_tag(FALSE); /* a variable */
3028 /* if (constantypedefs && structdef == snone)*/
3031 switch ((unsigned int)structtype) {
3033 set_construct(C_ENUMERATION);
3036 set_construct(C_CLASS);
3039 set_construct(C_VARIABLE);
3043 /* Force reset of st_C_enum structtype value. */
3044 structtype = st_none;
3050 /* The following instruction invalidates the token.
3051 Probably the token should be invalidated in all
3052 other cases where some state machine is reset. */
3055 if (structdef == stagseen)
3059 if (definedef != dnone)
3061 switch ((unsigned int)objdef) {
3064 make_C_tag(TRUE); /* an Objective C method */
3072 switch ((unsigned int)fvdef) {
3078 if ((globals && cblev == 0)
3079 || (members && cblev == 1))
3080 make_C_tag(FALSE); /* a variable */
3085 if (structdef == stagseen)
3089 if (definedef != dnone)
3091 if (cblev == 0 && typdef == tend) {
3093 set_construct(C_TYPE);
3096 make_C_tag(FALSE); /* a typedef */
3099 switch ((unsigned int)fvdef) {
3106 if ((globals && cblev == 0)
3107 || (members && cblev == 1))
3108 make_C_tag(FALSE); /* a variable */
3110 if (constantypedefs && structdef == snone) {
3112 switch ((unsigned int)structtype) {
3114 set_construct(C_ENUMERATION);
3117 set_construct(C_CLASS);
3120 set_construct(C_VARIABLE);
3124 /* Force reset of st_C_enum structtype value. */
3125 structtype = st_none;
3135 if (structdef == stagseen)
3139 if (definedef != dnone)
3141 if (objdef == otagseen && parlev == 0)
3142 objdef = oparenseen;
3143 switch ((unsigned int)fvdef) {
3148 if (tok.valid && *lp != '*') {
3149 /* This handles constructs like:
3150 typedef void OperatorFun (int fun); */
3153 set_construct(C_TYPE);
3165 } /* switch (typdef) */
3181 if (definedef != dnone)
3183 if (objdef == ocatseen && parlev == 1) {
3184 make_C_tag(TRUE); /* an Objective C category */
3187 if (--parlev == 0) {
3188 switch ((unsigned int)fvdef) {
3198 if (cblev == 0 && typdef == tend) {
3200 set_construct(C_TYPE);
3203 make_C_tag(FALSE); /* a typedef */
3205 } else if (parlev < 0) /* can happen due to ill-conceived #if's. */
3209 if (definedef != dnone)
3211 if (typdef == ttypedseen)
3213 switch (structdef) {
3214 case skeyseen: /* unnamed struct */
3215 structdef = sinbody;
3216 structtag = "_anonymous_";
3219 case scolonseen: /* named struct */
3220 structdef = sinbody;
3221 make_C_tag(FALSE); /* a struct */
3230 switch ((unsigned int)fvdef) {
3233 set_construct(C_FUNCTION);
3234 /* Ensure function name is recorded.
3235 -- Bob Weiner, Altrasoft */
3238 make_C_tag(TRUE); /* a function */
3244 switch ((unsigned int)objdef) {
3246 make_C_tag(TRUE); /* an Objective C class */
3251 make_C_tag(TRUE); /* an Objective C method */
3255 /* Neutralize `extern "C" {' grot. */
3256 if (cblev == 0 && structdef == snone
3268 if (definedef != dnone)
3270 if (fvdef == fstartlist)
3271 fvdef = fvnone; /* avoid tagging `foo' in `foo (*bar()) ()' */
3274 if (definedef != dnone)
3276 if (!noindentypedefs && lp == newlb.buffer + 1) {
3277 cblev = 0; /* reset curly brace level if first column */
3278 parlev = 0; /* also reset paren level, just in case... */
3279 } else if (cblev > 0)
3282 if (typdef == tinbody)
3284 /* Memory leakage here: the string pointed by structtag is
3285 never released, because I fear to miss something and
3286 break things while freeing the area. The amount of
3287 memory leaked here is the sum of the lengths of the
3289 if (structdef == sinbody)
3290 free (structtag); */
3293 structtag = "<error>";
3295 /* Next line added to avoid any state carryover between
3296 functions. -- Bob Weiner, Altrasoft, 11/19/1997 */
3298 oo_browser_construct = C_NULL;
3303 if (definedef != dnone)
3315 if ((globals && cblev == 0)
3316 || (members && cblev == 1))
3318 make_C_tag(FALSE); /* a variable */
3322 switch ((unsigned int)structtype) {
3328 set_construct(C_CLASS);
3331 /* a global variable */
3337 /* ootags categorizes each tag found whereas etags doesn't.
3338 Set the is_method flag if this tag has been marked as
3339 such by an earlier section of code.
3340 -- Steve Baur, Altrasoft, 5/7/1998 */
3342 (oo_browser_construct ==
3346 /* Force reset of st_C_enum structtype value. */
3347 structtype = st_none;
3356 fvdef = is_method ? fignore : vignore;
3367 if (objdef == oinbody && cblev == 0) {
3368 objdef = omethodsign;
3385 if (definedef != dnone)
3390 /* The above characters cannot follow a function tag in C, so
3391 unmark this as a function entry. For C++, these characters
3392 may follow an `operator' function construct, so skip the
3393 unmarking conditional below.
