1 /* Tags file maker to go with GNU Emacs -*- coding: latin-1 -*-
2 Copyright (C) 1984, 87, 88, 89, 93, 94, 95, 98, 99, 2000, 2001
3 Free Software Foundation, Inc. and Ken Arnold
5 This file is not considered part of GNU Emacs.
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 2 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, write to the Free Software Foundation,
19 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
23 * Ctags originally by Ken Arnold.
24 * Fortran added by Jim Kleckner.
25 * Ed Pelegri-Llopart added C typedefs.
26 * Gnu Emacs TAGS format and modifications by RMS?
27 * 1989 Sam Kendall added C++.
28 * 1993 Francesco Potortì reorganised C and C++ based on work by Joe Wells.
29 * 1994 Regexp tags by Tom Tromey.
30 * 2001 Nested classes by Francesco Potortì based on work by Mykola Dzyuba.
32 * Francesco Potortì <pot@gnu.org> has maintained it since 1993.
35 char pot_etags_version[] = "@(#) pot revision number is 14.15";
45 # define NDEBUG /* disable assert */
48 #if defined(__STDC__) && (__STDC__ || defined(__SUNPRO_C))
49 # define P_(proto) proto
56 /* On some systems, Emacs defines static as nothing for the sake
57 of unexec. We don't want that here since we don't use unexec. */
59 # define ETAGS_REGEXPS /* use the regexp features */
60 # define LONG_OPTIONS /* accept long options */
61 #endif /* HAVE_CONFIG_H */
64 # define _GNU_SOURCE 1 /* enables some compiler checks on GNU */
67 /* WIN32_NATIVE is for XEmacs.
68 MSDOS, WINDOWSNT, DOS_NT are for Emacs. */
73 #endif /* WIN32_NATIVE */
79 # include <sys/param.h>
81 # ifndef HAVE_CONFIG_H
83 # include <sys/config.h>
95 # define MAXPATHLEN _MAX_PATH
101 # endif /* undef HAVE_GETCWD */
102 #else /* !WINDOWSNT */
107 extern char *getenv ();
109 #endif /* !WINDOWSNT */
114 # if defined (HAVE_GETCWD) && !defined (WINDOWSNT)
115 extern char *getcwd (char *buf, size_t size);
117 #endif /* HAVE_UNISTD_H */
125 #include <sys/types.h>
126 #include <sys/stat.h>
130 # undef assert /* some systems have a buggy assert.h */
131 # define assert(x) ((void) 0)
134 #if !defined (S_ISREG) && defined (S_IFREG)
135 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
141 # define getopt_long(argc,argv,optstr,lopts,lind) getopt (argc, argv, optstr)
143 extern int optind, opterr;
144 #endif /* LONG_OPTIONS */
148 #endif /* ETAGS_REGEXPS */
150 /* Define CTAGS to make the program "ctags" compatible with the usual one.
151 Leave it undefined to make the program "etags", which makes emacs-style
152 tag tables and tags typedefs, #defines and struct/union/enum by default. */
160 /* Exit codes for success and failure. */
169 #define streq(s,t) (assert((s)!=NULL || (t)!=NULL), !strcmp (s, t))
170 #define strneq(s,t,n) (assert((s)!=NULL || (t)!=NULL), !strncmp (s, t, n))
172 #define CHARS 256 /* 2^sizeof(char) */
173 #define CHAR(x) ((unsigned int)(x) & (CHARS - 1))
174 #define iswhite(c) (_wht[CHAR(c)]) /* c is white */
175 #define notinname(c) (_nin[CHAR(c)]) /* c is not in a name */
176 #define begtoken(c) (_btk[CHAR(c)]) /* c can start token */
177 #define intoken(c) (_itk[CHAR(c)]) /* c can be in token */
178 #define endtoken(c) (_etk[CHAR(c)]) /* c ends tokens */
180 #define ISALNUM(c) isalnum (CHAR(c))
181 #define ISALPHA(c) isalpha (CHAR(c))
182 #define ISDIGIT(c) isdigit (CHAR(c))
183 #define ISLOWER(c) islower (CHAR(c))
185 #define lowcase(c) tolower (CHAR(c))
186 #define upcase(c) toupper (CHAR(c))
190 * xnew, xrnew -- allocate, reallocate storage
192 * SYNOPSIS: Type *xnew (int n, Type);
193 * void xrnew (OldPointer, int n, Type);
196 # include "chkmalloc.h"
197 # define xnew(n,Type) ((Type *) trace_malloc (__FILE__, __LINE__, \
198 (n) * sizeof (Type)))
199 # define xrnew(op,n,Type) ((op) = (Type *) trace_realloc (__FILE__, __LINE__, \
200 (char *) (op), (n) * sizeof (Type)))
202 # define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type)))
203 # define xrnew(op,n,Type) ((op) = (Type *) xrealloc ( \
204 (char *) (op), (n) * sizeof (Type)))
209 typedef void Lang_function P_((FILE *));
214 char *command; /* Takes one arg and decompresses to stdout */
220 Lang_function *function;
226 typedef struct node_st
227 { /* sorting structure */
228 char *name; /* function or type name */
229 char *file; /* file name */
230 bool is_func; /* use pattern or line no */
231 bool been_warned; /* set if noticed dup */
232 int lno; /* line number tag is on */
233 long cno; /* character number line starts on */
234 char *pat; /* search pattern */
235 struct node_st *left, *right; /* left and right sons */
239 * A `linebuffer' is a structure which holds a line of text.
240 * `readline_internal' reads a line from a stream into a linebuffer
241 * and works regardless of the length of the line.
242 * SIZE is the size of BUFFER, LEN is the length of the string in
243 * BUFFER after readline reads it.
252 /* Many compilers barf on this:
253 Lang_function Ada_funcs;
254 so let's write it this way */
255 static void Ada_funcs P_((FILE *));
256 static void Asm_labels P_((FILE *));
257 static void C_entries P_((int c_ext, FILE *));
258 static void default_C_entries P_((FILE *));
259 static void plain_C_entries P_((FILE *));
260 static void Cjava_entries P_((FILE *));
261 static void Cobol_paragraphs P_((FILE *));
262 static void Cplusplus_entries P_((FILE *));
263 static void Cstar_entries P_((FILE *));
264 static void Erlang_functions P_((FILE *));
265 static void Fortran_functions P_((FILE *));
266 static void Yacc_entries P_((FILE *));
267 static void Lisp_functions P_((FILE *));
268 static void Makefile_targets P_((FILE *));
269 static void Pascal_functions P_((FILE *));
270 static void Perl_functions P_((FILE *));
271 static void Postscript_functions P_((FILE *));
272 static void Prolog_functions P_((FILE *));
273 static void Python_functions P_((FILE *));
274 static void Scheme_functions P_((FILE *));
275 static void TeX_commands P_((FILE *));
276 static void Texinfo_nodes P_((FILE *));
277 static void just_read_file P_((FILE *));
279 static void print_language_names P_((void));
280 static void print_version P_((void));
281 static void print_help P_((void));
282 int main P_((int, char **));
283 static int number_len P_((long));
285 static compressor *get_compressor_from_suffix P_((char *, char **));
286 static language *get_language_from_langname P_((char *));
287 static language *get_language_from_interpreter P_((char *));
288 static language *get_language_from_filename P_((char *));
289 static int total_size_of_entries P_((node *));
290 static long readline P_((linebuffer *, FILE *));
291 static long readline_internal P_((linebuffer *, FILE *));
292 static void get_tag P_((char *));
295 static void analyse_regex P_((char *, bool));
296 static void add_regex P_((char *, bool, language *));
297 static void free_patterns P_((void));
298 #endif /* ETAGS_REGEXPS */
299 static void error P_((const char *, const char *));
300 static void suggest_asking_for_help P_((void));
301 void fatal P_((char *, char *));
302 static void pfatal P_((char *));
303 static void add_node P_((node *, node **));
305 static void init P_((void));
306 static void initbuffer P_((linebuffer *));
307 static void find_entries P_((char *, FILE *));
308 static void free_tree P_((node *));
309 static void pfnote P_((char *, bool, char *, int, int, long));
310 static void new_pfnote P_((char *, int, bool, char *, int, int, long));
311 static void process_file P_((char *));
312 static void put_entries P_((node *));
313 static void takeprec P_((void));
315 static char *concat P_((char *, char *, char *));
316 static char *skip_spaces P_((char *));
317 static char *skip_non_spaces P_((char *));
318 static char *savenstr P_((char *, int));
319 static char *savestr P_((char *));
320 static char *etags_strchr P_((const char *, int));
321 static char *etags_strrchr P_((const char *, int));
322 static char *etags_getcwd P_((void));
323 static char *relative_filename P_((char *, char *));
324 static char *absolute_filename P_((char *, char *));
325 static char *absolute_dirname P_((char *, char *));
326 static bool filename_is_absolute P_((char *f));
327 static void canonicalize_filename P_((char *));
328 static void linebuffer_setlen P_((linebuffer *, int));
329 long *xmalloc P_((unsigned int));
330 long *xrealloc P_((char *, unsigned int));
333 char searchar = '/'; /* use /.../ searches */
335 char *tagfile; /* output file */
336 char *progname; /* name this program was invoked with */
337 char *cwd; /* current working directory */
338 char *tagfiledir; /* directory of tagfile */
339 FILE *tagf; /* ioptr for tags file */
341 char *curfile; /* current input file name */
342 language *curlang; /* current language */
344 int lineno; /* line number of current line */
345 long charno; /* current character number */
346 long linecharno; /* charno of start of current line */
347 char *dbp; /* pointer to start of current tag */
349 node *head; /* the head of the binary tree of tags */
351 linebuffer lb; /* the current line */
353 /* boolean "functions" (see init) */
354 bool _wht[CHARS], _nin[CHARS], _itk[CHARS], _btk[CHARS], _etk[CHARS];
357 *white = " \f\t\n\r\v",
359 *nonam = " \f\t\n\r(=,[;",
360 /* token ending chars */
361 *endtk = " \t\n\r\"'#()[]{}=-+%*/&|^~!<>;,.:?",
362 /* token starting chars */
363 *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~@",
364 /* valid in-token chars */
365 *midtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789";
367 bool append_to_tagfile; /* -a: append to tags */
368 /* The following four default to TRUE for etags, but to FALSE for ctags. */
369 bool typedefs; /* -t: create tags for C and Ada typedefs */
370 bool typedefs_or_cplusplus; /* -T: create tags for C typedefs, level */
371 /* 0 struct/enum/union decls, and C++ */
372 /* member functions. */
373 bool constantypedefs; /* -d: create tags for C #define, enum */
374 /* constants and variables. */
375 /* -D: opposite of -d. Default under ctags. */
376 bool declarations; /* --declarations: tag them and extern in C&Co*/
377 bool globals; /* create tags for global variables */
378 bool members; /* create tags for C member variables */
379 bool update; /* -u: update tags */
380 bool vgrind_style; /* -v: create vgrind style index output */
381 bool no_warnings; /* -w: suppress warnings */
382 bool cxref_style; /* -x: create cxref style output */
383 bool cplusplus; /* .[hc] means C++, not C */
384 bool noindentypedefs; /* -I: ignore indentation in C */
385 bool packages_only; /* --packages-only: in Ada, only tag packages*/
388 struct option longopts[] =
390 { "packages-only", no_argument, &packages_only, TRUE },
391 { "append", no_argument, NULL, 'a' },
392 { "backward-search", no_argument, NULL, 'B' },
393 { "c++", no_argument, NULL, 'C' },
394 { "cxref", no_argument, NULL, 'x' },
395 { "defines", no_argument, NULL, 'd' },
396 { "declarations", no_argument, &declarations, TRUE },
397 { "no-defines", no_argument, NULL, 'D' },
398 { "globals", no_argument, &globals, TRUE },
399 { "no-globals", no_argument, &globals, FALSE },
400 { "help", no_argument, NULL, 'h' },
401 { "help", no_argument, NULL, 'H' },
402 { "ignore-indentation", no_argument, NULL, 'I' },
403 { "include", required_argument, NULL, 'i' },
404 { "language", required_argument, NULL, 'l' },
405 { "members", no_argument, &members, TRUE },
406 { "no-members", no_argument, &members, FALSE },
407 { "no-warn", no_argument, NULL, 'w' },
408 { "output", required_argument, NULL, 'o' },
410 { "regex", required_argument, NULL, 'r' },
411 { "no-regex", no_argument, NULL, 'R' },
412 { "ignore-case-regex", required_argument, NULL, 'c' },
413 #endif /* ETAGS_REGEXPS */
414 { "typedefs", no_argument, NULL, 't' },
415 { "typedefs-and-c++", no_argument, NULL, 'T' },
416 { "update", no_argument, NULL, 'u' },
417 { "version", no_argument, NULL, 'V' },
418 { "vgrind", no_argument, NULL, 'v' },
421 #endif /* LONG_OPTIONS */
424 /* Structure defining a regular expression. Elements are
425 the compiled pattern, and the name string. */
426 typedef struct pattern
428 struct pattern *p_next;
431 struct re_pattern_buffer *pattern;
432 struct re_registers regs;
437 /* List of all regexps. */
438 pattern *p_head = NULL;
440 /* How many characters in the character set. (From regex.c.) */
441 #define CHAR_SET_SIZE 256
442 /* Translation table for case-insensitive matching. */
443 char lc_trans[CHAR_SET_SIZE];
444 #endif /* ETAGS_REGEXPS */
446 compressor compressors[] =
448 { "z", "gzip -d -c"},
449 { "Z", "gzip -d -c"},
450 { "gz", "gzip -d -c"},
451 { "GZ", "gzip -d -c"},
452 { "bz2", "bzip2 -d -c" },
460 /* Non-NULL if language fixed. */
461 language *forced_lang = NULL;
464 char *Ada_suffixes [] =
465 { "ads", "adb", "ada", NULL };
468 char *Asm_suffixes [] = { "a", /* Unix assembler */
469 "asm", /* Microcontroller assembly */
470 "def", /* BSO/Tasking definition includes */
471 "inc", /* Microcontroller include files */
472 "ins", /* Microcontroller include files */
473 "s", "sa", /* Unix assembler */
474 "S", /* cpp-processed Unix assembler */
475 "src", /* BSO/Tasking C compiler output */
479 /* Note that .c and .h can be considered C++, if the --c++ flag was
480 given, or if the `class' keyowrd is met inside the file.
