lib-src/etags.c

   1 /* Tags file maker to go with GNU Emacs           -*- coding: latin-1 -*-
   2    Copyright (C) 1984, 1987-1989, 1993-1995, 1998-2001, 2002
   3    Free Software Foundation, Inc. and Ken Arnold
   4
   5  This file is not considered part of GNU Emacs.
   6
   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.
  11
  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.
  16
  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. */
  20
  21 /*
  22  * Authors:
  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  * 1992 Joseph B. Wells improved C and C++ parsing.
  29  * 1993 Francesco Potortì reorganised C and C++.
  30  * 1994 Line-by-line regexp tags by Tom Tromey.
  31  * 2001 Nested classes by Francesco Potortì (concept by Mykola Dzyuba).
  32  * 2002 #line directives by Francesco Potortì.
  33  *
  34  * Francesco Potortì <pot@gnu.org> has maintained and improved it since 1993.
  35  *
  36  */
  37
  38 char pot_etags_version[] = "@(#) pot revision number is 16.56";
  39
  40 #define TRUE    1
  41 #define FALSE   0
  42
  43 #ifdef DEBUG
  44 #  undef DEBUG
  45 #  define DEBUG TRUE
  46 #else
  47 #  define DEBUG  FALSE
  48 #  define NDEBUG                /* disable assert */
  49 #endif
  50
  51 #ifdef HAVE_CONFIG_H
  52 # include <config.h>
  53   /* On some systems, Emacs defines static as nothing for the sake
  54      of unexec.  We don't want that here since we don't use unexec. */
  55 # undef static
  56 # define ETAGS_REGEXPS          /* use the regexp features */
  57 # define LONG_OPTIONS           /* accept long options */
  58 # ifndef PTR                    /* for Xemacs */
  59 #   define PTR void *
  60 # endif
  61 # ifndef __P                    /* for Xemacs */
  62 #   define __P(args) args
  63 # endif
  64 #else
  65 # if defined(__STDC__) && (__STDC__ || defined(__SUNPRO_C))
  66 #   define __P(args) args       /* use prototypes */
  67 #   define PTR void *           /* for generic pointers */
  68 # else
  69 #   define __P(args) ()         /* no prototypes */
  70 #   define const                /* remove const for old compilers' sake */
  71 #   define PTR long *           /* don't use void* */
  72 # endif
  73 #endif /* !HAVE_CONFIG_H */
  74
  75 #ifndef _GNU_SOURCE
  76 # define _GNU_SOURCE 1          /* enables some compiler checks on GNU */
  77 #endif
  78
  79 /* WIN32_NATIVE is for Xemacs.
  80    MSDOS, WINDOWSNT, DOS_NT are for Emacs. */
  81 #ifdef WIN32_NATIVE
  82 # undef MSDOS
  83 # undef  WINDOWSNT
  84 # define WINDOWSNT
  85 #endif /* WIN32_NATIVE */
  86
  87 #ifdef MSDOS
  88 # undef MSDOS
  89 # define MSDOS TRUE
  90 # include <fcntl.h>
  91 # include <sys/param.h>
  92 # include <io.h>
  93 # ifndef HAVE_CONFIG_H
  94 #   define DOS_NT
  95 #   include <sys/config.h>
  96 # endif
  97 #else
  98 # define MSDOS FALSE
  99 #endif /* MSDOS */
 100
 101 #ifdef WINDOWSNT
 102 # include <stdlib.h>
 103 # include <fcntl.h>
 104 # include <string.h>
 105 # include <direct.h>
 106 # include <io.h>
 107 # define MAXPATHLEN _MAX_PATH
 108 # undef HAVE_NTGUI
 109 # undef  DOS_NT
 110 # define DOS_NT
 111 # ifndef HAVE_GETCWD
 112 #   define HAVE_GETCWD
 113 # endif /* undef HAVE_GETCWD */
 114 #else /* !WINDOWSNT */
 115 # ifdef STDC_HEADERS
 116 #  include <stdlib.h>
 117 #  include <string.h>
 118 # else
 119     extern char *getenv ();
 120 # endif
 121 #endif /* !WINDOWSNT */
 122
 123 #ifdef HAVE_UNISTD_H
 124 # include <unistd.h>
 125 #else
 126 # if defined (HAVE_GETCWD) && !defined (WINDOWSNT)
 127     extern char *getcwd (char *buf, size_t size);
 128 # endif
 129 #endif /* HAVE_UNISTD_H */
 130
 131 #include <stdio.h>
 132 #include <ctype.h>
 133 #include <errno.h>
 134 #ifndef errno
 135   extern int errno;
 136 #endif
 137 #include <sys/types.h>
 138 #include <sys/stat.h>
 139
 140 #include <assert.h>
 141 #ifdef NDEBUG
 142 # undef  assert                 /* some systems have a buggy assert.h */
 143 # define assert(x) ((void) 0)
 144 #endif
 145
 146 #if !defined (S_ISREG) && defined (S_IFREG)
 147 # define S_ISREG(m)     (((m) & S_IFMT) == S_IFREG)
 148 #endif
 149
 150 #ifdef LONG_OPTIONS
 151 # include <getopt.h>
 152 #else
 153 # define getopt_long(argc,argv,optstr,lopts,lind) getopt (argc, argv, optstr)
 154   extern char *optarg;
 155   extern int optind, opterr;
 156 #endif /* LONG_OPTIONS */
 157
 158 #ifdef ETAGS_REGEXPS
 159 # ifndef HAVE_CONFIG_H          /* this is a standalone compilation */
 160 #   ifdef __CYGWIN__            /* compiling on Cygwin */
 161                              !!! NOTICE !!!