3394 -- Steve Baur, Altrasoft, 5/7/1998 */
3395 if (fvdef != finlist && fvdef != fignore
3396 && fvdef != vignore)
3403 if (objdef == otagseen) {
3404 make_C_tag(TRUE); /* an Objective C class */
3407 /* If a macro spans multiple lines don't reset its state. */
3409 CNL_SAVE_DEFINEDEF();
3419 } /* while not eof */
3423 * Process either a C++ file or a C file depending on the setting
3426 void default_C_entries(inf)
3429 C_entries(cplusplus ? C_PLPL : 0, inf);
3432 /* Always do plain ANSI C. */
3433 void plain_C_entries(inf)
3439 /* Always do C++. */
3440 void Cplusplus_entries(inf)
3443 C_entries(C_PLPL, inf);
3446 /* Always do Java. */
3447 void Cjava_entries(inf)
3450 C_entries(C_JAVA, inf);
3454 void Cstar_entries(inf)
3457 C_entries(C_STAR, inf);
3460 /* Always do Yacc. */
3461 void Yacc_entries(inf)
3464 C_entries(YACC, inf);
3467 /* A useful macro. */
3468 #define LOOP_ON_INPUT_LINES(file_pointer, line_buffer, char_pointer) \
3469 for (lineno = charno = 0; /* loop initialization */ \
3470 !feof (file_pointer) /* loop test */ \
3471 && (lineno++, /* instructions at start of loop */ \
3472 linecharno = charno, \
3473 charno += readline (&line_buffer, file_pointer), \
3474 char_pointer = lb.buffer, \
3479 * Read a file, but do no processing. This is used to do regexp
3480 * matching on files that have no language defined.
3482 void just_read_file(inf)
3485 register char *dummy;
3487 LOOP_ON_INPUT_LINES(inf, lb, dummy)
3491 /* Fortran parsing */
3493 bool tail PP((char *cp));
3497 register int len = 0;
3499 while (*cp && lowcase(*cp) == lowcase(dbp[len]))
3501 if (*cp == '\0' && !intoken(dbp[len])) {
3510 dbp = skip_spaces(dbp);
3514 dbp = skip_spaces(dbp);
3515 if (strneq(dbp, "(*)", 3)) {
3519 if (!isdigit(*dbp)) {
3520 --dbp; /* force failure */
3525 while (isdigit(*dbp));
3528 void getit PP((FILE * inf));
3534 dbp = skip_spaces(dbp);
3537 linecharno = charno;
3538 charno += readline(&lb, inf);
3543 dbp = skip_spaces(dbp);
3546 && *dbp != '_' && *dbp != '$')
3548 for (cp = dbp + 1; *cp && intoken(*cp); cp++)
3550 pfnote((CTAGS) ? savenstr(dbp, cp - dbp) : NULL, TRUE,
3551 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
3554 void Fortran_functions(inf)
3557 LOOP_ON_INPUT_LINES(inf, lb, dbp) {
3559 dbp++; /* Ratfor escape to fortran */
3560 dbp = skip_spaces(dbp);
3563 switch (lowcase(*dbp)) {
3565 if (tail("integer"))
3573 if (tail("logical"))
3577 if (tail("complex") || tail("character"))
3581 if (tail("double")) {
3582 dbp = skip_spaces(dbp);
3585 if (tail("precision"))
3593 dbp = skip_spaces(dbp);
3596 switch (lowcase(*dbp)) {
3598 if (tail("function"))
3602 if (tail("subroutine"))
3610 if (tail("program")) {
3614 if (tail("procedure"))
3624 * Bob Weiner, Motorola Inc., 4/3/94
3625 * Unix and microcontroller assembly tag handling
3626 * look for '^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]'
3628 void Asm_labels(FILE * inf)
3632 LOOP_ON_INPUT_LINES(inf, lb, cp) {
3633 /* If first char is alphabetic or one of [_.$], test for colon
3634 following identifier. */
3635 if (isalpha(*cp) || *cp == '_' || *cp == '.' || *cp == '$') {
3636 /* Read past label. */
3638 while (isalnum(*cp) || *cp == '_' || *cp == '.'
3641 if (*cp == ':' || isspace(*cp)) {
3642 /* Found end of label, so copy it and add it to the table. */
3645 cp - lb.buffer) : NULL, TRUE,
3646 lb.buffer, cp - lb.buffer + 1, lineno,
3654 * Perl support by Bart Robinson <lomew@cs.utah.edu>
3655 * enhanced by Michael Ernst <mernst@alum.mit.edu>
3656 * Perl sub names: look for /^sub[ \t\n]+[^ \t\n{]+/
3657 * Perl variable names: /^(my|local).../
3659 void Perl_functions(inf)
3664 LOOP_ON_INPUT_LINES(inf, lb, cp) {
3666 && *cp++ == 'u' && *cp++ == 'b' && isspace(*cp++)) {
3667 cp = skip_spaces(cp);
3670 && !isspace(*cp) && *cp != '{'
3675 cp - lb.buffer) : NULL, TRUE,
3676 lb.buffer, cp - lb.buffer + 1, lineno,
3679 } else if (globals /* only if tagging global vars is enabled */
3680 && ((cp = lb.buffer, *cp++ == 'm' && *cp++ == 'y')
3685 && *cp++ == 'a' && *cp++ == 'l'))
3686 && (*cp == '(' || isspace(*cp))) {
3687 /* After "my" or "local", but before any following paren or space. */
3688 char *varname = NULL;
3690 cp = skip_spaces(cp);
3691 if (*cp == '$' || *cp == '@' || *cp == '%') {
3692 char *varstart = ++cp;
3693 while (isalnum(*cp) || *cp == '_')
3695 varname = savenstr(varstart, cp - varstart);
3697 /* Should be examining a variable list at this point;
3698 could insist on seeing an open parenthesis. */
3699 while (*cp != '\0' && *cp != ';' && *cp != '='
3704 /* Perhaps I should back cp up one character, so the TAGS table
3705 doesn't mention (and so depend upon) the following char. */
3706 pfnote((CTAGS) ? savenstr(lb.buffer, cp - lb.buffer) :
3707 varname, FALSE, lb.buffer, cp - lb.buffer + 1,
3708 lineno, linecharno);
3714 * Python support by Eric S. Raymond <esr@thyrsus.com>
3715 * Look for /^def[ \t\n]+[^ \t\n(:]+/ or /^class[ \t\n]+[^ \t\n(:]+/
3717 void Python_functions(inf)
3722 LOOP_ON_INPUT_LINES(inf, lb, cp) {
3724 && *cp++ == 'e' && *cp++ == 'f' && isspace(*cp++)) {
3725 cp = skip_spaces(cp);
3726 while (*cp != '\0' && !isspace(*cp) && *cp != '('
3729 pfnote((char *)NULL, TRUE,
3730 lb.buffer, cp - lb.buffer + 1, lineno,
3738 && *cp++ == 's' && *cp++ == 's' && isspace(*cp++)) {
3739 cp = skip_spaces(cp);
3740 while (*cp != '\0' && !isspace(*cp) && *cp != '('
3743 pfnote((char *)NULL, TRUE,
3744 lb.buffer, cp - lb.buffer + 1, lineno,
3750 /* Idea by Corny de Souza
3751 * Cobol tag functions
3752 * We could look for anything that could be a paragraph name.