481 That is why default_C_entries is called for these. */
482 char *default_C_suffixes [] =
485 char *Cplusplus_suffixes [] =
486 { "C", "c++", "cc", "cpp", "cxx", "H", "h++", "hh", "hpp", "hxx",
487 "M", /* Objective C++ */
488 "pdb", /* Postscript with C syntax */
491 char *Cjava_suffixes [] =
494 char *Cobol_suffixes [] =
495 { "COB", "cob", NULL };
497 char *Cstar_suffixes [] =
498 { "cs", "hs", NULL };
500 char *Erlang_suffixes [] =
501 { "erl", "hrl", NULL };
503 char *Fortran_suffixes [] =
504 { "F", "f", "f90", "for", NULL };
506 char *Lisp_suffixes [] =
507 { "cl", "clisp", "el", "l", "lisp", "LSP", "lsp", "ml", NULL };
509 char *Makefile_filenames [] =
510 { "Makefile", "makefile", "GNUMakefile", "Makefile.in", "Makefile.am", NULL};
512 char *Pascal_suffixes [] =
513 { "p", "pas", NULL };
515 char *Perl_suffixes [] =
516 { "pl", "pm", NULL };
517 char *Perl_interpreters [] =
518 { "perl", "@PERL@", NULL };
520 char *plain_C_suffixes [] =
521 { "lm", /* Objective lex file */
522 "m", /* Objective C file */
523 "pc", /* Pro*C file */
526 char *Postscript_suffixes [] =
527 { "ps", "psw", NULL }; /* .psw is for PSWrap */
529 char *Prolog_suffixes [] =
532 char *Python_suffixes [] =
535 /* Can't do the `SCM' or `scm' prefix with a version number. */
536 char *Scheme_suffixes [] =
537 { "oak", "sch", "scheme", "SCM", "scm", "SM", "sm", "ss", "t", NULL };
539 char *TeX_suffixes [] =
540 { "bib", "clo", "cls", "ltx", "sty", "TeX", "tex", NULL };
542 char *Texinfo_suffixes [] =
543 { "texi", "texinfo", "txi", NULL };
545 char *Yacc_suffixes [] =
546 { "y", "y++", "ym", "yxx", "yy", NULL }; /* .ym is Objective yacc file */
549 * Table of languages.
551 * It is ok for a given function to be listed under more than one
552 * name. I just didn't.
555 language lang_names [] =
557 { "ada", Ada_funcs, NULL, Ada_suffixes, NULL },
558 { "asm", Asm_labels, NULL, Asm_suffixes, NULL },
559 { "c", default_C_entries, NULL, default_C_suffixes, NULL },
560 { "c++", Cplusplus_entries, NULL, Cplusplus_suffixes, NULL },
561 { "c*", Cstar_entries, NULL, Cstar_suffixes, NULL },
562 { "cobol", Cobol_paragraphs, NULL, Cobol_suffixes, NULL },
563 { "erlang", Erlang_functions, NULL, Erlang_suffixes, NULL },
564 { "fortran", Fortran_functions, NULL, Fortran_suffixes, NULL },
565 { "java", Cjava_entries, NULL, Cjava_suffixes, NULL },
566 { "lisp", Lisp_functions, NULL, Lisp_suffixes, NULL },
567 { "makefile", Makefile_targets, Makefile_filenames, NULL, NULL },
568 { "pascal", Pascal_functions, NULL, Pascal_suffixes, NULL },
569 { "perl", Perl_functions, NULL, Perl_suffixes, Perl_interpreters },
570 { "postscript", Postscript_functions, NULL, Postscript_suffixes, NULL },
571 { "proc", plain_C_entries, NULL, plain_C_suffixes, NULL },
572 { "prolog", Prolog_functions, NULL, Prolog_suffixes, NULL },
573 { "python", Python_functions, NULL, Python_suffixes, NULL },
574 { "scheme", Scheme_functions, NULL, Scheme_suffixes, NULL },
575 { "tex", TeX_commands, NULL, TeX_suffixes, NULL },
576 { "texinfo", Texinfo_nodes, NULL, Texinfo_suffixes, NULL },
577 { "yacc", Yacc_entries, NULL, Yacc_suffixes, NULL },
578 { "auto", NULL }, /* default guessing scheme */
579 { "none", just_read_file }, /* regexp matching only */
580 { NULL, NULL } /* end of list */
585 print_language_names ()
590 puts ("\nThese are the currently supported languages, along with the\n\
591 default file names and dot suffixes:");
592 for (lang = lang_names; lang->name != NULL; lang++)
594 printf (" %-*s", 10, lang->name);
595 if (lang->filenames != NULL)
596 for (name = lang->filenames; *name != NULL; name++)
597 printf (" %s", *name);
598 if (lang->suffixes != NULL)
599 for (ext = lang->suffixes; *ext != NULL; ext++)
600 printf (" .%s", *ext);
603 puts ("Where `auto' means use default language for files based on file\n\
604 name suffix, and `none' means only do regexp processing on files.\n\
605 If no language is specified and no matching suffix is found,\n\
606 the first line of the file is read for a sharp-bang (#!) sequence\n\
607 followed by the name of an interpreter. If no such sequence is found,\n\
608 Fortran is tried first; if no tags are found, C is tried next.\n\
609 When parsing any C file, a \"class\" keyword switches to C++.\n\
610 Compressed files are supported using gzip and bzip2.");
614 # define EMACS_NAME "GNU Emacs"
617 # define VERSION "21"
622 printf ("%s (%s %s)\n", (CTAGS) ? "ctags" : "etags", EMACS_NAME, VERSION);
623 puts ("Copyright (C) 1999 Free Software Foundation, Inc. and Ken Arnold");
624 puts ("This program is distributed under the same terms as Emacs");
632 printf ("Usage: %s [options] [[regex-option ...] file-name] ...\n\
634 These are the options accepted by %s.\n", progname, progname);
636 puts ("You may use unambiguous abbreviations for the long option names.");
638 puts ("Long option names do not work with this executable, as it is not\n\
639 linked with GNU getopt.");
640 #endif /* LONG_OPTIONS */
641 puts ("A - as file name means read names from stdin (one per line).");
643 printf (" Absolute names are stored in the output file as they are.\n\
644 Relative ones are stored relative to the output file's directory.");
647 puts ("-a, --append\n\
648 Append tag entries to existing tags file.");
650 puts ("--packages-only\n\
651 For Ada files, only generate tags for packages .");
654 puts ("-B, --backward-search\n\
655 Write the search commands for the tag entries using '?', the\n\
656 backward-search command instead of '/', the forward-search command.");
658 /* This option is mostly obsolete, because etags can now automatically
659 detect C++. Retained for backward compatibility and for debugging and
660 experimentation. In principle, we could want to tag as C++ even
661 before any "class" keyword.
663 Treat files whose name suffix defaults to C language as C++ files.");
666 puts ("--declarations\n\
667 In C and derived languages, create tags for function declarations,");
669 puts ("\tand create tags for extern variables if --globals is used.");
672 ("\tand create tags for extern variables unless --no-globals is used.");
675 puts ("-d, --defines\n\
676 Create tag entries for C #define constants and enum constants, too.");
678 puts ("-D, --no-defines\n\
679 Don't create tag entries for C #define constants and enum constants.\n\
680 This makes the tags file smaller.");
684 puts ("-i FILE, --include=FILE\n\
685 Include a note in tag file indicating that, when searching for\n\
686 a tag, one should also consult the tags file FILE after\n\
687 checking the current file.");
688 puts ("-l LANG, --language=LANG\n\
689 Force the following files to be considered as written in the\n\
690 named language up to the next --language=LANG option.");
695 Create tag entries for global variables in some languages.");
697 puts ("--no-globals\n\
698 Do not create tag entries for global variables in some\n\
699 languages. This makes the tags file smaller.");
701 Create tag entries for member variables in C and derived languages.");
704 puts ("-r /REGEXP/, --regex=/REGEXP/ or --regex=@regexfile\n\
705 Make a tag for each line matching pattern REGEXP in the following\n\
706 files. {LANGUAGE}/REGEXP/ uses REGEXP for LANGUAGE files only.\n\
707 regexfile is a file containing one REGEXP per line.\n\
708 REGEXP is anchored (as if preceded by ^).\n\
709 The form /REGEXP/NAME/ creates a named tag.\n\
710 For example Tcl named tags can be created with:\n\
711 --regex=/proc[ \\t]+\\([^ \\t]+\\)/\\1/.");
712 puts ("-c /REGEXP/, --ignore-case-regex=/REGEXP/ or --ignore-case-regex=@regexfile\n\
713 Like -r, --regex but ignore case when matching expressions.");
714 puts ("-R, --no-regex\n\
715 Don't create tags from regexps for the following files.");
716 #endif /* ETAGS_REGEXPS */
717 puts ("-o FILE, --output=FILE\n\
718 Write the tags to FILE.");
719 puts ("-I, --ignore-indentation\n\
720 Don't rely on indentation quite as much as normal. Currently,\n\
721 this means not to assume that a closing brace in the first\n\
722 column is the final brace of a function or structure\n\
723 definition in C and C++.");
727 puts ("-t, --typedefs\n\
728 Generate tag entries for C and Ada typedefs.");
729 puts ("-T, --typedefs-and-c++\n\
730 Generate tag entries for C typedefs, C struct/enum/union tags,\n\
731 and C++ member functions.");
732 puts ("-u, --update\n\
733 Update the tag entries for the given files, leaving tag\n\
734 entries for other files in place. Currently, this is\n\
735 implemented by deleting the existing entries for the given\n\
736 files and then rewriting the new entries at the end of the\n\
737 tags file. It is often faster to simply rebuild the entire\n\
738 tag file than to use this.");
739 puts ("-v, --vgrind\n\
740 Generates an index of items intended for human consumption,\n\
741 similar to the output of vgrind. The index is sorted, and\n\
742 gives the page number of each item.");
743 puts ("-w, --no-warn\n\
744 Suppress warning messages about entries defined in multiple\n\
746 puts ("-x, --cxref\n\
747 Like --vgrind, but in the style of cxref, rather than vgrind.\n\
748 The output uses line numbers instead of page numbers, but\n\
749 beyond that the differences are cosmetic; try both to see\n\
753 puts ("-V, --version\n\
754 Print the version of the program.\n\
756 Print this help message.");
758 print_language_names ();
761 puts ("Report bugs to bug-gnu-emacs@gnu.org");
775 /* This structure helps us allow mixing of --lang and file names. */
778 enum argument_type arg_type;
780 language *lang; /* language of the regexp */
783 #ifdef VMS /* VMS specific functions */
787 /* This is a BUG! ANY arbitrary limit is a BUG!
788 Won't someone please fix this? */
789 #define MAX_FILE_SPEC_LEN 255
792 char body[MAX_FILE_SPEC_LEN + 1];
796 v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names
797 returning in each successive call the next file name matching the input
798 spec. The function expects that each in_spec passed
799 to it will be processed to completion; in particular, up to and
800 including the call following that in which the last matching name
801 is returned, the function ignores the value of in_spec, and will
802 only start processing a new spec with the following call.
803 If an error occurs, on return out_spec contains the value
804 of in_spec when the error occurred.
806 With each successive file name returned in out_spec, the
807 function's return value is one. When there are no more matching
808 names the function returns zero. If on the first call no file
809 matches in_spec, or there is any other error, -1 is returned.
814 #define OUTSIZE MAX_FILE_SPEC_LEN
820 static long context = 0;
821 static struct dsc$descriptor_s o;
822 static struct dsc$descriptor_s i;
823 static bool pass1 = TRUE;
830 o.dsc$a_pointer = (char *) out;
831 o.dsc$w_length = (short)OUTSIZE;
832 i.dsc$a_pointer = in;
833 i.dsc$w_length = (short)strlen(in);
834 i.dsc$b_dtype = DSC$K_DTYPE_T;
835 i.dsc$b_class = DSC$K_CLASS_S;
836 o.dsc$b_dtype = DSC$K_DTYPE_VT;
837 o.dsc$b_class = DSC$K_CLASS_VS;
839 if ((status = lib$find_file(&i, &o, &context, 0, 0)) == RMS$_NORMAL)
841 out->body[out->curlen] = EOS;
844 else if (status == RMS$_NMF)
848 strcpy(out->body, in);
851 lib$find_file_end(&context);
857 v1.01 nmm 19-Aug-85 gfnames - return in successive calls the
858 name of each file specified by the provided arg expanding wildcards.
861 gfnames (arg, p_error)
865 static vspec filename = {MAX_FILE_SPEC_LEN, "\0"};
867 switch (fn_exp (&filename, arg))
871 return filename.body;
877 return filename.body;
881 #ifndef OLD /* Newer versions of VMS do provide `system'. */
885 error ("%s", "system() function not implemented under VMS");
889 #define VERSION_DELIM ';'
890 char *massage_name (s)
896 if (*s == VERSION_DELIM)
914 unsigned int nincluded_files;
915 char **included_files;
918 int current_arg, file_count;
919 linebuffer filename_lb;
925 _fmode = O_BINARY; /* all of files are treated as binary files */
930 included_files = xnew (argc, char *);
934 /* Allocate enough no matter what happens. Overkill, but each one
936 argbuffer = xnew (argc, argument);
939 /* Set syntax for regular expression routines. */
940 re_set_syntax (RE_SYNTAX_EMACS | RE_INTERVALS);
941 /* Translation table for case-insensitive search. */
942 for (i = 0; i < CHAR_SET_SIZE; i++)
943 lc_trans[i] = lowcase (i);
944 #endif /* ETAGS_REGEXPS */
947 * If etags, always find typedefs and structure tags. Why not?
948 * Also default to find macro constants, enum constants and
953 typedefs = typedefs_or_cplusplus = constantypedefs = TRUE;
955 declarations = FALSE;
965 optstring = "-aCdDf:Il:o:r:c:RStTi:BuvxwVhH";
967 optstring = "-aCdDf:Il:o:StTi:BuvxwVhH";
968 #endif /* ETAGS_REGEXPS */
971 optstring = optstring + 1;
972 #endif /* LONG_OPTIONS */
974 opt = getopt_long (argc, argv, optstring, longopts, 0);
981 /* If getopt returns 0, then it has already processed a
982 long-named option. We should do nothing. */
986 /* This means that a file name has been seen. Record it. */
987 argbuffer[current_arg].arg_type = at_filename;
988 argbuffer[current_arg].what = optarg;
993 /* Common options. */
994 case 'a': append_to_tagfile = TRUE; break;
995 case 'C': cplusplus = TRUE; break;
996 case 'd': constantypedefs = TRUE; break;
997 case 'D': constantypedefs = FALSE; break;
998 case 'f': /* for compatibility with old makefiles */
1002 error ("-o option may only be given once.", (char *)NULL);
1003 suggest_asking_for_help ();
1008 case 'S': /* for backward compatibility */
1009 noindentypedefs = TRUE;
1013 language *lang = get_language_from_langname (optarg);
1016 argbuffer[current_arg].lang = lang;
1017 argbuffer[current_arg].arg_type = at_language;
1022 #ifdef ETAGS_REGEXPS
1024 argbuffer[current_arg].arg_type = at_regexp;
1025 argbuffer[current_arg].what = optarg;
1029 argbuffer[current_arg].arg_type = at_regexp;
1030 argbuffer[current_arg].what = NULL;
1034 argbuffer[current_arg].arg_type = at_icregexp;
1035 argbuffer[current_arg].what = optarg;
1038 #endif /* ETAGS_REGEXPS */
1050 typedefs = typedefs_or_cplusplus = TRUE;
1055 included_files[nincluded_files++] = optarg;
1058 /* Ctags options. */
1059 case 'B': searchar = '?'; break;
1060 case 'u': update = TRUE; break;
1061 case 'v': vgrind_style = TRUE; /*FALLTHRU*/
1062 case 'x': cxref_style = TRUE; break;
1063 case 'w': no_warnings = TRUE; break;
1066 suggest_asking_for_help ();
1070 for (; optind < argc; ++optind)
1072 argbuffer[current_arg].arg_type = at_filename;
1073 argbuffer[current_arg].what = argv[optind];
1078 if (nincluded_files == 0 && file_count == 0)
1080 error ("no input files specified.", (char *)NULL);
1081 suggest_asking_for_help ();
1084 if (tagfile == NULL)
1085 tagfile = CTAGS ? "tags" : "TAGS";
1086 cwd = etags_getcwd (); /* the current working directory */
1087 if (cwd[strlen (cwd) - 1] != '/')
1090 cwd = concat (oldcwd, "/", "");
1093 if (streq (tagfile, "-"))
1096 tagfiledir = absolute_dirname (tagfile, cwd);
1098 init (); /* set up boolean "functions" */
1101 initbuffer (&filename_lb);
1105 if (streq (tagfile, "-"))
1109 /* Switch redirected `stdout' to binary mode (setting `_fmode'
1110 doesn't take effect until after `stdout' is already open). */
1111 if (!isatty (fileno (stdout)))
1112 setmode (fileno (stdout), O_BINARY);
1116 tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1122 * Loop through files finding functions.