 162  the regex.h distributed with Cygwin is not compatible with etags, alas!
 163 If you want regular expression support, you should delete this notice and
 164               arrange to use the GNU regex.h and regex.c.
 165 #   endif
 166 # endif
 167 # include <regex.h>
 168 #endif /* ETAGS_REGEXPS */
 169
 170 /* Define CTAGS to make the program "ctags" compatible with the usual one.
 171  Leave it undefined to make the program "etags", which makes emacs-style
 172  tag tables and tags typedefs, #defines and struct/union/enum by default. */
 173 #ifdef CTAGS
 174 # undef  CTAGS
 175 # define CTAGS TRUE
 176 #else
 177 # define CTAGS FALSE
 178 #endif
 179
 180 /* Exit codes for success and failure.  */
 181 #ifdef VMS
 182 # define        GOOD    1
 183 # define        BAD     0
 184 #else
 185 # define        GOOD    0
 186 # define        BAD     1
 187 #endif
 188
 189 #define streq(s,t)      (assert((s)!=NULL || (t)!=NULL), !strcmp (s, t))
 190 #define strcaseeq(s,t)  (assert((s)!=NULL && (t)!=NULL), !etags_strcasecmp (s, t))
 191 #define strneq(s,t,n)   (assert((s)!=NULL || (t)!=NULL), !strncmp (s, t, n))
 192 #define strncaseeq(s,t,n) (assert((s)!=NULL && (t)!=NULL), !etags_strncasecmp (s, t, n))
 193
 194 #define CHARS 256               /* 2^sizeof(char) */
 195 #define CHAR(x)         ((unsigned int)(x) & (CHARS - 1))
 196 #define iswhite(c)      (_wht[CHAR(c)]) /* c is white (see white) */
 197 #define notinname(c)    (_nin[CHAR(c)]) /* c is not in a name (see nonam) */
 198 #define begtoken(c)     (_btk[CHAR(c)]) /* c can start token (see begtk) */
 199 #define intoken(c)      (_itk[CHAR(c)]) /* c can be in token (see midtk) */
 200 #define endtoken(c)     (_etk[CHAR(c)]) /* c ends tokens (see endtk) */
 201
 202 #define ISALNUM(c)      isalnum (CHAR(c))
 203 #define ISALPHA(c)      isalpha (CHAR(c))
 204 #define ISDIGIT(c)      isdigit (CHAR(c))
 205 #define ISLOWER(c)      islower (CHAR(c))
 206
 207 #define lowcase(c)      tolower (CHAR(c))
 208 #define upcase(c)       toupper (CHAR(c))
 209
 210
 211 /*
 212  *      xnew, xrnew -- allocate, reallocate storage
 213  *
 214  * SYNOPSIS:    Type *xnew (int n, Type);
 215  *              void xrnew (OldPointer, int n, Type);
 216  */
 217 #if DEBUG
 218 # include "chkmalloc.h"
 219 # define xnew(n,Type)     ((Type *) trace_malloc (__FILE__, __LINE__, \
 220                                                   (n) * sizeof (Type)))
 221 # define xrnew(op,n,Type) ((op) = (Type *) trace_realloc (__FILE__, __LINE__, \
 222                                         (char *) (op), (n) * sizeof (Type)))
 223 #else
 224 # define xnew(n,Type)     ((Type *) xmalloc ((n) * sizeof (Type)))
 225 # define xrnew(op,n,Type) ((op) = (Type *) xrealloc ( \
 226                                         (char *) (op), (n) * sizeof (Type)))
 227 #endif
 228
 229 #define bool int
 230
 231 typedef void Lang_function __P((FILE *));
 232
 233 typedef struct
 234 {
 235   char *suffix;                 /* file name suffix for this compressor */
 236   char *command;                /* takes one arg and decompresses to stdout */
 237 } compressor;
 238
 239 typedef struct
 240 {
 241   char *name;                   /* language name */
 242   char *help;                   /* detailed help for the language */
 243   Lang_function *function;      /* parse function */
 244   char **suffixes;              /* name suffixes of this language's files */
 245   char **filenames;             /* names of this language's files */
 246   char **interpreters;          /* interpreters for this language */
 247   bool metasource;              /* source used to generate other sources */
 248 } language;
 249
 250 typedef struct fdesc
 251 {
 252   struct fdesc *next;           /* for the linked list */
 253   char *infname;                /* uncompressed input file name */
 254   char *infabsname;             /* absolute uncompressed input file name */
 255   char *infabsdir;              /* absolute dir of input file */
 256   char *taggedfname;            /* file name to write in tagfile */
 257   language *lang;               /* language of file */
 258   char *prop;                   /* file properties to write in tagfile */
 259   bool usecharno;               /* etags tags shall contain char number */
 260   bool written;                 /* entry written in the tags file */
 261 } fdesc;
 262
 263 typedef struct node_st