3753 * i.e. anything that starts in column 8 is one word and ends in a full stop.
3755 void Cobol_paragraphs(inf)
3758 register char *bp, *ep;
3760 LOOP_ON_INPUT_LINES(inf, lb, bp) {
3765 /* If eoln, compiler option or comment ignore whole line. */
3766 if (bp[-1] != ' ' || !isalnum(bp[0]))
3769 for (ep = bp; isalnum(*ep) || *ep == '-'; ep++)
3772 pfnote((CTAGS) ? savenstr(bp, ep - bp) : NULL, TRUE,
3773 lb.buffer, ep - lb.buffer + 1, lineno,
3778 /* Added by Mosur Mohan, 4/22/88 */
3779 /* Pascal parsing */
3782 * Locates tags for procedures & functions. Doesn't do any type- or
3783 * var-definitions. It does look for the keyword "extern" or
3784 * "forward" immediately following the procedure statement; if found,
3785 * the tag is skipped.
3787 void Pascal_functions(inf)
3790 linebuffer tline; /* mostly copied from C_entries */
3792 int save_lineno, save_len;
3793 char c, *cp, *namebuf;
3795 bool /* each of these flags is TRUE iff: */
3796 incomment, /* point is inside a comment */
3797 inquote, /* point is inside '..' string */
3798 get_tagname, /* point is after PROCEDURE/FUNCTION
3799 keyword, so next item = potential tag */
3800 found_tag, /* point is after a potential tag */
3801 inparms, /* point is within parameter-list */
3802 verify_tag; /* point has passed the parm-list, so the
3803 next token will determine whether this
3804 is a FORWARD/EXTERN to be ignored, or
3805 whether it is a real tag */
3807 save_lcno = save_lineno = save_len = 0; /* keep compiler quiet */
3808 namebuf = NULL; /* keep compiler quiet */
3815 incomment = inquote = FALSE;
3816 found_tag = FALSE; /* have a proc name; check if extern */
3817 get_tagname = FALSE; /* have found "procedure" keyword */
3818 inparms = FALSE; /* found '(' after "proc" */
3819 verify_tag = FALSE; /* check if "extern" is ahead */
3821 while (!feof(inf)) { /* long main loop to get next char */
3823 if (c == '\0') { /* if end of line */
3825 linecharno = charno;
3826 charno += readline(&lb, inf);
3830 if (!((found_tag && verify_tag)
3832 c = *dbp++; /* only if don't need *dbp pointing
3833 to the beginning of the name of
3834 the procedure or function */
3837 if (c == '}') /* within { } comments */
3839 else if (c == '*' && *dbp == ')') { /* within (* *) comments */
3844 } else if (inquote) {
3851 inquote = TRUE; /* found first quote */
3853 case '{': /* found open { comment */
3857 if (*dbp == '*') { /* found open (* comment */
3860 } else if (found_tag) /* found '(' after tag, i.e., parm-list */
3863 case ')': /* end of parms list */
3868 if (found_tag && !inparms) { /* end of proc or fn stmt */
3876 if (found_tag && verify_tag && (*dbp != ' ')) {
3877 /* check if this is an "extern" declaration */
3880 if (lowcase(*dbp == 'e')) {
3881 if (tail("extern")) { /* superfluous, really! */
3885 } else if (lowcase(*dbp) == 'f') {
3886 if (tail("forward")) { /* check for forward reference */
3891 if (found_tag && verify_tag) { /* not external proc, so make tag */
3894 pfnote(namebuf, TRUE,
3895 tline.buffer, save_len, save_lineno,
3900 if (get_tagname) { /* grab name of proc or fn */
3904 /* save all values for later tagging */
3905 grow_linebuffer(&tline, lb.len + 1);
3906 strncpy(tline.buffer, lb.buffer, lb.len);
3907 save_lineno = lineno;
3908 save_lcno = linecharno;
3910 /* grab block name */
3911 for (cp = dbp + 1; *cp != '\0' && !endtoken(*cp); cp++)
3913 namebuf = (CTAGS) ? savenstr(dbp, cp - dbp) : NULL;
3914 dbp = cp; /* set dbp to e-o-token */
3915 save_len = dbp - lb.buffer + 1;
3916 get_tagname = FALSE;
3920 /* and proceed to check for "extern" */
3921 } else if (!incomment && !inquote && !found_tag) {
3922 /* check for proc/fn keywords */
3923 switch (lowcase(c)) {
3925 if (tail("rocedure")) /* c = 'p', dbp has advanced */