1124 for (i = 0; i < current_arg; ++i)
1126 switch (argbuffer[i].arg_type)
1129 forced_lang = argbuffer[i].lang;
1131 #ifdef ETAGS_REGEXPS
1133 analyse_regex (argbuffer[i].what, FALSE);
1136 analyse_regex (argbuffer[i].what, TRUE);
1141 while ((this_file = gfnames (argbuffer[i].what, &got_err)) != NULL)
1145 error ("can't find file %s\n", this_file);
1150 this_file = massage_name (this_file);
1153 this_file = argbuffer[i].what;
1155 /* Input file named "-" means read file names from stdin
1156 (one per line) and use them. */
1157 if (streq (this_file, "-"))
1158 while (readline_internal (&filename_lb, stdin) > 0)
1159 process_file (filename_lb.buffer);
1161 process_file (this_file);
1169 #ifdef ETAGS_REGEXPS
1171 #endif /* ETAGS_REGEXPS */
1175 while (nincluded_files-- > 0)
1176 fprintf (tagf, "\f\n%s,include\n", *included_files++);
1182 /* If CTAGS, we are here. process_file did not write the tags yet,
1183 because we want them ordered. Let's do it now. */
1195 for (i = 0; i < current_arg; ++i)
1197 if (argbuffer[i].arg_type != at_filename)
1200 "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS",
1201 tagfile, argbuffer[i].what, tagfile);
1202 if (system (cmd) != GOOD)
1203 fatal ("failed to execute shell command", (char *)NULL);
1205 append_to_tagfile = TRUE;
1208 tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1219 sprintf (cmd, "sort %s -o %s", tagfile, tagfile);
1220 exit (system (cmd));
1228 * Return a compressor given the file name. If EXTPTR is non-zero,
1229 * return a pointer into FILE where the compressor-specific
1230 * extension begins. If no compressor is found, NULL is returned
1231 * and EXTPTR is not significant.
1232 * Idea by Vladimir Alexiev <vladimir@cs.ualberta.ca> (1998)
1235 get_compressor_from_suffix (file, extptr)
1240 char *slash, *suffix;
1242 /* This relies on FN to be after canonicalize_filename,
1243 so we don't need to consider backslashes on DOS_NT. */
1244 slash = etags_strrchr (file, '/');
1245 suffix = etags_strrchr (file, '.');
1246 if (suffix == NULL || suffix < slash)
1251 /* Let those poor souls who live with DOS 8+3 file name limits get
1252 some solace by treating foo.cgz as if it were foo.c.gz, etc.
1253 Only the first do loop is run if not MSDOS */
1256 for (compr = compressors; compr->suffix != NULL; compr++)
1257 if (streq (compr->suffix, suffix))
1260 break; /* do it only once: not really a loop */
1263 } while (*suffix != '\0');
1270 * Return a language given the name.
1273 get_language_from_langname (name)
1279 error ("empty language name", (char *)NULL);
1282 for (lang = lang_names; lang->name != NULL; lang++)
1283 if (streq (name, lang->name))
1285 error ("unknown language \"%s\"", name);
1293 * Return a language given the interpreter name.
1296 get_language_from_interpreter (interpreter)
1302 if (interpreter == NULL)
1304 for (lang = lang_names; lang->name != NULL; lang++)
1305 if (lang->interpreters != NULL)
1306 for (iname = lang->interpreters; *iname != NULL; iname++)
1307 if (streq (*iname, interpreter))
1316 * Return a language given the file name.
1319 get_language_from_filename (file)
1323 char **name, **ext, *suffix;
1325 /* Try whole file name first. */
1326 for (lang = lang_names; lang->name != NULL; lang++)
1327 if (lang->filenames != NULL)
1328 for (name = lang->filenames; *name != NULL; name++)
1329 if (streq (*name, file))
1332 /* If not found, try suffix after last dot. */
1333 suffix = etags_strrchr (file, '.');
1337 for (lang = lang_names; lang->name != NULL; lang++)
1338 if (lang->suffixes != NULL)
1339 for (ext = lang->suffixes; *ext != NULL; ext++)
1340 if (streq (*ext, suffix))
1348 * This routine is called on each file argument.
1354 struct stat stat_buf;
1357 char *compressed_name, *uncompressed_name;
1358 char *ext, *real_name;
1361 canonicalize_filename (file);
1362 if (streq (file, tagfile) && !streq (tagfile, "-"))
1364 error ("skipping inclusion of %s in self.", file);
1367 if ((compr = get_compressor_from_suffix (file, &ext)) == NULL)
1369 compressed_name = NULL;
1370 real_name = uncompressed_name = savestr (file);
1374 real_name = compressed_name = savestr (file);
1375 uncompressed_name = savenstr (file, ext - file);
1378 /* If the canonicalised uncompressed name has already be dealt with,
1379 skip it silently, else add it to the list. */
1381 typedef struct processed_file
1384 struct processed_file *next;
1386 static processed_file *pf_head = NULL;
1387 register processed_file *fnp;
1389 for (fnp = pf_head; fnp != NULL; fnp = fnp->next)
1390 if (streq (uncompressed_name, fnp->filename))
1393 pf_head = xnew (1, struct processed_file);
1394 pf_head->filename = savestr (uncompressed_name);
1395 pf_head->next = fnp;
1398 if (stat (real_name, &stat_buf) != 0)
1400 /* Reset real_name and try with a different name. */
1402 if (compressed_name != NULL) /* try with the given suffix */
1404 if (stat (uncompressed_name, &stat_buf) == 0)
1405 real_name = uncompressed_name;
1407 else /* try all possible suffixes */
1409 for (compr = compressors; compr->suffix != NULL; compr++)
1411 compressed_name = concat (file, ".", compr->suffix);
1412 if (stat (compressed_name, &stat_buf) != 0)
1416 char *suf = compressed_name + strlen (file);
1417 size_t suflen = strlen (compr->suffix) + 1;
1418 for ( ; suf[1]; suf++, suflen--)
1420 memmove (suf, suf + 1, suflen);
1421 if (stat (compressed_name, &stat_buf) == 0)
1423 real_name = compressed_name;
1427 if (real_name != NULL)
1430 free (compressed_name);
1431 compressed_name = NULL;
1435 real_name = compressed_name;
1440 if (real_name == NULL)
1445 } /* try with a different name */
1447 if (!S_ISREG (stat_buf.st_mode))
1449 error ("skipping %s: it is not a regular file.", real_name);
1452 if (real_name == compressed_name)
1454 char *cmd = concat (compr->command, " ", real_name);
1455 inf = (FILE *) popen (cmd, "r");
1459 inf = fopen (real_name, "r");
1466 find_entries (uncompressed_name, inf);
1468 if (real_name == compressed_name)
1477 if (filename_is_absolute (uncompressed_name))
1479 /* file is an absolute file name. Canonicalise it. */
1480 filename = absolute_filename (uncompressed_name, cwd);
1484 /* file is a file name relative to cwd. Make it relative
1485 to the directory of the tags file. */
1486 filename = relative_filename (uncompressed_name, tagfiledir);
1488 fprintf (tagf, "\f\n%s,%d\n", filename, total_size_of_entries (head));
1496 if (compressed_name) free(compressed_name);
1497 if (uncompressed_name) free(uncompressed_name);
1502 * This routine sets up the boolean pseudo-functions which work
1503 * by setting boolean flags dependent upon the corresponding character.
1504 * Every char which is NOT in that string is not a white char. Therefore,
1505 * all of the array "_wht" is set to FALSE, and then the elements
1506 * subscripted by the chars in "white" are set to TRUE. Thus "_wht"
1507 * of a char is TRUE if it is the string "white", else FALSE.
1515 for (i = 0; i < CHARS; i++)
1516 iswhite(i) = notinname(i) = begtoken(i) = intoken(i) = endtoken(i) = FALSE;
1517 for (sp = white; *sp != '\0'; sp++) iswhite (*sp) = TRUE;
1518 for (sp = nonam; *sp != '\0'; sp++) notinname (*sp) = TRUE;
1519 notinname('\0') = notinname('\n');
1520 for (sp = begtk; *sp != '\0'; sp++) begtoken (*sp) = TRUE;
1521 begtoken('\0') = begtoken('\n');
1522 for (sp = midtk; *sp != '\0'; sp++) intoken (*sp) = TRUE;
1523 intoken('\0') = intoken('\n');
1524 for (sp = endtk; *sp != '\0'; sp++) endtoken (*sp) = TRUE;
1525 endtoken('\0') = endtoken('\n');
1529 * This routine opens the specified file and calls the function
1530 * which finds the function and type definitions.
1532 node *last_node = NULL;
1535 find_entries (file, inf)
1541 node *old_last_node;
1543 /* Memory leakage here: the string pointed by curfile is
1544 never released, because curfile is copied into np->file
1545 for each node, to be used in CTAGS mode. The amount of
1546 memory leaked here is the sum of the lengths of the
1548 curfile = savestr (file);
1550 /* If user specified a language, use it. */
1552 if (lang != NULL && lang->function != NULL)
1555 lang->function (inf);
1559 /* Try to guess the language given the file name. */
1560 lang = get_language_from_filename (file);
1561 if (lang != NULL && lang->function != NULL)
1564 lang->function (inf);
1568 /* Look for sharp-bang as the first two characters. */
1569 if (readline_internal (&lb, inf) > 0
1571 && lb.buffer[0] == '#'
1572 && lb.buffer[1] == '!')
1576 /* Set lp to point at the first char after the last slash in the
1577 line or, if no slashes, at the first nonblank. Then set cp to
1578 the first successive blank and terminate the string. */
1579 lp = etags_strrchr (lb.buffer+2, '/');
1583 lp = skip_spaces (lb.buffer + 2);
1584 cp = skip_non_spaces (lp);
1587 if (strlen (lp) > 0)
1589 lang = get_language_from_interpreter (lp);
1590 if (lang != NULL && lang->function != NULL)
1593 lang->function (inf);
1598 /* We rewind here, even if inf may be a pipe. We fail if the
1599 length of the first line is longer than the pipe block size,
1600 which is unlikely. */
1604 old_last_node = last_node;
1605 curlang = get_language_from_langname ("fortran");
1606 Fortran_functions (inf);
1608 /* No Fortran entries found. Try C. */
1609 if (old_last_node == last_node)
1611 /* We do not tag if rewind fails.
1612 Only the file name will be recorded in the tags file. */
1614 curlang = get_language_from_langname (cplusplus ? "c++" : "c");
1615 default_C_entries (inf);
1623 pfnote (name, is_func, linestart, linelen, lno, cno)
1624 char *name; /* tag name, or NULL if unnamed */
1625 bool is_func; /* tag is a function */
1626 char *linestart; /* start of the line where tag is */
1627 int linelen; /* length of the line where tag is */
1628 int lno; /* line number */
1629 long cno; /* character number */
1633 if (CTAGS && name == NULL)
1636 np = xnew (1, node);
1638 /* If ctags mode, change name "main" to M<thisfilename>. */
1639 if (CTAGS && !cxref_style && streq (name, "main"))
1641 register char *fp = etags_strrchr (curfile, '/');
1642 np->name = concat ("M", fp == NULL ? curfile : fp + 1, "");
1643 fp = etags_strrchr (np->name, '.');
1644 if (fp != NULL && fp[1] != '\0' && fp[2] == '\0')
1649 np->been_warned = FALSE;
1651 np->is_func = is_func;
1653 /* Our char numbers are 0-base, because of C language tradition?
1654 ctags compatibility? old versions compatibility? I don't know.
1655 Anyway, since emacs's are 1-base we expect etags.el to take care
1656 of the difference. If we wanted to have 1-based numbers, we would
1657 uncomment the +1 below. */
1658 np->cno = cno /* + 1 */ ;
1659 np->left = np->right = NULL;
1660 if (CTAGS && !cxref_style)
1662 if (strlen (linestart) < 50)
1663 np->pat = concat (linestart, "$", "");
1665 np->pat = savenstr (linestart, 50);
1668 np->pat = savenstr (linestart, linelen);
1670 add_node (np, &head);
1674 * TAGS format specification
1675 * Idea by Sam Kendall <kendall@mv.mv.com> (1997)
1677 * pfnote should emit the optimized form [unnamed tag] only if:
1678 * 1. name does not contain any of the characters " \t\r\n(),;";
1679 * 2. linestart contains name as either a rightmost, or rightmost but
1680 * one character, substring;
1681 * 3. the character, if any, immediately before name in linestart must
1682 * be one of the characters " \t(),;";
1683 * 4. the character, if any, immediately after name in linestart must
1684 * also be one of the characters " \t(),;".
1686 * The real implementation uses the notinname() macro, which recognises
1687 * characters slightly different form " \t\r\n(),;". See the variable
1690 #define traditional_tag_style TRUE
1692 new_pfnote (name, namelen, is_func, linestart, linelen, lno, cno)
1693 char *name; /* tag name, or NULL if unnamed */
1694 int namelen; /* tag length */
1695 bool is_func; /* tag is a function */
1696 char *linestart; /* start of the line where tag is */
1697 int linelen; /* length of the line where tag is */
1698 int lno; /* line number */
1699 long cno; /* character number */
1707 for (cp = name; !notinname (*cp); cp++)
1709 if (*cp == '\0') /* rule #1 */
1711 cp = linestart + linelen - namelen;
1712 if (notinname (linestart[linelen-1]))
1713 cp -= 1; /* rule #4 */
1714 if (cp >= linestart /* rule #2 */
1716 || notinname (cp[-1])) /* rule #3 */
1717 && strneq (name, cp, namelen)) /* rule #2 */
1718 named = FALSE; /* use unnamed tag */
1723 name = savenstr (name, namelen);
1726 pfnote (name, is_func, linestart, linelen, lno, cno);
1731 * recurse on left children, iterate on right children.
1739 register node *node_right = np->right;
1740 free_tree (np->left);
1741 if (np->name != NULL)
1751 * Adds a node to the tree of nodes. In etags mode, we don't keep
1752 * it sorted; we just keep a linear list. In ctags mode, maintain
1753 * an ordered tree, with no attempt at balancing.
1755 * add_node is the only function allowed to add nodes, so it can
1759 add_node (np, cur_node_p)
1760 node *np, **cur_node_p;
1763 register node *cur_node = *cur_node_p;
1765 if (cur_node == NULL)
1775 if (last_node == NULL)
1776 fatal ("internal error in add_node", (char *)NULL);
1777 last_node->right = np;
1783 dif = strcmp (np->name, cur_node->name);
1786 * If this tag name matches an existing one, then
1787 * do not add the node, but maybe print a warning.