 264 {                               /* sorting structure */
 265   struct node_st *left, *right; /* left and right sons */
 266   fdesc *fdp;                   /* description of file to whom tag belongs */
 267   char *name;                   /* tag name */
 268   char *regex;                  /* search regexp */
 269   bool valid;                   /* write this tag on the tag file */
 270   bool is_func;                 /* function tag: use regexp in CTAGS mode */
 271   bool been_warned;             /* warning already given for duplicated tag */
 272   int lno;                      /* line number tag is on */
 273   long cno;                     /* character number line starts on */
 274 } node;
 275
 276 /*
 277  * A `linebuffer' is a structure which holds a line of text.
 278  * `readline_internal' reads a line from a stream into a linebuffer
 279  * and works regardless of the length of the line.
 280  * SIZE is the size of BUFFER, LEN is the length of the string in
 281  * BUFFER after readline reads it.
 282  */
 283 typedef struct
 284 {
 285   long size;
 286   int len;
 287   char *buffer;
 288 } linebuffer;
 289
 290 /* Used to support mixing of --lang and file names. */
 291 typedef struct
 292 {
 293   enum {
 294     at_language,                /* a language specification */
 295     at_regexp,                  /* a regular expression */
 296     at_filename,                /* a file name */
 297     at_stdin,                   /* read from stdin here */
 298     at_end                      /* stop parsing the list */
 299   } arg_type;                   /* argument type */
 300   language *lang;               /* language associated with the argument */
 301   char *what;                   /* the argument itself */
 302 } argument;
 303
 304 #ifdef ETAGS_REGEXPS
 305 /* Structure defining a regular expression. */
 306 typedef struct regexp
 307 {
 308   struct regexp *p_next;        /* pointer to next in list */
 309   language *lang;               /* if set, use only for this language */
 310   char *pattern;                /* the regexp pattern */
 311   char *name;                   /* tag name */
 312   struct re_pattern_buffer *pat; /* the compiled pattern */
 313   struct re_registers regs;     /* re registers */
 314   bool error_signaled;          /* already signaled for this regexp */
 315   bool force_explicit_name;     /* do not allow implict tag name */
 316   bool ignore_case;             /* ignore case when matching */
 317   bool multi_line;              /* do a multi-line match on the whole file */
 318 } regexp;
 319 #endif /* ETAGS_REGEXPS */
 320
 321
 322 /* Many compilers barf on this:
 323         Lang_function Ada_funcs;
 324    so let's write it this way */
 325 static void Ada_funcs __P((FILE *));
 326 static void Asm_labels __P((FILE *));
 327 static void C_entries __P((int c_ext, FILE *));
 328 static void default_C_entries __P((FILE *));
 329 static void plain_C_entries __P((FILE *));
 330 static void Cjava_entries __P((FILE *));
 331 static void Cobol_paragraphs __P((FILE *));
 332 static void Cplusplus_entries __P((FILE *));
 333 static void Cstar_entries __P((FILE *));
 334 static void Erlang_functions __P((FILE *));
 335 static void Fortran_functions __P((FILE *));
 336 static void HTML_labels __P((FILE *));
 337 static void Lisp_functions __P((FILE *));
 338 static void Makefile_targets __P((FILE *));
 339 static void Pascal_functions __P((FILE *));
 340 static void Perl_functions __P((FILE *));
 341 static void PHP_functions __P((FILE *));
 342 static void PS_functions __P((FILE *));
 343 static void Prolog_functions __P((FILE *));
 344 static void Python_functions __P((FILE *));
 345 static void Scheme_functions __P((FILE *));
 346 static void TeX_commands __P((FILE *));
 347 static void Texinfo_nodes __P((FILE *));
 348 static void Yacc_entries __P((FILE *));
 349 static void just_read_file __P((FILE *));
 350
 351 static void print_language_names __P((void));
 352 static void print_version __P((void));
 353 static void print_help __P((argument *));
 354 int main __P((int, char **));
 355
 356 static compressor *get_compressor_from_suffix __P((char *, char **));
 357 static language *get_language_from_langname __P((const char *));
 358 static language *get_language_from_interpreter __P((char *));
 359 static language *get_language_from_filename __P((char *, bool));