3929 if (tail("unction"))
3936 } /* while not eof */
3942 * lisp tag functions
3943 * look for (def or (DEF, quote or QUOTE
3945 int L_isdef PP((char *strp));
3947 register char *strp;
3949 return ((strp[1] == 'd' || strp[1] == 'D')
3950 && (strp[2] == 'e' || strp[2] == 'E')
3951 && (strp[3] == 'f' || strp[3] == 'F'));
3953 int L_isquote PP((char *strp));
3955 register char *strp;
3957 return ((*++strp == 'q' || *strp == 'Q')
3958 && (*++strp == 'u' || *strp == 'U')
3959 && (*++strp == 'o' || *strp == 'O')
3960 && (*++strp == 't' || *strp == 'T')
3961 && (*++strp == 'e' || *strp == 'E')
3962 && isspace(*++strp));
3965 void L_getit PP((void));
3970 if (*dbp == '\'') /* Skip prefix quote */
3972 else if (*dbp == '(') {
3974 dbp += 7; /* Skip "(quote " */
3976 dbp += 1; /* Skip "(" before name in (defstruct (foo)) */
3977 dbp = skip_spaces(dbp);
3980 for (cp = dbp /*+1 */ ;
3981 *cp != '\0' && *cp != '(' && *cp != ' ' && *cp != ')'; cp++)
3986 pfnote((CTAGS) ? savenstr(dbp, cp - dbp) : NULL, TRUE,
3987 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
3990 void Lisp_functions(inf)
3993 LOOP_ON_INPUT_LINES(inf, lb, dbp) {
3994 if (dbp[0] == '(') {
3996 dbp = skip_non_spaces(dbp);
3997 dbp = skip_spaces(dbp);
4000 /* Check for (foo::defmumble name-defined ... */
4003 while (*dbp != '\0' && !isspace(*dbp)
4004 && *dbp != ':' && *dbp != '('
4009 while (*dbp == ':');
4011 if (L_isdef(dbp - 1)) {
4012 dbp = skip_non_spaces(dbp);
4013 dbp = skip_spaces(dbp);
4023 * Postscript tag functions
4024 * Just look for lines where the first character is '/'
4025 * Richard Mlynarik <mly@adoc.xerox.com>
4027 void Postscript_functions(inf)
4030 register char *bp, *ep;
4032 LOOP_ON_INPUT_LINES(inf, lb, bp) {
4035 *ep != '\0' && *ep != ' ' && *ep != '{'; ep++)
4037 pfnote((CTAGS) ? savenstr(bp, ep - bp) : NULL, TRUE,
4038 lb.buffer, ep - lb.buffer + 1, lineno,
4045 * Scheme tag functions
4046 * look for (def... xyzzy
4047 * look for (def... (xyzzy
4048 * look for (def ... ((...(xyzzy ....
4049 * look for (set! xyzzy
4052 void get_scheme PP((void));
4054 void Scheme_functions(inf)
4057 LOOP_ON_INPUT_LINES(inf, lb, dbp) {
4058 if (dbp[0] == '(' && (dbp[1] == 'D' || dbp[1] == 'd')
4059 && (dbp[2] == 'E' || dbp[2] == 'e')
4060 && (dbp[3] == 'F' || dbp[3] == 'f')) {
4061 dbp = skip_non_spaces(dbp);
4062 /* Skip over open parens and white space */
4063 while (isspace(*dbp) || *dbp == '(')
4067 if (dbp[0] == '(' && (dbp[1] == 'S' || dbp[1] == 's')
4068 && (dbp[2] == 'E' || dbp[2] == 'e')
4069 && (dbp[3] == 'T' || dbp[3] == 't')
4070 && (dbp[4] == '!' || dbp[4] == '!')
4071 && (isspace(dbp[5]))) {
4072 dbp = skip_non_spaces(dbp);
4073 dbp = skip_spaces(dbp);
4085 /* Go till you get to white space or a syntactic break */
4087 *cp != '\0' && *cp != '(' && *cp != ')' && !isspace(*cp); cp++)
4089 pfnote((CTAGS) ? savenstr(dbp, cp - dbp) : NULL, TRUE,
4090 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4093 /* Find tags in TeX and LaTeX input files. */
4095 /* TEX_toktab is a table of TeX control sequences that define tags.
4096 Each TEX_tabent records one such control sequence.
4097 CONVERT THIS TO USE THE Stab TYPE!! */
4103 struct TEX_tabent *TEX_toktab = NULL; /* Table with tag tokens */
4105 /* Default set of control sequences to put into TEX_toktab.
4106 The value of environment var TEXTAGS is prepended to this. */
4108 char *TEX_defenv = "\
4109 :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem\
4110 :part:appendix:entry:index";
4112 void TEX_mode PP((FILE * inf));
4113 struct TEX_tabent *TEX_decode_env PP((char *evarname, char *defenv));
4114 int TEX_Token PP((char *cp));
4116 char TEX_esc = '\\';
4117 char TEX_opgrp = '{';
4118 char TEX_clgrp = '}';
4121 * TeX/LaTeX scanning loop.