1791 if (streq (np->file, cur_node->file))
1795 fprintf (stderr, "Duplicate entry in file %s, line %d: %s\n",
1796 np->file, lineno, np->name);
1797 fprintf (stderr, "Second entry ignored\n");
1800 else if (!cur_node->been_warned && !no_warnings)
1804 "Duplicate entry in files %s and %s: %s (Warning only)\n",
1805 np->file, cur_node->file, np->name);
1806 cur_node->been_warned = TRUE;
1811 /* Actually add the node */
1812 add_node (np, dif < 0 ? &cur_node->left : &cur_node->right);
1826 /* Output subentries that precede this one */
1827 put_entries (np->left);
1829 /* Output this entry */
1833 if (np->name != NULL)
1834 fprintf (tagf, "%s\177%s\001%d,%ld\n",
1835 np->pat, np->name, np->lno, np->cno);
1837 fprintf (tagf, "%s\177%d,%ld\n",
1838 np->pat, np->lno, np->cno);
1842 if (np->name == NULL)
1843 error ("internal error: NULL name in ctags mode.", (char *)NULL);
1848 fprintf (stdout, "%s %s %d\n",
1849 np->name, np->file, (np->lno + 63) / 64);
1851 fprintf (stdout, "%-16s %3d %-16s %s\n",
1852 np->name, np->lno, np->file, np->pat);
1856 fprintf (tagf, "%s\t%s\t", np->name, np->file);
1860 putc (searchar, tagf);
1863 for (sp = np->pat; *sp; sp++)
1865 if (*sp == '\\' || *sp == searchar)
1869 putc (searchar, tagf);
1872 { /* a typedef; text pattern inadequate */
1873 fprintf (tagf, "%d", np->lno);
1879 /* Output subentries that follow this one */
1880 put_entries (np->right);
1883 /* Length of a number's decimal representation. */
1889 while ((num /= 10) > 0)
1895 * Return total number of characters that put_entries will output for
1896 * the nodes in the subtree of the specified node. Works only if
1897 * we are not ctags, but called only in that case. This count
1898 * is irrelevant with the new tags.el, but is still supplied for
1899 * backward compatibility.
1902 total_size_of_entries (np)
1910 for (total = 0; np != NULL; np = np->right)
1912 /* Count left subentries. */
1913 total += total_size_of_entries (np->left);
1915 /* Count this entry */
1916 total += strlen (np->pat) + 1;
1917 total += number_len ((long) np->lno) + 1 + number_len (np->cno) + 1;
1918 if (np->name != NULL)
1919 total += 1 + strlen (np->name); /* \001name */
1927 #define C_EXT 0x00fff /* C extensions */
1928 #define C_PLAIN 0x00000 /* C */
1929 #define C_PLPL 0x00001 /* C++ */
1930 #define C_STAR 0x00003 /* C* */
1931 #define C_JAVA 0x00005 /* JAVA */
1932 #define C_AUTO 0x01000 /* C, but switch to C++ if `class' is met */
1933 #define YACC 0x10000 /* yacc file */
1936 * The C symbol tables.
1941 st_C_objprot, st_C_objimpl, st_C_objend,
1947 st_C_struct, st_C_extern, st_C_enum, st_C_define, st_C_typedef, st_C_typespec
1950 static unsigned int hash P_((const char *, unsigned int));
1951 static struct C_stab_entry * in_word_set P_((const char *, unsigned int));
1952 static enum sym_type C_symtype P_((char *, int, int));
1954 /* Feed stuff between (but not including) %[ and %] lines to:
1955 gperf -c -k 1,3 -o -p -r -t
1956 then put a `static' keyword in front of the in_word_set function.
1958 struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
1962 while, 0, st_C_ignore
1963 switch, 0, st_C_ignore
1964 return, 0, st_C_ignore
1965 @interface, 0, st_C_objprot
1966 @protocol, 0, st_C_objprot
1967 @implementation,0, st_C_objimpl
1968 @end, 0, st_C_objend
1969 import, C_JAVA, st_C_ignore
1970 package, C_JAVA, st_C_ignore
1971 friend, C_PLPL, st_C_ignore
1972 extends, C_JAVA, st_C_javastruct
1973 implements, C_JAVA, st_C_javastruct
1974 interface, C_JAVA, st_C_struct
1975 class, 0, st_C_class
1976 namespace, C_PLPL, st_C_struct
1977 domain, C_STAR, st_C_struct
1978 union, 0, st_C_struct
1979 struct, 0, st_C_struct
1980 extern, 0, st_C_extern
1982 typedef, 0, st_C_typedef
1983 define, 0, st_C_define
1984 operator, C_PLPL, st_C_operator
1985 bool, C_PLPL, st_C_typespec
1986 long, 0, st_C_typespec
1987 short, 0, st_C_typespec
1988 int, 0, st_C_typespec
1989 char, 0, st_C_typespec
1990 float, 0, st_C_typespec
1991 double, 0, st_C_typespec
1992 signed, 0, st_C_typespec
1993 unsigned, 0, st_C_typespec
1994 auto, 0, st_C_typespec
1995 void, 0, st_C_typespec
1996 static, 0, st_C_typespec
1997 const, 0, st_C_typespec
1998 volatile, 0, st_C_typespec
1999 explicit, C_PLPL, st_C_typespec
2000 mutable, C_PLPL, st_C_typespec
2001 typename, C_PLPL, st_C_typespec
2002 # DEFUN used in emacs, the next three used in glibc (SYSCALL only for mach).
2003 DEFUN, 0, st_C_gnumacro
2004 SYSCALL, 0, st_C_gnumacro
2005 ENTRY, 0, st_C_gnumacro
2006 PSEUDO, 0, st_C_gnumacro
2007 # These are defined inside C functions, so currently they are not met.
2008 # EXFUN used in glibc, DEFVAR_* in emacs.
2009 #EXFUN, 0, st_C_gnumacro
2010 #DEFVAR_, 0, st_C_gnumacro
2012 and replace lines between %< and %> with its output,
2013 then make in_word_set static. */
2015 /* C code produced by gperf version 2.7.1 (19981006 egcs) */
2016 /* Command-line: gperf -c -k 1,3 -o -p -r -t */
2017 struct C_stab_entry { char *name; int c_ext; enum sym_type type; };
2019 #define TOTAL_KEYWORDS 46
2020 #define MIN_WORD_LENGTH 2
2021 #define MAX_WORD_LENGTH 15
2022 #define MIN_HASH_VALUE 13
2023 #define MAX_HASH_VALUE 121
2024 /* maximum key range = 109, duplicates = 0 */
2031 register const char *str;
2032 register unsigned int len;
2034 static unsigned char asso_values[] =
2036 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2037 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2038 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2039 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2040 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2041 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2042 122, 122, 122, 122, 57, 122, 122, 122, 55, 6,
2043 60, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2044 51, 122, 122, 10, 2, 122, 122, 122, 122, 122,
2045 122, 122, 122, 122, 122, 122, 122, 2, 52, 59,
2046 49, 38, 56, 41, 122, 22, 122, 122, 9, 32,
2047 33, 60, 26, 122, 1, 28, 46, 59, 44, 51,
2048 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2049 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2050 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2051 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2052 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2053 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2054 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2055 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2056 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2057 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2058 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2059 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2060 122, 122, 122, 122, 122, 122, 122, 122, 122, 122,
2061 122, 122, 122, 122, 122, 122
2063 register int hval = len;
2069 hval += asso_values[(unsigned char)str[2]];
2072 hval += asso_values[(unsigned char)str[0]];
2081 static struct C_stab_entry *
2082 in_word_set (str, len)
2083 register const char *str;
2084 register unsigned int len;
2086 static struct C_stab_entry wordlist[] =
2088 {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2089 {""}, {""}, {""}, {""},
2090 {"ENTRY", 0, st_C_gnumacro},
2091 {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2093 {"if", 0, st_C_ignore},
2095 {"SYSCALL", 0, st_C_gnumacro},
2096 {""}, {""}, {""}, {""}, {""}, {""}, {""},
2097 {"struct", 0, st_C_struct},
2098 {"static", 0, st_C_typespec},
2099 {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2100 {"long", 0, st_C_typespec},
2101 {""}, {""}, {""}, {""}, {""},
2102 {"auto", 0, st_C_typespec},
2103 {"return", 0, st_C_ignore},
2104 {"import", C_JAVA, st_C_ignore},
2106 {"switch", 0, st_C_ignore},
2108 {"implements", C_JAVA, st_C_javastruct},
2110 {"for", 0, st_C_ignore},
2111 {"volatile", 0, st_C_typespec},
2113 {"PSEUDO", 0, st_C_gnumacro},
2115 {"char", 0, st_C_typespec},
2116 {"class", 0, st_C_class},
2117 {"@protocol", 0, st_C_objprot},
2119 {"void", 0, st_C_typespec},
2120 {"int", 0, st_C_typespec},
2121 {"explicit", C_PLPL, st_C_typespec},
2123 {"namespace", C_PLPL, st_C_struct},
2124 {"signed", 0, st_C_typespec},
2126 {"interface", C_JAVA, st_C_struct},
2127 {"while", 0, st_C_ignore},
2128 {"typedef", 0, st_C_typedef},
2129 {"typename", C_PLPL, st_C_typespec},
2131 {"friend", C_PLPL, st_C_ignore},
2132 {"mutable", C_PLPL, st_C_typespec},
2133 {"union", 0, st_C_struct},
2134 {"domain", C_STAR, st_C_struct},
2136 {"extern", 0, st_C_extern},
2137 {"extends", C_JAVA, st_C_javastruct},
2138 {"package", C_JAVA, st_C_ignore},
2139 {"short", 0, st_C_typespec},
2140 {"@end", 0, st_C_objend},
2141 {"unsigned", 0, st_C_typespec},
2143 {"const", 0, st_C_typespec},
2145 {"@interface", 0, st_C_objprot},
2146 {"enum", 0, st_C_enum},
2148 {"@implementation",0, st_C_objimpl},
2150 {"operator", C_PLPL, st_C_operator},
2151 {""}, {""}, {""}, {""},
2152 {"define", 0, st_C_define},
2154 {"double", 0, st_C_typespec},
2156 {"bool", C_PLPL, st_C_typespec},
2158 {"DEFUN", 0, st_C_gnumacro},
2159 {"float", 0, st_C_typespec}
2162 if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH)
2164 register int key = hash (str, len);
2166 if (key <= MAX_HASH_VALUE && key >= 0)
2168 register const char *s = wordlist[key].name;
2170 if (*str == *s && !strncmp (str + 1, s + 1, len - 1))
2171 return &wordlist[key];
2178 static enum sym_type
2179 C_symtype (str, len, c_ext)
2184 register struct C_stab_entry *se = in_word_set (str, len);
2186 if (se == NULL || (se->c_ext && !(c_ext & se->c_ext)))
2193 * C functions and variables are recognized using a simple
2194 * finite automaton. fvdef is its state variable.
2198 fvnone, /* nothing seen */
2199 fdefunkey, /* Emacs DEFUN keyword seen */
2200 fdefunname, /* Emacs DEFUN name seen */
2201 foperator, /* func: operator keyword seen (cplpl) */
2202 fvnameseen, /* function or variable name seen */
2203 fstartlist, /* func: just after open parenthesis */
2204 finlist, /* func: in parameter list */
2205 flistseen, /* func: after parameter list */
2206 fignore, /* func: before open brace */
2207 vignore /* var-like: ignore until ';' */
2210 bool fvextern; /* func or var: extern keyword seen; */
2213 * typedefs are recognized using a simple finite automaton.
2214 * typdef is its state variable.
2218 tnone, /* nothing seen */
2219 tkeyseen, /* typedef keyword seen */
2220 ttypeseen, /* defined type seen */
2221 tinbody, /* inside typedef body */
2222 tend, /* just before typedef tag */
2223 tignore /* junk after typedef tag */
2227 * struct-like structures (enum, struct and union) are recognized
2228 * using another simple finite automaton. `structdef' is its state
2233 snone, /* nothing seen yet,
2234 or in struct body if cblev > 0 */
2235 skeyseen, /* struct-like keyword seen */
2236 stagseen, /* struct-like tag seen */
2237 sintemplate, /* inside template (ignore) */
2238 scolonseen /* colon seen after struct-like tag */
2242 * When objdef is different from onone, objtag is the name of the class.
2244 char *objtag = "<uninited>";
2247 * Yet another little state machine to deal with preprocessor lines.
2251 dnone, /* nothing seen */
2252 dsharpseen, /* '#' seen as first char on line */
2253 ddefineseen, /* '#' and 'define' seen */
2254 dignorerest /* ignore rest of line */
2258 * State machine for Objective C protocols and implementations.
2259 * Idea by Tom R.Hageman <tom@basil.icce.rug.nl> (1995)
2263 onone, /* nothing seen */
2264 oprotocol, /* @interface or @protocol seen */
2265 oimplementation, /* @implementations seen */
2266 otagseen, /* class name seen */
2267 oparenseen, /* parenthesis before category seen */
2268 ocatseen, /* category name seen */
2269 oinbody, /* in @implementation body */
2270 omethodsign, /* in @implementation body, after +/- */
2271 omethodtag, /* after method name */
2272 omethodcolon, /* after method colon */
2273 omethodparm, /* after method parameter */
2274 oignore /* wait for @end */
2279 * Use this structure to keep info about the token read, and how it
2280 * should be tagged. Used by the make_C_tag function to build a tag.
2291 } token; /* latest token read */
2292 linebuffer token_name; /* its name */
2295 * Variables and functions for dealing with nested structures.
2296 * Idea by Mykola Dzyuba <mdzyuba@yahoo.com> (2001)
2298 static void pushclass_above P_((int, char *, int));
2299 static void popclass_above P_((int));
2300 static void write_classname P_((linebuffer *, char *qualifier));
2303 char **cname; /* nested class names */
2304 int *cblev; /* nested class curly brace level */
2305 int nl; /* class nesting level (elements used) */
2306 int size; /* length of the array */
2307 } cstack; /* stack for nested declaration tags */
2308 /* Current struct nesting depth (namespace, class, struct, union, enum). */
2309 #define nestlev (cstack.nl)
2310 /* After struct keyword or in struct body, not inside an nested function. */
2311 #define instruct (structdef == snone && nestlev > 0 \
2312 && cblev == cstack.cblev[nestlev-1] + 1)
2315 pushclass_above (cblev, str, len)
2322 popclass_above (cblev);
2324 if (nl >= cstack.size)
2326 int size = cstack.size *= 2;
2327 xrnew (cstack.cname, size, char *);
2328 xrnew (cstack.cblev, size, int);
2330 assert (nl == 0 || cstack.cblev[nl-1] < cblev);
2331 cstack.cname[nl] = (str == NULL) ? NULL : savenstr (str, len);
2332 cstack.cblev[nl] = cblev;
2337 popclass_above (cblev)
2342 for (nl = cstack.nl - 1;
2343 nl >= 0 && cstack.cblev[nl] >= cblev;
2346 if (cstack.cname[nl] != NULL)
2347 free (cstack.cname[nl]);
2353 write_classname (cn, qualifier)
2358 int qlen = strlen (qualifier);
2360 if (cstack.nl == 0 || cstack.cname[0] == NULL)
2364 cn->buffer[0] = '\0';
2368 len = strlen (cstack.cname[0]);
2369 linebuffer_setlen (cn, len);
2370 strcpy (cn->buffer, cstack.cname[0]);
2372 for (i = 1; i < cstack.nl; i++)
2377 s = cstack.cname[i];
2382 linebuffer_setlen (cn, len);
2383 strncat (cn->buffer, qualifier, qlen);
2384 strncat (cn->buffer, s, slen);
2389 static bool consider_token P_((char *, int, int, int *, int, int, bool *));
2390 static void make_C_tag P_((bool));
2394 * checks to see if the current token is at the start of a
2395 * function or variable, or corresponds to a typedef, or
2396 * is a struct/union/enum tag, or #define, or an enum constant.