 360 static void readline __P((linebuffer *, FILE *));
 361 static long readline_internal __P((linebuffer *, FILE *));
 362 static bool nocase_tail __P((char *));
 363 static void get_tag __P((char *, char **));
 364
 365 #ifdef ETAGS_REGEXPS
 366 static void analyse_regex __P((char *));
 367 static void free_regexps __P((void));
 368 static void regex_tag_multiline __P((void));
 369 #endif /* ETAGS_REGEXPS */
 370 static void error __P((const char *, const char *));
 371 static void suggest_asking_for_help __P((void));
 372 void fatal __P((char *, char *));
 373 static void pfatal __P((char *));
 374 static void add_node __P((node *, node **));
 375
 376 static void init __P((void));
 377 static void process_file_name __P((char *, language *));
 378 static void process_file __P((FILE *, char *, language *));
 379 static void find_entries __P((FILE *));
 380 static void free_tree __P((node *));
 381 static void free_fdesc __P((fdesc *));
 382 static void pfnote __P((char *, bool, char *, int, int, long));
 383 static void make_tag __P((char *, int, bool, char *, int, int, long));
 384 static void invalidate_nodes __P((fdesc *, node **));
 385 static void put_entries __P((node *));
 386
 387 static char *concat __P((char *, char *, char *));
 388 static char *skip_spaces __P((char *));
 389 static char *skip_non_spaces __P((char *));
 390 static char *savenstr __P((char *, int));
 391 static char *savestr __P((char *));
 392 static char *etags_strchr __P((const char *, int));
 393 static char *etags_strrchr __P((const char *, int));
 394 static int etags_strcasecmp __P((const char *, const char *));
 395 static int etags_strncasecmp __P((const char *, const char *, int));
 396 static char *etags_getcwd __P((void));
 397 static char *relative_filename __P((char *, char *));
 398 static char *absolute_filename __P((char *, char *));
 399 static char *absolute_dirname __P((char *, char *));
 400 static bool filename_is_absolute __P((char *f));
 401 static void canonicalize_filename __P((char *));
 402 static void linebuffer_init __P((linebuffer *));
 403 static void linebuffer_setlen __P((linebuffer *, int));
 404 static PTR xmalloc __P((unsigned int));
 405 static PTR xrealloc __P((char *, unsigned int));
 406
 407 \f
 408 static char searchar = '/';     /* use /.../ searches */
 409
 410 static char *tagfile;           /* output file */
 411 static char *progname;          /* name this program was invoked with */
 412 static char *cwd;               /* current working directory */
 413 static char *tagfiledir;        /* directory of tagfile */
 414 static FILE *tagf;              /* ioptr for tags file */
 415
 416 static fdesc *fdhead;           /* head of file description list */
 417 static fdesc *curfdp;           /* current file description */
 418 static int lineno;              /* line number of current line */
 419 static long charno;             /* current character number */
 420 static long linecharno;         /* charno of start of current line */
 421 static char *dbp;               /* pointer to start of current tag */
 422
 423 static const int invalidcharno = -1;
 424
 425 static node *nodehead;          /* the head of the binary tree of tags */
 426 static node *last_node;         /* the last node created */
 427
 428 static linebuffer lb;           /* the current line */
 429 static linebuffer filebuf;      /* a buffer containing the whole file */
 430 static linebuffer token_name;   /* a buffer containing a tag name */
 431
 432 /* boolean "functions" (see init)       */
 433 static bool _wht[CHARS], _nin[CHARS], _itk[CHARS], _btk[CHARS], _etk[CHARS];
 434 static char
 435   /* white chars */
 436   *white = " \f\t\n\r\v",
 437   /* not in a name */
 438   *nonam = " \f\t\n\r()=,;",    /* look at make_tag before modifying! */
 439   /* token ending chars */
 440   *endtk = " \t\n\r\"'#()[]{}=-+%*/&|^~!<>;,.:?",
 441   /* token starting chars */
 442   *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~@",
 443   /* valid in-token chars */
 444   *midtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789";
 445
 446 static bool append_to_tagfile;  /* -a: append to tags */
 447 /* The next four default to TRUE for etags, but to FALSE for ctags.  */
 448 static bool typedefs;           /* -t: create tags for C and Ada typedefs */
 449 static bool typedefs_or_cplusplus; /* -T: create tags for C typedefs, level */