4123 void TeX_functions(inf)
4129 /* Select either \ or ! as escape character. */
4132 /* Initialize token table once from environment. */
4134 TEX_toktab = TEX_decode_env("TEXTAGS", TEX_defenv);
4136 LOOP_ON_INPUT_LINES(inf, lb, cp) {
4138 /* Look at each esc in line. */
4139 while ((cp = etags_strchr(cp, TEX_esc)) != NULL) {
4142 linecharno += cp - lasthit;
4144 i = TEX_Token(lasthit);
4146 /* We seem to include the TeX command in the tag name.
4148 for (p = lasthit + TEX_toktab[i].len;
4149 *p != '\0' && *p != TEX_clgrp;
4152 pfnote( /*savenstr (lasthit, p-lasthit) */
4154 lb.buffer, lb.len, lineno, linecharno);
4155 break; /* We only tag a line once */
4161 #define TEX_LESC '\\'
4162 #define TEX_SESC '!'
4165 /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
4166 chars accordingly. */
4172 while ((c = getc(inf)) != EOF) {
4173 /* Skip to next line if we hit the TeX comment char. */
4177 else if (c == TEX_LESC || c == TEX_SESC)
4181 if (c == TEX_LESC) {
4193 /* Read environment and prepend it to the default string.
4194 Build token table. */
4195 struct TEX_tabent *TEX_decode_env(evarname, defenv)
4199 register char *env, *p;
4201 struct TEX_tabent *tab;
4204 /* Append default string to environment. */
4205 env = getenv(evarname);
4210 env = concat(oldenv, defenv, "");
4214 /* Allocate a token table */
4215 for (size = 1, p = env; p;)
4216 if ((p = etags_strchr(p, ':')) && *++p != '\0')
4218 /* Add 1 to leave room for null terminator. */
4219 tab = xnew(size + 1, struct TEX_tabent);
4221 /* Unpack environment string into token table. Be careful about */
4222 /* zero-length strings (leading ':', "::" and trailing ':') */
4223 for (i = 0; *env;) {
4224 p = etags_strchr(env, ':');
4225 if (!p) /* End of environment string. */
4226 p = env + strlen(env);
4227 if (p - env > 0) { /* Only non-zero strings. */
4228 tab[i].name = savenstr(env, p - env);
4229 tab[i].len = strlen(tab[i].name);
4235 tab[i].name = NULL; /* Mark end of table. */
4243 /* If the text at CP matches one of the tag-defining TeX command names,
4244 return the pointer to the first occurrence of that command in TEX_toktab.
4245 Otherwise return -1.
4246 Keep the capital `T' in `token' for dumb truncating compilers
4247 (this distinguishes it from `TEX_toktab' */
4253 for (i = 0; TEX_toktab[i].len > 0; i++)
4254 if (strneq(TEX_toktab[i].name, cp, TEX_toktab[i].len))
4260 * Prolog support (rewritten) by Anders Lindgren, Mar. 96
4262 * Assumes that the predicate starts at column 0.
4263 * Only the first clause of a predicate is added.
4265 int prolog_pred PP((char *s, char *last));
4266 void prolog_skip_comment PP((linebuffer * plb, FILE * inf));
4267 int prolog_atom PP((char *s, int pos));
4269 void Prolog_functions(inf)
4280 LOOP_ON_INPUT_LINES(inf, lb, cp) {
4281 if (cp[0] == '\0') /* Empty line */
4283 else if (isspace(cp[0])) /* Not a predicate */
4285 else if (cp[0] == '/' && cp[1] == '*') /* comment. */
4286 prolog_skip_comment(&lb, inf);
4287 else if ((len = prolog_pred(cp, last)) > 0) {
4288 /* Predicate. Store the function name so that we only
4289 generate a tag for the first clause. */
4291 last = xnew(len + 1, char);
4292 else if (len + 1 > allocated)
4293 last = xrnew(last, len + 1, char);
4294 allocated = len + 1;
4295 strncpy(last, cp, len);
4302 void prolog_skip_comment(plb, inf)
4309 for (cp = plb->buffer; *cp != '\0'; cp++)
4310 if (cp[0] == '*' && cp[1] == '/')
4313 linecharno += readline(plb, inf);
4319 * A predicate definition is added if it matches:
4320 * <beginning of line><Prolog Atom><whitespace>(
4322 * It is added to the tags database if it doesn't match the
4323 * name of the previous clause header.
4325 * Return the size of the name of the predicate, or 0 if no header
4328 int prolog_pred(s, last)
4330 char *last; /* Name of last clause. */
4335 pos = prolog_atom(s, 0);
4340 pos = skip_spaces(s + pos) - s;
4342 if ((s[pos] == '(') || (s[pos] == '.')) {
4346 /* Save only the first clause. */
4347 if (last == NULL || len != (int)strlen(last)
4348 || !strneq(s, last, len)) {
4349 pfnote((CTAGS) ? savenstr(s, len) : NULL, TRUE,
4350 s, pos, lineno, linecharno);
4358 * Consume a Prolog atom.
4359 * Return the number of bytes consumed, or -1 if there was an error.
4361 * A prolog atom, in this context, could be one of:
4362 * - An alphanumeric sequence, starting with a lower case letter.
4363 * - A quoted arbitrary string. Single quotes can escape themselves.
4364 * Backslash quotes everything.
4366 int prolog_atom(s, pos)
4374 if (islower(s[pos]) || (s[pos] == '_')) {
4375 /* The atom is unquoted. */
4377 while (isalnum(s[pos]) || (s[pos] == '_')) {
4380 return pos - origpos;
4381 } else if (s[pos] == '\'') {
4385 if (s[pos] == '\'') {
4389 pos++; /* A double quote */
4390 } else if (s[pos] == '\0')
4391 /* Multiline quoted atoms are ignored. */
4393 else if (s[pos] == '\\') {
4394 if (s[pos + 1] == '\0')
4400 return pos - origpos;
4406 * Support for Erlang -- Anders Lindgren, Feb 1996.