2398 * *IS_FUNC gets TRUE iff the token is a function or #define macro
2399 * with args. C_EXTP points to which language we are looking at.
2410 consider_token (str, len, c, c_extp, cblev, parlev, is_func_or_var)
2411 register char *str; /* IN: token pointer */
2412 register int len; /* IN: token length */
2413 register int c; /* IN: first char after the token */
2414 int *c_extp; /* IN, OUT: C extensions mask */
2415 int cblev; /* IN: curly brace level */
2416 int parlev; /* IN: parenthesis level */
2417 bool *is_func_or_var; /* OUT: function or variable found */
2419 /* When structdef is stagseen, scolonseen, or snone with cblev > 0,
2420 structtype is the type of the preceding struct-like keyword, and
2421 structcblev is the curly brace level where it has been seen. */
2422 static enum sym_type structtype;
2423 static int structcblev;
2424 static enum sym_type toktype;
2427 toktype = C_symtype (str, len, *c_extp);
2430 * Advance the definedef state machine.
2435 /* We're not on a preprocessor line. */
2436 if (toktype == st_C_gnumacro)
2443 if (toktype == st_C_define)
2445 definedef = ddefineseen;
2449 definedef = dignorerest;
2454 * Make a tag for any macro, unless it is a constant
2455 * and constantypedefs is FALSE.
2457 definedef = dignorerest;
2458 *is_func_or_var = (c == '(');
2459 if (!*is_func_or_var && !constantypedefs)
2466 error ("internal error: definedef value.", (char *)NULL);
2475 if (toktype == st_C_typedef)
2497 if (structdef == snone && fvdef == fvnone)
2516 * This structdef business is NOT invoked when we are ctags and the
2517 * file is plain C. This is because a struct tag may have the same
2518 * name as another tag, and this loses with ctags.
2522 case st_C_javastruct:
2523 if (structdef == stagseen)
2524 structdef = scolonseen;
2528 && (*c_extp & C_AUTO) /* automatic detection of C++ language */
2529 && definedef == dnone && structdef == snone
2530 && typdef == tnone && fvdef == fvnone)
2531 *c_extp = (*c_extp | C_PLPL) & ~C_AUTO;
2537 && (typdef == tkeyseen
2538 || (typedefs_or_cplusplus && structdef == snone)))
2540 structdef = skeyseen;
2541 structtype = toktype;
2542 structcblev = cblev;
2547 if (structdef == skeyseen)
2549 structdef = stagseen;
2553 if (typdef != tnone)
2556 /* Detect Objective C constructs. */
2566 objdef = oimplementation;
2570 case oimplementation:
2571 /* Save the class tag for functions or variables defined inside. */
2572 objtag = savenstr (str, len);
2576 /* Save the class tag for categories. */
2577 objtag = savenstr (str, len);
2579 *is_func_or_var = TRUE;
2583 *is_func_or_var = TRUE;
2590 objdef = omethodtag;
2591 linebuffer_setlen (&token_name, len);
2592 strncpy (token_name.buffer, str, len);
2593 token_name.buffer[len] = '\0';
2599 objdef = omethodparm;
2604 objdef = omethodtag;
2605 linebuffer_setlen (&token_name, token_name.len + len);
2606 strncat (token_name.buffer, str, len);
2611 if (toktype == st_C_objend)
2613 /* Memory leakage here: the string pointed by objtag is
2614 never released, because many tests would be needed to
2615 avoid breaking on incorrect input code. The amount of
2616 memory leaked here is the sum of the lengths of the
2624 /* A function, variable or enum constant? */
2631 if (fvdef != finlist && fvdef != fignore && fvdef != vignore)
2632 fvdef = fvnone; /* should be useless */
2640 *is_func_or_var = TRUE;
2644 && structdef == snone
2645 && structtype == st_C_enum && cblev > structcblev)
2646 return TRUE; /* enum constant */
2652 fvdef = fdefunname; /* GNU macro */
2653 *is_func_or_var = TRUE;
2656 if ((strneq (str, "asm", 3) && endtoken (str[3]))
2657 || (strneq (str, "__asm__", 7) && endtoken (str[7])))
2662 if ((*c_extp & C_PLPL) && strneq (str+len-10, "::operator", 10))
2665 *is_func_or_var = TRUE;
2668 if (cblev > 0 && !instruct)
2670 fvdef = fvnameseen; /* function or variable */
2671 *is_func_or_var = TRUE;
2682 * C_entries often keeps pointers to tokens or lines which are older than
2683 * the line currently read. By keeping two line buffers, and switching
2684 * them at end of line, it is possible to use those pointers.
2692 #define current_lb_is_new (newndx == curndx)
2693 #define switch_line_buffers() (curndx = 1 - curndx)
2695 #define curlb (lbs[curndx].lb)
2696 #define newlb (lbs[newndx].lb)
2697 #define curlinepos (lbs[curndx].linepos)
2698 #define newlinepos (lbs[newndx].linepos)
2700 #define CNL_SAVE_DEFINEDEF() \
2702 curlinepos = charno; \
2704 linecharno = charno; \
2705 charno += readline (&curlb, inf); \
2706 lp = curlb.buffer; \
2713 CNL_SAVE_DEFINEDEF(); \
2714 if (savetoken.valid) \
2716 token = savetoken; \
2717 savetoken.valid = FALSE; \
2719 definedef = dnone; \
2727 /* This function should never be called when token.valid is FALSE, but
2728 we must protect against invalid input or internal errors. */
2729 if (DEBUG || token.valid)
2731 if (traditional_tag_style)
2733 /* This was the original code. Now we call new_pfnote instead,
2734 which uses the new method for naming tags (see new_pfnote). */
2737 if (CTAGS || token.named)
2738 name = savestr (token_name.buffer);
2739 if (DEBUG && !token.valid)
2742 name = concat (name, "##invalid##", "");
2744 name = savestr ("##invalid##");
2746 pfnote (name, isfun, token.line,
2747 token.offset+token.length+1, token.lineno, token.linepos);
2750 new_pfnote (token_name.buffer, token_name.len, isfun, token.line,
2751 token.offset+token.length+1, token.lineno, token.linepos);
2752 token.valid = FALSE;
2759 * This routine finds functions, variables, typedefs,
2760 * #define's, enum constants and struct/union/enum definitions in
2761 * C syntax and adds them to the list.
2764 C_entries (c_ext, inf)
2765 int c_ext; /* extension of C */
2766 FILE *inf; /* input file */
2768 register char c; /* latest char read; '\0' for end of line */
2769 register char *lp; /* pointer one beyond the character `c' */
2770 int curndx, newndx; /* indices for current and new lb */
2771 register int tokoff; /* offset in line of start of current token */
2772 register int toklen; /* length of current token */
2773 char *qualifier; /* string used to qualify names */
2774 int qlen; /* length of qualifier */
2775 int cblev; /* current curly brace level */
2776 int parlev; /* current parenthesis level */
2777 int typdefcblev; /* cblev where a typedef struct body begun */
2778 bool incomm, inquote, inchar, quotednl, midtoken;
2780 bool yacc_rules; /* in the rules part of a yacc file */
2781 struct tok savetoken; /* token saved during preprocessor handling */
2784 initbuffer (&token_name);
2785 initbuffer (&lbs[0].lb);
2786 initbuffer (&lbs[1].lb);
2787 if (cstack.size == 0)
2789 cstack.size = (DEBUG) ? 1 : 4;
2791 cstack.cname = xnew (cstack.size, char *);
2792 cstack.cblev = xnew (cstack.size, int);
2795 tokoff = toklen = 0; /* keep compiler quiet */
2796 curndx = newndx = 0;
2802 fvdef = fvnone; fvextern = FALSE; typdef = tnone;
2803 structdef = snone; definedef = dnone; objdef = onone;
2805 midtoken = inquote = inchar = incomm = quotednl = FALSE;
2806 token.valid = savetoken.valid = FALSE;
2810 cplpl = (c_ext & C_PLPL) == C_PLPL;
2811 cjava = (c_ext & C_JAVA) == C_JAVA;
2813 { qualifier = "."; qlen = 1; }
2815 { qualifier = "::"; qlen = 2; }
2823 /* If we're at the end of the line, the next character is a
2824 '\0'; don't skip it, because it's the thing that tells us
2825 to read the next line. */
2846 /* Newlines inside comments do not end macro definitions in
2848 CNL_SAVE_DEFINEDEF ();
2861 /* Newlines inside strings do not end macro definitions
2862 in traditional cpp, even though compilers don't
2863 usually accept them. */
2864 CNL_SAVE_DEFINEDEF ();
2874 /* Hmmm, something went wrong. */
2903 if (fvdef != finlist && fvdef != fignore && fvdef !=vignore)
2916 else if (/* cplpl && */ *lp == '/')
2924 if ((c_ext & YACC) && *lp == '%')
2926 /* Entering or exiting rules section in yacc file. */
2928 definedef = dnone; fvdef = fvnone; fvextern = FALSE;
2929 typdef = tnone; structdef = snone;
2930 midtoken = inquote = inchar = incomm = quotednl = FALSE;
2932 yacc_rules = !yacc_rules;
2938 if (definedef == dnone)
2941 bool cpptoken = TRUE;
2943 /* Look back on this line. If all blanks, or nonblanks
2944 followed by an end of comment, this is a preprocessor
2946 for (cp = newlb.buffer; cp < lp-1; cp++)
2949 if (*cp == '*' && *(cp+1) == '/')
2958 definedef = dsharpseen;
2959 } /* if (definedef == dnone) */
2965 /* Consider token only if some involved conditions are satisfied. */
2966 if (typdef != tignore
2967 && definedef != dignorerest
2969 && structdef != sintemplate
2970 && (definedef != dnone
2971 || structdef != scolonseen))
2977 if (c == ':' && cplpl && *lp == ':' && begtoken (lp[1]))
2980 * This handles :: in the middle, but not at the
2981 * beginning of an identifier. Also, space-separated
2982 * :: is not recognised.
2987 goto still_in_token;
2991 bool funorvar = FALSE;
2994 || consider_token (newlb.buffer + tokoff, toklen, c,
2995 &c_ext, cblev, parlev, &funorvar))
2997 if (fvdef == foperator)
3000 lp = skip_spaces (lp-1);
3004 && !iswhite (*lp) && *lp != '(')
3007 toklen += lp - oldlp;
3009 token.named = FALSE;
3010 if ((c_ext & C_EXT) /* not pure C */
3011 && nestlev > 0 && definedef == dnone)
3012 /* in struct body */
3014 write_classname (&token_name, qualifier);
3015 linebuffer_setlen (&token_name,
3016 token_name.len+qlen+toklen);
3017 strcat (token_name.buffer, qualifier);
3018 strncat (token_name.buffer,
3019 newlb.buffer + tokoff, toklen);
3022 else if (objdef == ocatseen)
3023 /* Objective C category */
3025 int len = strlen (objtag) + 2 + toklen;
3026 linebuffer_setlen (&token_name, len);
3027 strcpy (token_name.buffer, objtag);
3028 strcat (token_name.buffer, "(");
3029 strncat (token_name.buffer,
3030 newlb.buffer + tokoff, toklen);
3031 strcat (token_name.buffer, ")");
3034 else if (objdef == omethodtag
3035 || objdef == omethodparm)
3036 /* Objective C method */
3040 else if (fvdef == fdefunname)
3041 /* GNU DEFUN and similar macros */
3043 bool defun = (newlb.buffer[tokoff] == 'F');
3047 /* Rewrite the tag so that emacs lisp DEFUNs
3048 can be found by their elisp name */
3055 linebuffer_setlen (&token_name, len);
3056 strncpy (token_name.buffer,
3057 newlb.buffer + off, len);
3058 token_name.buffer[len] = '\0';
3061 if (token_name.buffer[len] == '_')
3062 token_name.buffer[len] = '-';
3063 token.named = defun;
3067 linebuffer_setlen (&token_name, toklen);
3068 strncpy (token_name.buffer,
3069 newlb.buffer + tokoff, toklen);
3070 token_name.buffer[toklen] = '\0';
3071 /* Name macros and members. */
3072 token.named = (structdef == stagseen
3073 || typdef == ttypeseen
3076 && definedef == dignorerest)
3078 && definedef == dnone
3079 && structdef == snone
3082 token.lineno = lineno;
3083 token.offset = tokoff;
3084 token.length = toklen;
3085 token.line = newlb.buffer;
3086 token.linepos = newlinepos;
3089 if (definedef == dnone
3090 && (fvdef == fvnameseen
3091 || fvdef == foperator
3092 || structdef == stagseen
3094 || typdef == ttypeseen
3095 || objdef != onone))
3097 if (current_lb_is_new)
3098 switch_line_buffers ();
3100 else if (definedef != dnone
3101 || fvdef == fdefunname
3103 make_C_tag (funorvar);
3107 } /* if (endtoken (c)) */
3108 else if (intoken (c))
3114 } /* if (midtoken) */
3115 else if (begtoken (c))
3126 make_C_tag (TRUE); /* a function */
3133 if (structdef == stagseen && !cjava)
3135 popclass_above (cblev);
3142 if (!yacc_rules || lp == newlb.buffer + 1)
3144 tokoff = lp - 1 - newlb.buffer;
3149 } /* if (begtoken) */
3150 } /* if must look at token */
3153 /* Detect end of line, colon, comma, semicolon and various braces
3154 after having handled a token.*/
3158 if (yacc_rules && token.offset == 0 && token.valid)
3160 make_C_tag (FALSE); /* a yacc function */
3163 if (definedef != dnone)
3169 make_C_tag (TRUE); /* an Objective C class */
3173 objdef = omethodcolon;
3174 linebuffer_setlen (&token_name, token_name.len + 1);
3175 strcat (token_name.buffer, ":");
3178 if (structdef == stagseen)
3179 structdef = scolonseen;
3182 if (definedef != dnone)
3188 make_C_tag (FALSE); /* a typedef */
3198 if (typdef == tignore)
3202 if ((globals && cblev == 0 && (!fvextern || declarations))
3203 || (members && instruct))
3204 make_C_tag (FALSE); /* a variable */
3207 token.valid = FALSE;
3210 if ((declarations && typdef == tnone && !instruct)
3211 || (members && typdef != tignore && instruct))
3212 make_C_tag (TRUE); /* a function declaration */
3218 && structdef == stagseen && (c_ext & C_PLPL))
3219 make_C_tag (FALSE); /* forward declaration */
3221 /* The following instruction invalidates the token.