 450                                 /* 0 struct/enum/union decls, and C++ */
 451                                 /* member functions. */
 452 static bool constantypedefs;    /* -d: create tags for C #define, enum */
 453                                 /* constants and variables. */
 454                                 /* -D: opposite of -d.  Default under ctags. */
 455 static bool globals;            /* create tags for global variables */
 456 static bool declarations;       /* --declarations: tag them and extern in C&Co*/
 457 static bool members;            /* create tags for C member variables */
 458 static bool no_line_directive;  /* ignore #line directives (undocumented) */
 459 static bool update;             /* -u: update tags */
 460 static bool vgrind_style;       /* -v: create vgrind style index output */
 461 static bool no_warnings;        /* -w: suppress warnings */
 462 static bool cxref_style;        /* -x: create cxref style output */
 463 static bool cplusplus;          /* .[hc] means C++, not C */
 464 static bool ignoreindent;       /* -I: ignore indentation in C */
 465 static bool packages_only;      /* --packages-only: in Ada, only tag packages*/
 466
 467 #define STDIN 0x1001            /* returned by getopt_long on --parse-stdin */
 468 static bool parsing_stdin;      /* --parse-stdin used */
 469
 470 #ifdef ETAGS_REGEXPS
 471 static regexp *p_head;          /* list of all regexps */
 472 static bool need_filebuf;       /* some regexes are multi-line */
 473 #else
 474 # define need_filebuf FALSE
 475 #endif /* ETAGS_REGEXPS */
 476
 477 #ifdef LONG_OPTIONS
 478 static struct option longopts[] =
 479 {
 480   { "packages-only",      no_argument,       &packages_only,     TRUE  },
 481   { "c++",                no_argument,       NULL,               'C'   },
 482   { "declarations",       no_argument,       &declarations,      TRUE  },
 483   { "no-line-directive",  no_argument,       &no_line_directive, TRUE  },
 484   { "help",               no_argument,       NULL,               'h'   },
 485   { "help",               no_argument,       NULL,               'H'   },
 486   { "ignore-indentation", no_argument,       NULL,               'I'   },
 487   { "language",           required_argument, NULL,               'l'   },
 488   { "members",            no_argument,       &members,           TRUE  },
 489   { "no-members",         no_argument,       &members,           FALSE },
 490   { "output",             required_argument, NULL,               'o'   },
 491 #ifdef ETAGS_REGEXPS
 492   { "regex",              required_argument, NULL,               'r'   },
 493   { "no-regex",           no_argument,       NULL,               'R'   },
 494   { "ignore-case-regex",  required_argument, NULL,               'c'   },
 495 #endif /* ETAGS_REGEXPS */
 496   { "parse-stdin",        required_argument, NULL,               STDIN },
 497   { "version",            no_argument,       NULL,               'V'   },
 498
 499 #if CTAGS /* Etags options */
 500   { "backward-search",    no_argument,       NULL,               'B'   },
 501   { "cxref",              no_argument,       NULL,               'x'   },
 502   { "defines",            no_argument,       NULL,               'd'   },
 503   { "globals",            no_argument,       &globals,           TRUE  },
 504   { "typedefs",           no_argument,       NULL,               't'   },
 505   { "typedefs-and-c++",   no_argument,       NULL,               'T'   },
 506   { "update",             no_argument,       NULL,               'u'   },
 507   { "vgrind",             no_argument,       NULL,               'v'   },
 508   { "no-warn",            no_argument,       NULL,               'w'   },
 509
 510 #else /* Ctags options */
 511   { "append",             no_argument,       NULL,               'a'   },
 512   { "no-defines",         no_argument,       NULL,               'D'   },
 513   { "no-globals",         no_argument,       &globals,           FALSE },
 514   { "include",            required_argument, NULL,               'i'   },
 515 #endif
 516   { NULL }
 517 };
 518 #endif /* LONG_OPTIONS */
 519
 520 static compressor compressors[] =
 521 {
 522   { "z", "gzip -d -c"},
 523   { "Z", "gzip -d -c"},
 524   { "gz", "gzip -d -c"},
 525   { "GZ", "gzip -d -c"},
 526   { "bz2", "bzip2 -d -c" },
 527   { NULL }
 528 };
 529
 530 /*
 531  * Language stuff.