4408 * Generates tags for functions, defines, and records.
4410 * Assumes that Erlang functions start at column 0.
4412 int erlang_func PP((char *s, char *last));
4413 void erlang_attribute PP((char *s));
4414 int erlang_atom PP((char *s, int pos));
4416 void Erlang_functions(inf)
4427 LOOP_ON_INPUT_LINES(inf, lb, cp) {
4428 if (cp[0] == '\0') /* Empty line */
4430 else if (isspace(cp[0])) /* Not function nor attribute */
4432 else if (cp[0] == '%') /* comment */
4434 else if (cp[0] == '"') /* Sometimes, strings start in column one */
4436 else if (cp[0] == '-') { /* attribute, e.g. "-define" */
4437 erlang_attribute(cp);
4440 } else if ((len = erlang_func(cp, last)) > 0) {
4442 * Function. Store the function name so that we only
4443 * generates a tag for the first clause.
4446 last = xnew(len + 1, char);
4447 else if (len + 1 > allocated)
4448 last = xrnew(last, len + 1, char);
4449 allocated = len + 1;
4450 strncpy(last, cp, len);
4458 * A function definition is added if it matches:
4459 * <beginning of line><Erlang Atom><whitespace>(
4461 * It is added to the tags database if it doesn't match the
4462 * name of the previous clause header.
4464 * Return the size of the name of the function, or 0 if no function
4467 int erlang_func(s, last)
4469 char *last; /* Name of last clause. */
4474 pos = erlang_atom(s, 0);
4479 pos = skip_spaces(s + pos) - s;
4481 /* Save only the first clause. */
4482 if (s[pos++] == '(' && (last == NULL || len != (int)strlen(last)
4483 || !strneq(s, last, len))) {
4484 pfnote((CTAGS) ? savenstr(s, len) : NULL, TRUE,
4485 s, pos, lineno, linecharno);
4493 * Handle attributes. Currently, tags are generated for defines
4496 * They are on the form:
4497 * -define(foo, bar).
4498 * -define(Foo(M, N), M+N).
4499 * -record(graph, {vtab = notable, cyclic = true}).
4501 void erlang_attribute(s)
4507 if (strneq(s, "-define", 7) || strneq(s, "-record", 7)) {
4508 pos = skip_spaces(s + 7) - s;
4509 if (s[pos++] == '(') {
4510 pos = skip_spaces(s + pos) - s;
4511 len = erlang_atom(s, pos);
4513 pfnote((CTAGS) ? savenstr(&s[pos], len) : NULL,
4514 TRUE, s, pos + len, lineno, linecharno);
4521 * Consume an Erlang atom (or variable).
4522 * Return the number of bytes consumed, or -1 if there was an error.
4524 int erlang_atom(s, pos)
4532 if (isalpha(s[pos]) || s[pos] == '_') {
4533 /* The atom is unquoted. */
4535 while (isalnum(s[pos]) || s[pos] == '_')
4537 return pos - origpos;
4538 } else if (s[pos] == '\'') {
4542 if (s[pos] == '\'') {
4545 } else if (s[pos] == '\0')
4546 /* Multiline quoted atoms are ignored. */
4548 else if (s[pos] == '\\') {
4549 if (s[pos + 1] == '\0')
4555 return pos - origpos;
4560 #ifdef ETAGS_REGEXPS
4562 /* Take a string like "/blah/" and turn it into "blah", making sure
4563 that the first and last characters are the same, and handling
4564 quoted separator characters. Actually, stops on the occurrence of
4565 an unquoted separator. Also turns "\t" into a Tab character.
4566 Returns pointer to terminating separator. Works in place. Null
4567 terminates name string. */
4568 char *scan_separators PP((char *name));
4569 char *scan_separators(name)
4573 char *copyto = name;
4574 bool quoted = FALSE;
4576 for (++name; *name != '\0'; ++name) {
4580 else if (*name == sep)
4583 /* Something else is quoted, so preserve the quote. */
4588 } else if (*name == '\\')
4590 else if (*name == sep)
4596 /* Terminate copied string. */
4601 /* Look at the argument of --regex or --no-regex and do the right
4602 thing. Same for each line of a regexp file. */
4603 void analyse_regex(regex_arg)
4606 if (regex_arg == NULL) {
4607 free_patterns(); /* --no-regex: remove existing regexps */
4610 /* A real --regexp option or a line in a regexp file. */
4611 switch (regex_arg[0]) {
4612 /* Comments in regexp file or null arg to --regex. */
4618 /* Read a regex file. This is recursive and may result in a
4619 loop, which will stop when the file descriptors are exhausted. */
4623 linebuffer regexbuf;
4624 char *regexfile = regex_arg + 1;
4626 /* regexfile is a file containing regexps, one per line. */
4627 regexfp = fopen(regexfile, "r");
4628 if (regexfp == NULL) {
4632 initbuffer(®exbuf);
4633 while (readline_internal(®exbuf, regexfp) > 0)
4634 analyse_regex(regexbuf.buffer);
4635 free(regexbuf.buffer);
4640 /* Regexp to be used for a specific language only. */
4644 char *lang_name = regex_arg + 1;
4647 for (cp = lang_name; *cp != '}'; cp++)
4650 ("unterminated language name in regex: %s",
4655 lang = get_language_from_name(lang_name);
4658 add_regex(cp + 1, lang);
4662 /* Regexp to be used for any language. */
4664 add_regex(regex_arg, NULL);