3222 Probably the token should be invalidated in all other
3223 cases where some state machine is reset prematurely. */
3224 token.valid = FALSE;
3225 } /* switch (fvdef) */
3231 if (structdef == stagseen)
3235 if (definedef != dnone)
3241 make_C_tag (TRUE); /* an Objective C method */
3257 case fvnameseen: /* a variable */
3258 if ((globals && cblev == 0 && (!fvextern || declarations))
3259 || (members && instruct))
3262 case flistseen: /* a function */
3263 if ((declarations && typdef == tnone && !instruct)
3264 || (members && typdef != tignore && instruct))
3266 make_C_tag (TRUE); /* a function declaration */
3269 else if (!declarations)
3271 token.valid = FALSE;
3276 if (structdef == stagseen)
3280 if (definedef != dnone)
3282 if (structdef == stagseen)
3289 make_C_tag (FALSE); /* a typedef */
3301 if ((members && cblev == 1)
3302 || (globals && cblev == 0
3303 && (!fvextern || declarations)))
3304 make_C_tag (FALSE); /* a variable */
3313 if (definedef != dnone)
3315 if (objdef == otagseen && parlev == 0)
3316 objdef = oparenseen;
3320 if (typdef == ttypeseen
3324 /* This handles constructs like:
3325 typedef void OperatorFun (int fun); */
3342 if (definedef != dnone)
3344 if (objdef == ocatseen && parlev == 1)
3346 make_C_tag (TRUE); /* an Objective C category */
3360 || typdef == ttypeseen))
3363 make_C_tag (FALSE); /* a typedef */
3366 else if (parlev < 0) /* can happen due to ill-conceived #if's. */
3370 if (definedef != dnone)
3372 if (typdef == ttypeseen)
3374 typdefcblev = cblev;
3380 make_C_tag (TRUE); /* a function */
3389 make_C_tag (TRUE); /* an Objective C class */
3394 make_C_tag (TRUE); /* an Objective C method */
3398 /* Neutralize `extern "C" {' grot. */
3399 if (cblev == 0 && structdef == snone && nestlev == 0
3406 case skeyseen: /* unnamed struct */
3407 pushclass_above (cblev, NULL, 0);
3410 case stagseen: /* named struct or enum */
3411 case scolonseen: /* a class */
3412 pushclass_above (cblev, token.line+token.offset, token.length);
3414 make_C_tag (FALSE); /* a struct or enum */
3420 if (definedef != dnone)
3422 if (fvdef == fstartlist)
3423 fvdef = fvnone; /* avoid tagging `foo' in `foo (*bar()) ()' */
3426 if (definedef != dnone)
3428 if (!noindentypedefs && lp == newlb.buffer + 1)
3430 cblev = 0; /* reset curly brace level if first column */
3431 parlev = 0; /* also reset paren level, just in case... */
3435 popclass_above (cblev);
3437 if (typdef == tinbody && cblev <= typdefcblev)
3439 assert (cblev == typdefcblev);
3444 if (definedef != dnone)
3454 if ((members && cblev == 1)
3455 || (globals && cblev == 0 && (!fvextern || declarations)))
3456 make_C_tag (FALSE); /* a variable */
3463 if (cplpl && structdef == stagseen)
3465 structdef = sintemplate;
3470 if (structdef == sintemplate)
3472 structdef = stagseen;
3478 if (objdef == oinbody && cblev == 0)
3480 objdef = omethodsign;
3485 case '#': case '~': case '&': case '%': case '/': case '|':
3486 case '^': case '!': case '.': case '?': case ']':
3487 if (definedef != dnone)
3489 /* These surely cannot follow a function tag in C. */
3502 if (objdef == otagseen)
3504 make_C_tag (TRUE); /* an Objective C class */
3507 /* If a macro spans multiple lines don't reset its state. */
3509 CNL_SAVE_DEFINEDEF ();
3515 } /* while not eof */
3517 free (token_name.buffer);
3518 free (lbs[0].lb.buffer);
3519 free (lbs[1].lb.buffer);
3523 * Process either a C++ file or a C file depending on the setting
3527 default_C_entries (inf)
3530 C_entries (cplusplus ? C_PLPL : C_AUTO, inf);
3533 /* Always do plain C. */
3535 plain_C_entries (inf)
3541 /* Always do C++. */
3543 Cplusplus_entries (inf)
3546 C_entries (C_PLPL, inf);
3549 /* Always do Java. */
3554 C_entries (C_JAVA, inf);
3562 C_entries (C_STAR, inf);
3565 /* Always do Yacc. */
3570 C_entries (YACC, inf);
3574 /* A useful macro. */
3575 #define LOOP_ON_INPUT_LINES(file_pointer, line_buffer, char_pointer) \
3576 for (lineno = charno = 0; /* loop initialization */ \
3577 !feof (file_pointer) /* loop test */ \
3578 && (lineno++, /* instructions at start of loop */ \
3579 linecharno = charno, \
3580 charno += readline (&line_buffer, file_pointer), \
3581 char_pointer = lb.buffer, \
3587 * Read a file, but do no processing. This is used to do regexp
3588 * matching on files that have no language defined.
3591 just_read_file (inf)
3594 register char *dummy;
3596 LOOP_ON_INPUT_LINES (inf, lb, dummy)
3601 /* Fortran parsing */
3603 static bool tail P_((char *));
3604 static void takeprec P_((void));
3605 static void getit P_((FILE *));
3611 register int len = 0;
3613 while (*cp != '\0' && lowcase (*cp) == lowcase (dbp[len]))
3615 if (*cp == '\0' && !intoken (dbp[len]))
3626 dbp = skip_spaces (dbp);
3630 dbp = skip_spaces (dbp);
3631 if (strneq (dbp, "(*)", 3))
3636 if (!ISDIGIT (*dbp))
3638 --dbp; /* force failure */
3643 while (ISDIGIT (*dbp));
3652 dbp = skip_spaces (dbp);
3656 linecharno = charno;
3657 charno += readline (&lb, inf);
3662 dbp = skip_spaces (dbp);
3664 if (!ISALPHA (*dbp) && *dbp != '_' && *dbp != '$')
3666 for (cp = dbp + 1; *cp != '\0' && intoken (*cp); cp++)
3668 pfnote (savenstr (dbp, cp-dbp), TRUE,
3669 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
3674 Fortran_functions (inf)
3677 LOOP_ON_INPUT_LINES (inf, lb, dbp)
3680 dbp++; /* Ratfor escape to fortran */
3681 dbp = skip_spaces (dbp);
3684 switch (lowcase (*dbp))
3687 if (tail ("integer"))
3695 if (tail ("logical"))
3699 if (tail ("complex") || tail ("character"))
3703 if (tail ("double"))
3705 dbp = skip_spaces (dbp);
3708 if (tail ("precision"))
3714 dbp = skip_spaces (dbp);
3717 switch (lowcase (*dbp))
3720 if (tail ("function"))
3724 if (tail ("subroutine"))
3732 if (tail ("blockdata") || tail ("block data"))
3734 dbp = skip_spaces (dbp);
3735 if (*dbp == '\0') /* assume un-named */
3736 pfnote (savestr ("blockdata"), TRUE,
3737 lb.buffer, dbp - lb.buffer, lineno, linecharno);
3739 getit (inf); /* look for name */
3749 * Philippe Waroquiers <philippe.waroquiers@eurocontrol.be> (1998)
3752 static void adagetit P_((FILE *, char *));
3754 /* Once we are positioned after an "interesting" keyword, let's get
3755 the real tag value necessary. */
3757 adagetit (inf, name_qualifier)
3759 char *name_qualifier;
3767 dbp = skip_spaces (dbp);
3769 || (dbp[0] == '-' && dbp[1] == '-'))
3772 linecharno = charno;
3773 charno += readline (&lb, inf);
3782 /* Skipping body of procedure body or package body or ....
3783 resetting qualifier to body instead of spec. */
3784 name_qualifier = "/b";
3790 /* Skipping type of task type or protected type ... */
3798 for (cp = dbp; *cp != '\0' && *cp != '"'; cp++)
3803 dbp = skip_spaces (dbp);
3806 && (ISALPHA (*cp) || ISDIGIT (*cp) || *cp == '_' || *cp == '.'));
3814 name = concat (dbp, name_qualifier, "");
3816 pfnote (name, TRUE, lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
3827 bool inquote = FALSE;
3829 LOOP_ON_INPUT_LINES (inf, lb, dbp)
3831 while (*dbp != '\0')
3833 /* Skip a string i.e. "abcd". */
3834 if (inquote || (*dbp == '"'))
3836 dbp = etags_strchr ((inquote) ? dbp : dbp+1, '"');
3841 continue; /* advance char */
3846 break; /* advance line */
3850 /* Skip comments. */
3851 if (dbp[0] == '-' && dbp[1] == '-')
3852 break; /* advance line */
3854 /* Skip character enclosed in single quote i.e. 'a'
3855 and skip single quote starting an attribute i.e. 'Image. */
3864 /* Search for beginning of a token. */
3865 if (!begtoken (*dbp))
3868 continue; /* advance char */
3871 /* We are at the beginning of a token. */
3876 if (!packages_only && tail ("function"))
3877 adagetit (inf, "/f");
3879 break; /* from switch */
3880 continue; /* advance char */
3883 if (!packages_only && tail ("procedure"))
3884 adagetit (inf, "/p");
3885 else if (tail ("package"))
3886 adagetit (inf, "/s");
3887 else if (tail ("protected")) /* protected type */
3888 adagetit (inf, "/t");
3890 break; /* from switch */
3891 continue; /* advance char */
3894 if (!packages_only && tail ("task"))
3895 adagetit (inf, "/k");
3896 else if (typedefs && !packages_only && tail ("type"))
3898 adagetit (inf, "/t");
3899 while (*dbp != '\0')
3903 break; /* from switch */
3904 continue; /* advance char */
3907 /* Look for the end of the token. */
3908 while (!endtoken (*dbp))
3911 } /* advance char */
3912 } /* advance line */
3917 * Bob Weiner, Motorola Inc., 4/3/94
3918 * Unix and microcontroller assembly tag handling
3919 * look for '^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]'
3927 LOOP_ON_INPUT_LINES (inf, lb, cp)
3929 /* If first char is alphabetic or one of [_.$], test for colon
3930 following identifier. */
3931 if (ISALPHA (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
3933 /* Read past label. */
3935 while (ISALNUM (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
3937 if (*cp == ':' || iswhite (*cp))
3939 /* Found end of label, so copy it and add it to the table. */
3940 pfnote (savenstr(lb.buffer, cp-lb.buffer), TRUE,
3941 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
3950 * Perl sub names: look for /^sub[ \t\n]+[^ \t\n{]+/
3951 * Perl variable names: /^(my|local).../
3952 * Bart Robinson <lomew@cs.utah.edu> (1995)
3953 * Michael Ernst <mernst@alum.mit.edu> (1997)
3956 Perl_functions (inf)
3961 LOOP_ON_INPUT_LINES (inf, lb, cp)
3965 && *cp++ == 'b' && iswhite (*cp++))
3967 cp = skip_spaces (cp);
3972 && !iswhite (*cp) && *cp != '{' && *cp != '(')
3974 pfnote (savenstr (sp, cp-sp), TRUE,
3975 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
3978 else if (globals /* only if tagging global vars is enabled */
3979 && ((cp = lb.buffer,
3988 && (*cp == '(' || iswhite (*cp)))
3990 /* After "my" or "local", but before any following paren or space. */
3991 char *varname = NULL;
3993 cp = skip_spaces (cp);
3994 if (*cp == '$' || *cp == '@' || *cp == '%')
3996 char* varstart = ++cp;
3997 while (ISALNUM (*cp) || *cp == '_')
3999 varname = savenstr (varstart, cp-varstart);
4003 /* Should be examining a variable list at this point;
4004 could insist on seeing an open parenthesis. */
4005 while (*cp != '\0' && *cp != ';' && *cp != '=' && *cp != ')')
4009 /* Perhaps I should back cp up one character, so the TAGS table
4010 doesn't mention (and so depend upon) the following char. */
4011 pfnote ((CTAGS) ? savenstr (lb.buffer, cp-lb.buffer) : varname,
4012 FALSE, lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4020 * Look for /^[ \t\n]*def[ \t\n]+[^ \t\n(:]+/ or /^class[ \t\n]+[^ \t\n(:]+/
4021 * Eric S. Raymond <esr@thyrsus.com> (1997)
4022 * Sean Legassick <sean@informage.net> (2004)
4025 Python_functions (inf)
4030 LOOP_ON_INPUT_LINES (inf, lb, cp)
4032 cp = skip_spaces (cp);
4035 && *cp++ == 'f' && iswhite (*cp++))
4037 cp = skip_spaces (cp);
4038 while (*cp != '\0' && !iswhite (*cp) && *cp != '(' && *cp != ':')
4041 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4049 && *cp++ == 's' && iswhite (*cp++))
4051 cp = skip_spaces (cp);
4052 while (*cp != '\0' && !iswhite (*cp) && *cp != '(' && *cp != ':')
4055 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4061 /* Idea by Corny de Souza
4062 * Cobol tag functions
4063 * We could look for anything that could be a paragraph name.
4064 * i.e. anything that starts in column 8 is one word and ends in a full stop.
4067 Cobol_paragraphs (inf)
4070 register char *bp, *ep;
4072 LOOP_ON_INPUT_LINES (inf, lb, bp)
4078 /* If eoln, compiler option or comment ignore whole line. */
4079 if (bp[-1] != ' ' || !ISALNUM (bp[0]))
4082 for (ep = bp; ISALNUM (*ep) || *ep == '-'; ep++)
4085 pfnote (savenstr (bp, ep-bp), TRUE,
4086 lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
4093 * Idea by Assar Westerlund <assar@sics.se> (2001)
4096 Makefile_targets (inf)
4101 LOOP_ON_INPUT_LINES (inf, lb, bp)
4103 if (*bp == '\t' || *bp == '#')
4105 while (*bp != '\0' && *bp != '=' && *bp != ':')
4108 pfnote (savenstr (lb.buffer, bp - lb.buffer), TRUE,
4109 lb.buffer, bp - lb.buffer + 1, lineno, linecharno);
4114 /* Added by Mosur Mohan, 4/22/88 */
4115 /* Pascal parsing */
4118 * Locates tags for procedures & functions. Doesn't do any type- or
4119 * var-definitions. It does look for the keyword "extern" or
4120 * "forward" immediately following the procedure statement; if found,
4121 * the tag is skipped.