 532  */
 533
 534 /* Ada code */
 535 static char *Ada_suffixes [] =
 536   { "ads", "adb", "ada", NULL };
 537 static char Ada_help [] =
 538 "In Ada code, functions, procedures, packages, tasks and types are\n\
 539 tags.  Use the `--packages-only' option to create tags for\n\
 540 packages only.\n\
 541 Ada tag names have suffixes indicating the type of entity:\n\
 542         Entity type:    Qualifier:\n\
 543         ------------    ----------\n\
 544         function        /f\n\
 545         procedure       /p\n\
 546         package spec    /s\n\
 547         package body    /b\n\
 548         type            /t\n\
 549         task            /k\n\
 550 Thus, `M-x find-tag <RET> bidule/b <RET>' will go directly to the\n\
 551 body of the package `bidule', while `M-x find-tag <RET> bidule <RET>'\n\
 552 will just search for any tag `bidule'.";
 553
 554 /* Assembly code */
 555 static char *Asm_suffixes [] =
 556   { "a",        /* Unix assembler */
 557     "asm", /* Microcontroller assembly */
 558     "def", /* BSO/Tasking definition includes  */
 559     "inc", /* Microcontroller include files */
 560     "ins", /* Microcontroller include files */
 561     "s", "sa", /* Unix assembler */
 562     "S",   /* cpp-processed Unix assembler */
 563     "src", /* BSO/Tasking C compiler output */
 564     NULL
 565   };
 566 static char Asm_help [] =
 567 "In assembler code, labels appearing at the beginning of a line,\n\
 568 followed by a colon, are tags.";
 569
 570
 571 /* Note that .c and .h can be considered C++, if the --c++ flag was
 572    given, or if the `class' or `template' keyowrds are met inside the file.
 573    That is why default_C_entries is called for these. */
 574 static char *default_C_suffixes [] =
 575   { "c", "h", NULL };
 576 static char default_C_help [] =
 577 "In C code, any C function or typedef is a tag, and so are\n\
 578 definitions of `struct', `union' and `enum'.  `#define' macro\n\
 579 definitions and `enum' constants are tags unless you specify\n\
 580 `--no-defines'.  Global variables are tags unless you specify\n\
 581 `--no-globals'.  Use of `--no-globals' and `--no-defines'\n\
 582 can make the tags table file much smaller.\n\
 583 You can tag function declarations and external variables by\n\
 584 using `--declarations', and struct members by using `--members'.";
 585
 586 static char *Cplusplus_suffixes [] =
 587   { "C", "c++", "cc", "cpp", "cxx", "H", "h++", "hh", "hpp", "hxx",
 588     "M",                        /* Objective C++ */
 589     "pdb",                      /* Postscript with C syntax */
 590     NULL };
 591 static char Cplusplus_help [] =
 592 "In C++ code, all the tag constructs of C code are tagged\n\
 593 (use --help --lang=c --lang=c++ for full help).\n\
 594 In addition to C tags, member functions are also recognized, and\n\
 595 optionally member variables if you use the `--members' option.\n\
 596 Tags for variables and functions in classes are named `CLASS::VARIABLE'\n\
 597 and `CLASS::FUNCTION'.  `operator' definitions have tag names like\n\
 598 `operator+'.";
 599
 600 static char *Cjava_suffixes [] =
 601   { "java", NULL };
 602 static char Cjava_help [] =
 603 "In Java code, all the tags constructs of C and C++ code are\n\
 604 tagged (use --help --lang=c --lang=c++ --lang=java for full help).";
 605
 606
 607 static char *Cobol_suffixes [] =
 608   { "COB", "cob", NULL };
 609 static char Cobol_help [] =
 610 "In Cobol code, tags are paragraph names; that is, any word\n\
 611 starting in column 8 and followed by a period.";
 612
 613 static char *Cstar_suffixes [] =
 614   { "cs", "hs", NULL };
 615
 616 static char *Erlang_suffixes [] =
 617   { "erl", "hrl", NULL };
 618 static char Erlang_help [] =
 619 "In Erlang code, the tags are the functions, records and macros\n\
 620 defined in the file.";