4669 /* Turn a name, which is an ed-style (but Emacs syntax) regular
4670 expression, into a real regular expression by compiling it. */
4671 void add_regex(regexp_pattern, lang)
4672 char *regexp_pattern;
4677 struct re_pattern_buffer *patbuf;
4680 if (regexp_pattern[strlen(regexp_pattern) - 1] != regexp_pattern[0]) {
4681 error("%s: unterminated regexp", regexp_pattern);
4684 name = scan_separators(regexp_pattern);
4685 if (regexp_pattern[0] == '\0') {
4686 error("null regexp", (char *)NULL);
4689 (void)scan_separators(name);
4691 patbuf = xnew(1, struct re_pattern_buffer);
4692 patbuf->translate = NULL;
4693 patbuf->fastmap = NULL;
4694 patbuf->buffer = NULL;
4695 patbuf->allocated = 0;
4698 re_compile_pattern(regexp_pattern, strlen(regexp_pattern), patbuf);
4700 error("%s while compiling pattern", err);
4705 p_head = xnew(1, pattern);
4706 p_head->regex = savestr(regexp_pattern);
4707 p_head->p_next = pp;
4708 p_head->language = lang;
4709 p_head->pattern = patbuf;
4710 p_head->name_pattern = savestr(name);
4711 p_head->error_signaled = FALSE;
4715 * Do the substitutions indicated by the regular expression and
4718 char *substitute PP((char *in, char *out, struct re_registers * regs));
4719 char *substitute(in, out, regs)
4721 struct re_registers *regs;
4724 int size, dig, diglen;
4729 /* Pass 1: figure out how much to allocate by finding all \N strings. */
4730 if (out[size - 1] == '\\')
4731 fatal("pattern error in \"%s\"", out);
4732 for (t = etags_strchr(out, '\\');
4733 t != NULL; t = etags_strchr(t + 2, '\\'))
4734 if (isdigit(t[1])) {
4736 diglen = regs->end[dig] - regs->start[dig];
4741 /* Allocate space and do the substitutions. */
4742 result = xnew(size + 1, char);
4744 for (t = result; *out != '\0'; out++)
4745 if (*out == '\\' && isdigit(*++out)) {
4746 /* Using "dig2" satisfies my debugger. Bleah. */
4748 diglen = regs->end[dig] - regs->start[dig];
4749 strncpy(t, in + regs->start[dig], diglen);
4755 if (DEBUG && (t > result + size || t - result != strlen(result)))
4761 /* Deallocate all patterns. */
4762 void free_patterns()
4765 while (p_head != NULL) {
4766 pp = p_head->p_next;
4767 free(p_head->regex);
4768 free(p_head->name_pattern);
4775 #endif /* ETAGS_REGEXPS */
4776 /* Initialize a linebuffer for use */
4777 void initbuffer(lbp)
4781 lbp->buffer = xnew(200, char);
4785 * Read a line of text from `stream' into `lbp', excluding the
4786 * newline or CR-NL, if any. Return the number of characters read from
4787 * `stream', which is the length of the line including the newline.
4789 * On DOS or Windows we do not count the CR character, if any, before the
4790 * NL, in the returned length; this mirrors the behavior of emacs on those
4791 * platforms (for text files, it translates CR-NL to NL as it reads in the
4794 long readline_internal(lbp, stream)
4796 register FILE *stream;
4798 char *buffer = lbp->buffer;
4799 register char *p = lbp->buffer;
4800 register char *pend;
4803 pend = p + lbp->size; /* Separate to avoid 386/IX compiler bug. */
4806 register int c = getc(stream);
4808 /* We're at the end of linebuffer: expand it. */
4810 buffer = xrnew(buffer, lbp->size, char);
4811 p += buffer - lbp->buffer;
4812 pend = buffer + lbp->size;
4813 lbp->buffer = buffer;
4821 if (p > buffer && p[-1] == '\r') {
4832 lbp->len = p - buffer;
4834 return lbp->len + chars_deleted;
4838 * Like readline_internal, above, but in addition try to match the
4839 * input line against relevant regular expressions.
4841 long readline(lbp, stream)
4845 /* Read new line. */
4846 long result = readline_internal(lbp, stream);
4847 #ifdef ETAGS_REGEXPS
4851 /* Match against relevant patterns. */
4853 for (pp = p_head; pp != NULL; pp = pp->p_next) {
4854 /* Only use generic regexps or those for the current language. */
4855 if (pp->language != NULL && pp->language != curlang)
4859 re_match(pp->pattern, lbp->buffer, lbp->len, 0,
4864 if (!pp->error_signaled) {
4865 error("error while matching \"%s\"",
4867 pp->error_signaled = TRUE;
4874 /* Match occurred. Construct a tag. */
4875 if (pp->name_pattern[0] != '\0') {
4876 /* Make a named tag. */
4877 char *name = substitute(lbp->buffer,
4882 pfnote(name, TRUE, lbp->buffer,
4886 /* Make an unnamed tag. */
4887 pfnote((char *)NULL, TRUE,
4888 lbp->buffer, match, lineno,
4894 #endif /* ETAGS_REGEXPS */
4900 * Return a pointer to a space of size strlen(cp)+1 allocated
4901 * with xnew where the string CP has been copied.
4906 return savenstr(cp, strlen(cp));
4910 * Return a pointer to a space of size LEN+1 allocated with xnew where
4911 * the string CP has been copied for at most the first LEN characters.