4124 Pascal_functions (inf)
4127 linebuffer tline; /* mostly copied from C_entries */
4129 int save_lineno, save_len;
4130 char c, *cp, *namebuf;
4132 bool /* each of these flags is TRUE iff: */
4133 incomment, /* point is inside a comment */
4134 inquote, /* point is inside '..' string */
4135 get_tagname, /* point is after PROCEDURE/FUNCTION
4136 keyword, so next item = potential tag */
4137 found_tag, /* point is after a potential tag */
4138 inparms, /* point is within parameter-list */
4139 verify_tag; /* point has passed the parm-list, so the
4140 next token will determine whether this
4141 is a FORWARD/EXTERN to be ignored, or
4142 whether it is a real tag */
4144 save_lcno = save_lineno = save_len = 0; /* keep compiler quiet */
4145 namebuf = NULL; /* keep compiler quiet */
4150 initbuffer (&tline);
4152 incomment = inquote = FALSE;
4153 found_tag = FALSE; /* have a proc name; check if extern */
4154 get_tagname = FALSE; /* have found "procedure" keyword */
4155 inparms = FALSE; /* found '(' after "proc" */
4156 verify_tag = FALSE; /* check if "extern" is ahead */
4159 while (!feof (inf)) /* long main loop to get next char */
4162 if (c == '\0') /* if end of line */
4165 linecharno = charno;
4166 charno += readline (&lb, inf);
4170 if (!((found_tag && verify_tag)
4172 c = *dbp++; /* only if don't need *dbp pointing
4173 to the beginning of the name of
4174 the procedure or function */
4178 if (c == '}') /* within { } comments */
4180 else if (c == '*' && *dbp == ')') /* within (* *) comments */
4197 inquote = TRUE; /* found first quote */
4199 case '{': /* found open { comment */
4203 if (*dbp == '*') /* found open (* comment */
4208 else if (found_tag) /* found '(' after tag, i.e., parm-list */
4211 case ')': /* end of parms list */
4216 if (found_tag && !inparms) /* end of proc or fn stmt */
4223 if (found_tag && verify_tag && (*dbp != ' '))
4225 /* check if this is an "extern" declaration */
4228 if (lowcase (*dbp == 'e'))
4230 if (tail ("extern")) /* superfluous, really! */
4236 else if (lowcase (*dbp) == 'f')
4238 if (tail ("forward")) /* check for forward reference */
4244 if (found_tag && verify_tag) /* not external proc, so make tag */
4248 pfnote (namebuf, TRUE,
4249 tline.buffer, save_len, save_lineno, save_lcno);
4253 if (get_tagname) /* grab name of proc or fn */
4258 /* save all values for later tagging */
4259 linebuffer_setlen (&tline, lb.len);
4260 strcpy (tline.buffer, lb.buffer);
4261 save_lineno = lineno;
4262 save_lcno = linecharno;
4264 /* grab block name */
4265 for (cp = dbp + 1; *cp != '\0' && !endtoken (*cp); cp++)
4267 namebuf = savenstr (dbp, cp-dbp);
4268 dbp = cp; /* set dbp to e-o-token */
4269 save_len = dbp - lb.buffer + 1;
4270 get_tagname = FALSE;
4274 /* and proceed to check for "extern" */
4276 else if (!incomment && !inquote && !found_tag)
4278 /* check for proc/fn keywords */
4279 switch (lowcase (c))
4282 if (tail ("rocedure")) /* c = 'p', dbp has advanced */
4286 if (tail ("unction"))
4291 } /* while not eof */
4293 free (tline.buffer);
4298 * Lisp tag functions
4299 * look for (def or (DEF, quote or QUOTE
4302 static int L_isdef P_((char *));
4303 static int L_isquote P_((char *));
4304 static void L_getit P_((void));
4308 register char *strp;
4310 return ((strp[1] == 'd' || strp[1] == 'D')
4311 && (strp[2] == 'e' || strp[2] == 'E')
4312 && (strp[3] == 'f' || strp[3] == 'F'));
4317 register char *strp;
4319 return ((*++strp == 'q' || *strp == 'Q')
4320 && (*++strp == 'u' || *strp == 'U')
4321 && (*++strp == 'o' || *strp == 'O')
4322 && (*++strp == 't' || *strp == 'T')
4323 && (*++strp == 'e' || *strp == 'E')
4324 && iswhite (*++strp));
4332 if (*dbp == '\'') /* Skip prefix quote */
4334 else if (*dbp == '(')
4336 if (L_isquote (dbp))
4337 dbp += 7; /* Skip "(quote " */
4339 dbp += 1; /* Skip "(" before name in (defstruct (foo)) */
4340 dbp = skip_spaces (dbp);
4343 for (cp = dbp /*+1*/;
4344 *cp != '\0' && *cp != '(' && !iswhite(*cp) && *cp != ')';
4350 pfnote (savenstr (dbp, cp-dbp), TRUE,
4351 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4355 Lisp_functions (inf)
4358 LOOP_ON_INPUT_LINES (inf, lb, dbp)
4364 dbp = skip_non_spaces (dbp);
4365 dbp = skip_spaces (dbp);
4370 /* Check for (foo::defmumble name-defined ... */
4373 while (*dbp != '\0' && !iswhite (*dbp)
4374 && *dbp != ':' && *dbp != '(' && *dbp != ')');
4379 while (*dbp == ':');
4381 if (L_isdef (dbp - 1))
4383 dbp = skip_non_spaces (dbp);
4384 dbp = skip_spaces (dbp);
4395 * Postscript tag functions
4396 * Just look for lines where the first character is '/'
4397 * Also look at "defineps" for PSWrap
4398 * Richard Mlynarik <mly@adoc.xerox.com> (1997)
4399 * Ideas by Masatake Yamato <masata-y@is.aist-nara.ac.jp> (1999)
4402 Postscript_functions (inf)
4405 register char *bp, *ep;
4407 LOOP_ON_INPUT_LINES (inf, lb, bp)
4412 *ep != '\0' && *ep != ' ' && *ep != '{';
4415 pfnote (savenstr (bp, ep-bp), TRUE,
4416 lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
4418 else if (strneq (bp, "defineps", 8))
4420 bp = skip_non_spaces (bp);
4421 bp = skip_spaces (bp);
4429 * Scheme tag functions
4430 * look for (def... xyzzy
4431 * look for (def... (xyzzy
4432 * look for (def ... ((...(xyzzy ....
4433 * look for (set! xyzzy
4437 Scheme_functions (inf)
4442 LOOP_ON_INPUT_LINES (inf, lb, bp)
4445 && (bp[1] == 'D' || bp[1] == 'd')
4446 && (bp[2] == 'E' || bp[2] == 'e')
4447 && (bp[3] == 'F' || bp[3] == 'f'))
4449 bp = skip_non_spaces (bp);
4450 /* Skip over open parens and white space */
4451 while (iswhite (*bp) || *bp == '(')
4456 && (bp[1] == 'S' || bp[1] == 's')
4457 && (bp[2] == 'E' || bp[2] == 'e')
4458 && (bp[3] == 'T' || bp[3] == 't')
4459 && (bp[4] == '!' || bp[4] == '!')
4460 && (iswhite (bp[5])))
4462 bp = skip_non_spaces (bp);
4463 bp = skip_spaces (bp);
4470 /* Find tags in TeX and LaTeX input files. */
4472 /* TEX_toktab is a table of TeX control sequences that define tags.
4473 Each TEX_tabent records one such control sequence.
4474 CONVERT THIS TO USE THE Stab TYPE!! */
4481 struct TEX_tabent *TEX_toktab = NULL; /* Table with tag tokens */
4483 /* Default set of control sequences to put into TEX_toktab.
4484 The value of environment var TEXTAGS is prepended to this. */
4486 char *TEX_defenv = "\
4487 :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem\
4488 :part:appendix:entry:index";
4490 static void TEX_mode P_((FILE *));
4491 static struct TEX_tabent *TEX_decode_env P_((char *, char *));
4492 static int TEX_Token P_((char *));
4494 char TEX_esc = '\\';
4495 char TEX_opgrp = '{';
4496 char TEX_clgrp = '}';
4499 * TeX/LaTeX scanning loop.
4508 /* Select either \ or ! as escape character. */
4511 /* Initialize token table once from environment. */
4513 TEX_toktab = TEX_decode_env ("TEXTAGS", TEX_defenv);
4515 LOOP_ON_INPUT_LINES (inf, lb, cp)
4518 /* Look at each esc in line. */
4519 while ((cp = etags_strchr (cp, TEX_esc)) != NULL)
4523 linecharno += cp - lasthit;
4525 i = TEX_Token (lasthit);
4528 /* We seem to include the TeX command in the tag name.
4530 for (p = lasthit + TEX_toktab[i].len;
4531 *p != '\0' && *p != TEX_clgrp;
4534 pfnote (/*savenstr (lasthit, p-lasthit)*/ (char *)NULL, TRUE,
4535 lb.buffer, lb.len, lineno, linecharno);
4536 break; /* We only tag a line once */
4542 #define TEX_LESC '\\'
4543 #define TEX_SESC '!'
4546 /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
4547 chars accordingly. */
4554 while ((c = getc (inf)) != EOF)
4556 /* Skip to next line if we hit the TeX comment char. */
4560 else if (c == TEX_LESC || c == TEX_SESC )
4576 /* If the input file is compressed, inf is a pipe, and rewind may fail.
4577 No attempt is made to correct the situation. */
4581 /* Read environment and prepend it to the default string.
4582 Build token table. */
4583 static struct TEX_tabent *
4584 TEX_decode_env (evarname, defenv)
4588 register char *env, *p;
4590 struct TEX_tabent *tab;
4593 /* Append default string to environment. */
4594 env = getenv (evarname);
4600 env = concat (oldenv, defenv, "");
4603 /* Allocate a token table */
4604 for (size = 1, p = env; p;)
4605 if ((p = etags_strchr (p, ':')) && *++p != '\0')
4607 /* Add 1 to leave room for null terminator. */
4608 tab = xnew (size + 1, struct TEX_tabent);
4610 /* Unpack environment string into token table. Be careful about */
4611 /* zero-length strings (leading ':', "::" and trailing ':') */
4614 p = etags_strchr (env, ':');
4615 if (!p) /* End of environment string. */
4616 p = env + strlen (env);
4618 { /* Only non-zero strings. */
4619 tab[i].name = savenstr (env, p - env);
4620 tab[i].len = strlen (tab[i].name);
4627 tab[i].name = NULL; /* Mark end of table. */
4635 /* If the text at CP matches one of the tag-defining TeX command names,
4636 return the pointer to the first occurrence of that command in TEX_toktab.
4637 Otherwise return -1.
4638 Keep the capital `T' in `token' for dumb truncating compilers
4639 (this distinguishes it from `TEX_toktab' */
4646 for (i = 0; TEX_toktab[i].len > 0; i++)
4647 if (strneq (TEX_toktab[i].name, cp, TEX_toktab[i].len))
4653 /* Texinfo support. Dave Love, Mar. 2000. */
4659 LOOP_ON_INPUT_LINES (inf, lb, cp)
4665 && *cp++ == 'e' && iswhite (*cp++)))
4667 start = cp = skip_spaces(cp);
4668 while (*cp != '\0' && *cp != ',')
4670 pfnote (savenstr (start, cp - start), TRUE,
4671 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4678 * Prolog support (rewritten) by Anders Lindgren, Mar. 96
4680 * Assumes that the predicate starts at column 0.
4681 * Only the first clause of a predicate is added.
4683 static int prolog_pred P_((char *, char *));
4684 static void prolog_skip_comment P_((linebuffer *, FILE *));
4685 static int prolog_atom P_((char *, int));
4688 Prolog_functions (inf)
4699 LOOP_ON_INPUT_LINES (inf, lb, cp)
4701 if (cp[0] == '\0') /* Empty line */
4703 else if (iswhite (cp[0])) /* Not a predicate */
4705 else if (cp[0] == '/' && cp[1] == '*') /* comment. */
4706 prolog_skip_comment (&lb, inf);
4707 else if ((len = prolog_pred (cp, last)) > 0)
4709 /* Predicate. Store the function name so that we only
4710 generate a tag for the first clause. */
4712 last = xnew(len + 1, char);
4713 else if (len + 1 > allocated)
4714 xrnew (last, len + 1, char);
4715 allocated = len + 1;
4716 strncpy (last, cp, len);
4724 prolog_skip_comment (plb, inf)
4732 for (cp = plb->buffer; *cp != '\0'; cp++)
4733 if (cp[0] == '*' && cp[1] == '/')
4736 linecharno += readline (plb, inf);
4742 * A predicate definition is added if it matches:
4743 * <beginning of line><Prolog Atom><whitespace>(
4745 * It is added to the tags database if it doesn't match the
4746 * name of the previous clause header.
4748 * Return the size of the name of the predicate, or 0 if no header
4752 prolog_pred (s, last)
4754 char *last; /* Name of last clause. */
4759 pos = prolog_atom (s, 0);
4764 pos = skip_spaces (s + pos) - s;
4766 if ((s[pos] == '(') || (s[pos] == '.'))
4771 /* Save only the first clause. */
4773 || len != (int)strlen (last)
4774 || !strneq (s, last, len))
4776 pfnote (savenstr (s, len), TRUE, s, pos, lineno, linecharno);
4784 * Consume a Prolog atom.
4785 * Return the number of bytes consumed, or -1 if there was an error.
4787 * A prolog atom, in this context, could be one of:
4788 * - An alphanumeric sequence, starting with a lower case letter.
4789 * - A quoted arbitrary string. Single quotes can escape themselves.
4790 * Backslash quotes everything.
4793 prolog_atom (s, pos)
4801 if (ISLOWER(s[pos]) || (s[pos] == '_'))
4803 /* The atom is unquoted. */
4805 while (ISALNUM(s[pos]) || (s[pos] == '_'))
4809 return pos - origpos;
4811 else if (s[pos] == '\'')
4822 pos++; /* A double quote */
4824 else if (s[pos] == '\0')
4825 /* Multiline quoted atoms are ignored. */
4827 else if (s[pos] == '\\')
4829 if (s[pos+1] == '\0')
4836 return pos - origpos;
4844 * Support for Erlang -- Anders Lindgren, Feb 1996.
4846 * Generates tags for functions, defines, and records.
4848 * Assumes that Erlang functions start at column 0.
4850 static int erlang_func P_((char *, char *));
4851 static void erlang_attribute P_((char *));
4852 static int erlang_atom P_((char *, int));
4855 Erlang_functions (inf)
4866 LOOP_ON_INPUT_LINES (inf, lb, cp)
4868 if (cp[0] == '\0') /* Empty line */
4870 else if (iswhite (cp[0])) /* Not function nor attribute */
4872 else if (cp[0] == '%') /* comment */
4874 else if (cp[0] == '"') /* Sometimes, strings start in column one */
4876 else if (cp[0] == '-') /* attribute, e.g. "-define" */
4878 erlang_attribute (cp);
4881 else if ((len = erlang_func (cp, last)) > 0)
4884 * Function. Store the function name so that we only
4885 * generates a tag for the first clause.
4888 last = xnew (len + 1, char);
4889 else if (len + 1 > allocated)
4890 xrnew (last, len + 1, char);
4891 allocated = len + 1;
4892 strncpy (last, cp, len);
4900 * A function definition is added if it matches:
4901 * <beginning of line><Erlang Atom><whitespace>(
4903 * It is added to the tags database if it doesn't match the
4904 * name of the previous clause header.
4906 * Return the size of the name of the function, or 0 if no function
4910 erlang_func (s, last)
4912 char *last; /* Name of last clause. */
4917 pos = erlang_atom (s, 0);
4922 pos = skip_spaces (s + pos) - s;
4924 /* Save only the first clause. */
4927 || len != (int)strlen (last)
4928 || !strneq (s, last, len)))
4930 pfnote (savenstr (s, len), TRUE, s, pos, lineno, linecharno);
4939 * Handle attributes. Currently, tags are generated for defines
4942 * They are on the form:
4943 * -define(foo, bar).
4944 * -define(Foo(M, N), M+N).
4945 * -record(graph, {vtab = notable, cyclic = true}).
4948 erlang_attribute (s)
4954 if (strneq (s, "-define", 7) || strneq (s, "-record", 7))
4956 pos = skip_spaces (s + 7) - s;
4957 if (s[pos++] == '(')
4959 pos = skip_spaces (s + pos) - s;
4960 len = erlang_atom (s, pos);
4962 pfnote (savenstr (& s[pos], len), TRUE,
4963 s, pos + len, lineno, linecharno);
4971 * Consume an Erlang atom (or variable).