 621
 622 static char *Fortran_suffixes [] =
 623   { "F", "f", "f90", "for", NULL };
 624 static char Fortran_help [] =
 625 "In Fortran code, functions, subroutines and block data are tags.";
 626
 627 static char *HTML_suffixes [] =
 628   { "htm", "html", "shtml", NULL };
 629 static char HTML_help [] =
 630 "In HTML input files, the tags are the `title' and the `h1', `h2',\n\
 631 `h3' headers.  Also, tags are `name=' in anchors and all\n\
 632 occurrences of `id='.";
 633
 634 static char *Lisp_suffixes [] =
 635   { "cl", "clisp", "el", "l", "lisp", "LSP", "lsp", "ml", NULL };
 636 static char Lisp_help [] =
 637 "In Lisp code, any function defined with `defun', any variable\n\
 638 defined with `defvar' or `defconst', and in general the first\n\
 639 argument of any expression that starts with `(def' in column zero\n\
 640 is a tag.";
 641
 642 static char *Makefile_filenames [] =
 643   { "Makefile", "makefile", "GNUMakefile", "Makefile.in", "Makefile.am", NULL};
 644 static char Makefile_help [] =
 645 "In makefiles, targets are tags; additionally, variables are tags\n\
 646 unless you specify `--no-globals'.";
 647
 648 static char *Objc_suffixes [] =
 649   { "lm",                       /* Objective lex file */
 650     "m",                        /* Objective C file */
 651      NULL };
 652 static char Objc_help [] =
 653 "In Objective C code, tags include Objective C definitions for classes,\n\
 654 class categories, methods and protocols.  Tags for variables and\n\
 655 functions in classes are named `CLASS::VARIABLE' and `CLASS::FUNCTION'.";
 656
 657 static char *Pascal_suffixes [] =
 658   { "p", "pas", NULL };
 659 static char Pascal_help [] =
 660 "In Pascal code, the tags are the functions and procedures defined\n\
 661 in the file.";
 662
 663 static char *Perl_suffixes [] =
 664   { "pl", "pm", NULL };
 665 static char *Perl_interpreters [] =
 666   { "perl", "@PERL@", NULL };
 667 static char Perl_help [] =
 668 "In Perl code, the tags are the packages, subroutines and variables\n\
 669 defined by the `package', `sub', `my' and `local' keywords.  Use\n\
 670 `--globals' if you want to tag global variables.  Tags for\n\
 671 subroutines are named `PACKAGE::SUB'.  The name for subroutines\n\
 672 defined in the default package is `main::SUB'.";
 673
 674 static char *PHP_suffixes [] =
 675   { "php", "php3", "php4", NULL };
 676 static char PHP_help [] =
 677 "In PHP code, tags are functions, classes and defines.  When using\n\
 678 the `--members' option, vars are tags too.";
 679
 680 static char *plain_C_suffixes [] =
 681   { "pc",                       /* Pro*C file */
 682      NULL };
 683
 684 static char *PS_suffixes [] =
 685   { "ps", "psw", NULL };        /* .psw is for PSWrap */
 686 static char PS_help [] =
 687 "In PostScript code, the tags are the functions.";
 688
 689 static char *Prolog_suffixes [] =
 690   { "prolog", NULL };
 691 static char Prolog_help [] =
 692 "In Prolog code, tags are predicates and rules at the beginning of\n\
 693 line.";
 694
 695 static char *Python_suffixes [] =
 696   { "py", NULL };
 697 static char Python_help [] =
 698 "In Python code, `def' or `class' at the beginning of a line\n\
 699 generate a tag.";
 700
 701 /* Can't do the `SCM' or `scm' prefix with a version number. */
 702 static char *Scheme_suffixes [] =
 703   { "oak", "sch", "scheme", "SCM", "scm", "SM", "sm", "ss", "t", NULL };
 704 static char Scheme_help [] =
 705 "In Scheme code, tags include anything defined with `def' or with a\n\
 706 construct whose name starts with `def'.  They also include\n\
 707 variables set with `set!' at top level in the file.";
 708
 709 static char *TeX_suffixes [] =
 710   { "bib", "clo", "cls", "ltx", "sty", "TeX", "tex", NULL };
 711 static char TeX_help [] =
 712 "In LaTeX text, the argument of any of the commands `\\chapter',\n\