4913 char *savenstr(cp, len)
4919 dp = xnew(len + 1, char);
4920 strncpy(dp, cp, len);
4926 * Return the ptr in sp at which the character c last
4927 * appears; NULL if not found
4929 * Identical to System V strrchr, included for portability.
4931 char *etags_strrchr(sp, c)
4946 * Return the ptr in sp at which the character c first
4947 * appears; NULL if not found
4949 * Identical to System V strchr, included for portability.
4951 char *etags_strchr(sp, c)
4962 /* Skip spaces, return new pointer. */
4963 char *skip_spaces(cp)
4966 while (isspace(*cp)) /* isspace('\0')==FALSE */
4971 /* Skip non spaces, return new pointer. */
4972 char *skip_non_spaces(cp)
4975 while (!iswhite(*cp)) /* iswhite('\0')==TRUE */
4980 /* Print error message and exit. */
4995 void suggest_asking_for_help()
4997 fprintf(stderr, "\tTry `%s %s' for a complete list of options.\n",
5008 /* Print error message. `s1' is printf control string, `s2' is arg for it. */
5010 const char *s1, *s2;
5012 fprintf(stderr, "%s: ", progname);
5013 fprintf(stderr, s1, s2);
5014 fprintf(stderr, "\n");
5017 /* Return a newly-allocated string whose contents
5018 concatenate those of s1, s2, s3. */
5019 char *concat(s1, s2, s3)
5022 int len1 = strlen(s1), len2 = strlen(s2), len3 = strlen(s3);
5023 char *result = xnew(len1 + len2 + len3 + 1, char);
5026 strcpy(result + len1, s2);
5027 strcpy(result + len1 + len2, s3);
5028 result[len1 + len2 + len3] = '\0';
5033 /* Does the same work as the system V getcwd, but does not need to
5034 guess the buffer size in advance. */
5035 char *etags_getcwd()
5039 char *path = xnew(bufsize, char);
5041 while (getcwd(path, bufsize) == NULL) {
5042 if (errno != ERANGE)
5046 path = xnew(bufsize, char);
5049 canonicalize_filename(path);
5052 #else /* not HAVE_GETCWD */
5057 pipe = (FILE *) popen("pwd 2>/dev/null", "r");
5058 if (pipe == NULL || readline_internal(&path, pipe) == 0)
5063 #endif /* not HAVE_GETCWD */
5066 /* Return a newly allocated string containing the file name of FILE
5067 relative to the absolute directory DIR (which should end with a slash). */
5068 char *relative_filename(file, dir)
5071 char *fp, *dp, *afn, *res;
5075 /* Find the common root of file and dir (with a trailing slash). */
5076 afn = absolute_filename(file, cwd);
5079 while (*fp++ == *dp++)
5081 fp--, dp--; /* back to the first differing char */
5082 do /* look at the equal chars until '/' */
5085 fp ++; /* Advance past the '/' */
5087 /* Build a sequence of "../" strings for the resulting relative file name. */
5089 while ((dp = etags_strchr(dp + 1, '/')) != NULL)
5091 res_left = 3 * i + strlen(fp);
5092 res = xnew( res_left + 1, char);
5094 for ( ; i-- > 0 ; res_left -= 4 )
5095 strncat(res, "../", res_left );
5097 /* Add the file name relative to the common root of file and dir. */
5098 strncat(res, fp, res_left);
5104 /* Return a newly allocated string containing the absolute file name
5105 of FILE given DIR (which should end with a slash). */
5106 char *absolute_filename(file, dir)
5109 char *slashp, *cp, *res;
5111 if (filename_is_absolute(file))
5112 res = savestr(file);
5114 res = concat(dir, file, "");
5116 /* Delete the "/dirname/.." and "/." substrings. */
5117 slashp = etags_strchr(res, '/');
5118 while (slashp != NULL && slashp[0] != '\0') {
5119 if (slashp[1] == '.') {
5120 if (slashp[2] == '.'
5121 && (slashp[3] == '/' || slashp[3] == '\0')) {
5125 while (cp >= res && !filename_is_absolute(cp));
5127 cp = slashp; /* the absolute name begins with "/.." */
5128 strcpy(cp, slashp + 3);
5131 } else if (slashp[2] == '/' || slashp[2] == '\0') {
5132 strcpy(slashp, slashp + 2);
5137 slashp = etags_strchr(slashp + 1, '/');
5141 return savestr("/");
5146 /* Return a newly allocated string containing the absolute
5147 file name of dir where FILE resides given DIR (which should
5148 end with a slash). */
5149 char *absolute_dirname(file, dir)
5155 canonicalize_filename(file);
5156 slashp = etags_strrchr(file, '/');
5158 return savestr(dir);
5161 res = absolute_filename(file, dir);
5167 /* Whether the argument string is an absolute file name. The argument
5168 string must have been canonicalized with canonicalize_filename. */
5169 bool filename_is_absolute(fn)
5172 return (fn[0] == '/');
5175 /* Translate backslashes into slashes. Works in place. */
5176 void canonicalize_filename(fn)
5182 /* Increase the size of a linebuffer. */
5183 void grow_linebuffer(lbp, toksize)
5187 while (lbp->size < toksize)
5189 lbp->buffer = xrnew(lbp->buffer, lbp->size, char);
5192 /* Like malloc but get fatal error if memory is exhausted. */
5196 long *result = (long *)malloc(size);
5198 fatal("virtual memory exhausted", (char *)NULL);
5202 long *xrealloc(ptr, size)
5206 long *result = (long *)realloc(ptr, size);
5208 fatal("virtual memory exhausted", (char *)NULL);