4972 * Return the number of bytes consumed, or -1 if there was an error.
4975 erlang_atom (s, pos)
4983 if (ISALPHA (s[pos]) || s[pos] == '_')
4985 /* The atom is unquoted. */
4987 while (ISALNUM (s[pos]) || s[pos] == '_')
4989 return pos - origpos;
4991 else if (s[pos] == '\'')
5002 else if (s[pos] == '\0')
5003 /* Multiline quoted atoms are ignored. */
5005 else if (s[pos] == '\\')
5007 if (s[pos+1] == '\0')
5014 return pos - origpos;
5021 #ifdef ETAGS_REGEXPS
5023 static char *scan_separators P_((char *));
5024 static void analyse_regex P_((char *, bool));
5025 static void add_regex P_((char *, bool, language *));
5026 static char *substitute P_((char *, char *, struct re_registers *));
5028 /* Take a string like "/blah/" and turn it into "blah", making sure
5029 that the first and last characters are the same, and handling
5030 quoted separator characters. Actually, stops on the occurrence of
5031 an unquoted separator. Also turns "\t" into a Tab character.
5032 Returns pointer to terminating separator. Works in place. Null
5033 terminates name string. */
5035 scan_separators (name)
5039 char *copyto = name;
5040 bool quoted = FALSE;
5042 for (++name; *name != '\0'; ++name)
5048 else if (*name == sep)
5052 /* Something else is quoted, so preserve the quote. */
5058 else if (*name == '\\')
5060 else if (*name == sep)
5066 /* Terminate copied string. */
5071 /* Look at the argument of --regex or --no-regex and do the right
5072 thing. Same for each line of a regexp file. */
5074 analyse_regex (regex_arg, ignore_case)
5078 if (regex_arg == NULL)
5079 free_patterns (); /* --no-regex: remove existing regexps */
5081 /* A real --regexp option or a line in a regexp file. */
5082 switch (regex_arg[0])
5084 /* Comments in regexp file or null arg to --regex. */
5090 /* Read a regex file. This is recursive and may result in a
5091 loop, which will stop when the file descriptors are exhausted. */
5095 linebuffer regexbuf;
5096 char *regexfile = regex_arg + 1;
5098 /* regexfile is a file containing regexps, one per line. */
5099 regexfp = fopen (regexfile, "r");
5100 if (regexfp == NULL)
5105 initbuffer (®exbuf);
5106 while (readline_internal (®exbuf, regexfp) > 0)
5107 analyse_regex (regexbuf.buffer, ignore_case);
5108 free (regexbuf.buffer);
5113 /* Regexp to be used for a specific language only. */
5117 char *lang_name = regex_arg + 1;
5120 for (cp = lang_name; *cp != '}'; cp++)
5123 error ("unterminated language name in regex: %s", regex_arg);
5127 lang = get_language_from_langname (lang_name);
5130 add_regex (cp + 1, ignore_case, lang);
5134 /* Regexp to be used for any language. */
5136 add_regex (regex_arg, ignore_case, NULL);
5141 /* Turn a name, which is an ed-style (but Emacs syntax) regular
5142 expression, into a real regular expression by compiling it. */
5144 add_regex (regexp_pattern, ignore_case, lang)
5145 char *regexp_pattern;
5149 static struct re_pattern_buffer zeropattern;
5152 struct re_pattern_buffer *patbuf;
5156 if (regexp_pattern[strlen(regexp_pattern)-1] != regexp_pattern[0])
5158 error ("%s: unterminated regexp", regexp_pattern);
5161 name = scan_separators (regexp_pattern);
5162 if (regexp_pattern[0] == '\0')
5164 error ("null regexp", (char *)NULL);
5167 (void) scan_separators (name);
5169 patbuf = xnew (1, struct re_pattern_buffer);
5170 *patbuf = zeropattern;
5171 /* Translation table to fold case if appropriate. */
5172 patbuf->translate = (ignore_case) ? lc_trans : NULL;
5174 err = re_compile_pattern (regexp_pattern, strlen (regexp_pattern), patbuf);
5177 error ("%s while compiling pattern", err);
5182 p_head = xnew (1, pattern);
5183 p_head->regex = savestr (regexp_pattern);
5184 p_head->p_next = pp;
5185 p_head->language = lang;
5186 p_head->pattern = patbuf;
5187 p_head->name_pattern = savestr (name);
5188 p_head->error_signaled = FALSE;
5192 * Do the substitutions indicated by the regular expression and
5196 substitute (in, out, regs)
5198 struct re_registers *regs;
5201 int size, dig, diglen;
5204 size = strlen (out);
5206 /* Pass 1: figure out how much to allocate by finding all \N strings. */
5207 if (out[size - 1] == '\\')
5208 fatal ("pattern error in \"%s\"", out);
5209 for (t = etags_strchr (out, '\\');
5211 t = etags_strchr (t + 2, '\\'))
5215 diglen = regs->end[dig] - regs->start[dig];
5221 /* Allocate space and do the substitutions. */
5222 result = xnew (size + 1, char);
5224 for (t = result; *out != '\0'; out++)
5225 if (*out == '\\' && ISDIGIT (*++out))
5228 diglen = regs->end[dig] - regs->start[dig];
5229 strncpy (t, in + regs->start[dig], diglen);
5236 assert (t <= result + size && t - result == (int)strlen (result));
5241 /* Deallocate all patterns. */
5246 while (p_head != NULL)
5248 pp = p_head->p_next;
5249 free (p_head->regex);
5250 free (p_head->name_pattern);
5256 #endif /* ETAGS_REGEXPS */
5267 /* Go till you get to white space or a syntactic break */
5269 *cp != '\0' && *cp != '(' && *cp != ')' && !iswhite (*cp);
5272 pfnote (savenstr (bp, cp-bp), TRUE,
5273 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
5276 /* Initialize a linebuffer for use */
5281 lbp->size = (DEBUG) ? 3 : 200;
5282 lbp->buffer = xnew (lbp->size, char);
5283 lbp->buffer[0] = '\0';
5288 * Read a line of text from `stream' into `lbp', excluding the
5289 * newline or CR-NL, if any. Return the number of characters read from
5290 * `stream', which is the length of the line including the newline.
5292 * On DOS or Windows we do not count the CR character, if any, before the
5293 * NL, in the returned length; this mirrors the behavior of emacs on those
5294 * platforms (for text files, it translates CR-NL to NL as it reads in the
5298 readline_internal (lbp, stream)
5300 register FILE *stream;
5302 char *buffer = lbp->buffer;
5303 register char *p = lbp->buffer;
5304 register char *pend;
5307 pend = p + lbp->size; /* Separate to avoid 386/IX compiler bug. */
5311 register int c = getc (stream);
5314 /* We're at the end of linebuffer: expand it. */
5316 xrnew (buffer, lbp->size, char);
5317 p += buffer - lbp->buffer;
5318 pend = buffer + lbp->size;
5319 lbp->buffer = buffer;
5329 if (p > buffer && p[-1] == '\r')
5333 /* Assume CRLF->LF translation will be performed by Emacs
5334 when loading this file, so CRs won't appear in the buffer.
5335 It would be cleaner to compensate within Emacs;
5336 however, Emacs does not know how many CRs were deleted
5337 before any given point in the file. */
5352 lbp->len = p - buffer;
5354 return lbp->len + chars_deleted;
5358 * Like readline_internal, above, but in addition try to match the
5359 * input line against relevant regular expressions.
5362 readline (lbp, stream)
5366 /* Read new line. */
5367 long result = readline_internal (lbp, stream);
5368 #ifdef ETAGS_REGEXPS
5372 /* Match against relevant patterns. */
5374 for (pp = p_head; pp != NULL; pp = pp->p_next)
5376 /* Only use generic regexps or those for the current language. */
5377 if (pp->language != NULL && pp->language != curlang)
5380 match = re_match (pp->pattern, lbp->buffer, lbp->len, 0, &pp->regs);
5385 if (!pp->error_signaled)
5387 error ("error while matching \"%s\"", pp->regex);
5388 pp->error_signaled = TRUE;
5395 /* Match occurred. Construct a tag. */
5396 if (pp->name_pattern[0] != '\0')
5398 /* Make a named tag. */
5399 char *name = substitute (lbp->buffer,
5400 pp->name_pattern, &pp->regs);
5402 pfnote (name, TRUE, lbp->buffer, match, lineno, linecharno);
5406 /* Make an unnamed tag. */
5407 pfnote ((char *)NULL, TRUE,
5408 lbp->buffer, match, lineno, linecharno);
5413 #endif /* ETAGS_REGEXPS */
5420 * Return a pointer to a space of size strlen(cp)+1 allocated
5421 * with xnew where the string CP has been copied.
5427 return savenstr (cp, strlen (cp));
5431 * Return a pointer to a space of size LEN+1 allocated with xnew where
5432 * the string CP has been copied for at most the first LEN characters.
5441 dp = xnew (len + 1, char);
5442 strncpy (dp, cp, len);
5448 * Return the ptr in sp at which the character c last
5449 * appears; NULL if not found
5451 * Identical to POSIX strrchr, included for portability.
5454 etags_strrchr (sp, c)
5455 register const char *sp;
5458 register const char *r;
5471 * Return the ptr in sp at which the character c first
5472 * appears; NULL if not found
5474 * Identical to POSIX strchr, included for portability.
5477 etags_strchr (sp, c)
5478 register const char *sp;
5489 /* Skip spaces, return new pointer. */
5494 while (iswhite (*cp))
5499 /* Skip non spaces, return new pointer. */
5501 skip_non_spaces (cp)
5504 while (*cp != '\0' && !iswhite (*cp))
5509 /* Print error message and exit. */
5527 suggest_asking_for_help ()
5529 fprintf (stderr, "\tTry `%s %s' for a complete list of options.\n",
5540 /* Print error message. `s1' is printf control string, `s2' is arg for it. */
5543 const char *s1, *s2;
5545 fprintf (stderr, "%s: ", progname);
5546 fprintf (stderr, s1, s2);
5547 fprintf (stderr, "\n");
5550 /* Return a newly-allocated string whose contents
5551 concatenate those of s1, s2, s3. */
5556 int len1 = strlen (s1), len2 = strlen (s2), len3 = strlen (s3);
5557 char *result = xnew (len1 + len2 + len3 + 1, char);
5559 strcpy (result, s1);
5560 strcpy (result + len1, s2);
5561 strcpy (result + len1 + len2, s3);
5562 result[len1 + len2 + len3] = '\0';
5568 /* Does the same work as the system V getcwd, but does not need to
5569 guess the buffer size in advance. */
5575 char *path = xnew (bufsize, char);
5577 while (getcwd (path, bufsize) == NULL)
5579 if (errno != ERANGE)
5583 path = xnew (bufsize, char);
5586 canonicalize_filename (path);
5589 #else /* not HAVE_GETCWD */
5592 char *p, path[MAXPATHLEN + 1]; /* Fixed size is safe on MSDOS. */
5596 for (p = path; *p != '\0'; p++)
5602 return strdup (path);
5603 #else /* not MSDOS */
5608 pipe = (FILE *) popen ("pwd 2>/dev/null", "r");
5609 if (pipe == NULL || readline_internal (&path, pipe) == 0)
5614 #endif /* not MSDOS */
5615 #endif /* not HAVE_GETCWD */
5618 /* Return a newly allocated string containing the file name of FILE
5619 relative to the absolute directory DIR (which should end with a slash). */
5621 relative_filename (file, dir)
5624 char *fp, *dp, *afn, *res;
5627 /* Find the common root of file and dir (with a trailing slash). */
5628 afn = absolute_filename (file, cwd);
5631 while (*fp++ == *dp++)
5633 fp--, dp--; /* back to the first differing char */
5635 if (fp == afn && afn[0] != '/') /* cannot build a relative name */
5638 do /* look at the equal chars until '/' */
5642 /* Build a sequence of "../" strings for the resulting relative file name. */
5644 while ((dp = etags_strchr (dp + 1, '/')) != NULL)
5646 res = xnew (3*i + strlen (fp + 1) + 1, char);
5649 strcat (res, "../");
5651 /* Add the file name relative to the common root of file and dir. */
5652 strcat (res, fp + 1);
5658 /* Return a newly allocated string containing the absolute file name
5659 of FILE given DIR (which should end with a slash). */
5661 absolute_filename (file, dir)
5664 char *slashp, *cp, *res;
5666 if (filename_is_absolute (file))
5667 res = savestr (file);
5669 /* We don't support non-absolute file names with a drive
5670 letter, like `d:NAME' (it's too much hassle). */
5671 else if (file[1] == ':')
5672 fatal ("%s: relative file names with drive letters not supported", file);
5675 res = concat (dir, file, "");
5677 /* Delete the "/dirname/.." and "/." substrings. */
5678 slashp = etags_strchr (res, '/');
5679 while (slashp != NULL && slashp[0] != '\0')
5681 if (slashp[1] == '.')
5683 if (slashp[2] == '.'
5684 && (slashp[3] == '/' || slashp[3] == '\0'))
5689 while (cp >= res && !filename_is_absolute (cp));
5691 cp = slashp; /* the absolute name begins with "/.." */
5693 /* Under MSDOS and NT we get `d:/NAME' as absolute
5694 file name, so the luser could say `d:/../NAME'.
5695 We silently treat this as `d:/NAME'. */
5696 else if (cp[0] != '/')
5699 strcpy (cp, slashp + 3);
5703 else if (slashp[2] == '/' || slashp[2] == '\0')
5705 strcpy (slashp, slashp + 2);
5710 slashp = etags_strchr (slashp + 1, '/');
5714 return savestr ("/");
5719 /* Return a newly allocated string containing the absolute
5720 file name of dir where FILE resides given DIR (which should
5721 end with a slash). */
5723 absolute_dirname (file, dir)
5729 canonicalize_filename (file);
5730 slashp = etags_strrchr (file, '/');
5732 return savestr (dir);
5735 res = absolute_filename (file, dir);
5741 /* Whether the argument string is an absolute file name. The argument
5742 string must have been canonicalized with canonicalize_filename. */
5744 filename_is_absolute (fn)
5747 return (fn[0] == '/'
5749 || (ISALPHA(fn[0]) && fn[1] == ':' && fn[2] == '/')
5754 /* Translate backslashes into slashes. Works in place. */
5756 canonicalize_filename (fn)
5760 /* Canonicalize drive letter case. */
5761 if (fn[0] != '\0' && fn[1] == ':' && ISLOWER (fn[0]))
5762 fn[0] = upcase (fn[0]);
5763 /* Convert backslashes to slashes. */
5764 for (; *fn != '\0'; fn++)
5769 fn = NULL; /* shut up the compiler */
5773 /* Set the minimum size of a string contained in a linebuffer. */
5775 linebuffer_setlen (lbp, toksize)
5779 while (lbp->size <= toksize)
5782 xrnew (lbp->buffer, lbp->size, char);
5787 /* Like malloc but get fatal error if memory is exhausted. */
5792 long *result = (long *) malloc (size);
5794 fatal ("virtual memory exhausted", (char *)NULL);
5799 xrealloc (ptr, size)
5803 long *result = (long *) realloc (ptr, size);
5805 fatal ("virtual memory exhausted", (char *)NULL);