 713 `\\section', `\\subsection', `\\subsubsection', `\\eqno', `\\label',\n\
 714 `\\ref', `\\cite', `\\bibitem', `\\part', `\\appendix', `\\entry',\n\
 715 `\\index', `\\def', `\\newcommand', `\\renewcommand',\n\
 716 `\\newenvironment' or `\\renewenvironment' is a tag.\n\
 717 \n\
 718 Other commands can be specified by setting the environment variable\n\
 719 `TEXTAGS' to a colon-separated list like, for example,\n\
 720      TEXTAGS=\"mycommand:myothercommand\".";
 721
 722
 723 static char *Texinfo_suffixes [] =
 724   { "texi", "texinfo", "txi", NULL };
 725 static char Texinfo_help [] =
 726 "for texinfo files, lines starting with @node are tagged.";
 727
 728 static char *Yacc_suffixes [] =
 729   { "y", "y++", "ym", "yxx", "yy", NULL }; /* .ym is Objective yacc file */
 730 static char Yacc_help [] =
 731 "In Bison or Yacc input files, each rule defines as a tag the\n\
 732 nonterminal it constructs.  The portions of the file that contain\n\
 733 C code are parsed as C code (use --help --lang=c --lang=yacc\n\
 734 for full help).";
 735
 736 static char auto_help [] =
 737 "`auto' is not a real language, it indicates to use\n\
 738 a default language for files base on file name suffix and file contents.";
 739
 740 static char none_help [] =
 741 "`none' is not a real language, it indicates to only do\n\
 742 regexp processing on files.";
 743
 744 static char no_lang_help [] =
 745 "No detailed help available for this language.";
 746
 747
 748 /*
 749  * Table of languages.
 750  *
 751  * It is ok for a given function to be listed under more than one
 752  * name.  I just didn't.
 753  */
 754
 755 static language lang_names [] =
 756 {
 757   { "ada",       Ada_help,       Ada_funcs,         Ada_suffixes       },
 758   { "asm",       Asm_help,       Asm_labels,        Asm_suffixes       },
 759   { "c",         default_C_help, default_C_entries, default_C_suffixes },
 760   { "c++",       Cplusplus_help, Cplusplus_entries, Cplusplus_suffixes },
 761   { "c*",        no_lang_help,   Cstar_entries,     Cstar_suffixes     },
 762   { "cobol",     Cobol_help,     Cobol_paragraphs,  Cobol_suffixes     },
 763   { "erlang",    Erlang_help,    Erlang_functions,  Erlang_suffixes    },
 764   { "fortran",   Fortran_help,   Fortran_functions, Fortran_suffixes   },
 765   { "html",      HTML_help,      HTML_labels,       HTML_suffixes      },
 766   { "java",      Cjava_help,     Cjava_entries,     Cjava_suffixes     },
 767   { "lisp",      Lisp_help,      Lisp_functions,    Lisp_suffixes      },
 768   { "makefile",  Makefile_help,Makefile_targets,NULL,Makefile_filenames},
 769   { "objc",      Objc_help,      plain_C_entries,   Objc_suffixes      },
 770   { "pascal",    Pascal_help,    Pascal_functions,  Pascal_suffixes    },
 771   { "perl",Perl_help,Perl_functions,Perl_suffixes,NULL,Perl_interpreters},
 772   { "php",       PHP_help,       PHP_functions,     PHP_suffixes       },
 773   { "postscript",PS_help,        PS_functions,      PS_suffixes        },
 774   { "proc",      no_lang_help,   plain_C_entries,   plain_C_suffixes   },
 775   { "prolog",    Prolog_help,    Prolog_functions,  Prolog_suffixes    },
 776   { "python",    Python_help,    Python_functions,  Python_suffixes    },
 777   { "scheme",    Scheme_help,    Scheme_functions,  Scheme_suffixes    },
 778   { "tex",       TeX_help,       TeX_commands,      TeX_suffixes       },
 779   { "texinfo",   Texinfo_help,   Texinfo_nodes,     Texinfo_suffixes   },
 780   { "yacc",      Yacc_help,Yacc_entries,Yacc_suffixes,NULL,NULL,TRUE},
 781   { "auto",      auto_help },                      /* default guessing scheme */
 782   { "none",      none_help,      just_read_file }, /* regexp matching only */
 783   { NULL }                /* end of list */
 784 };
 785
 786 \f
 787 static void
 788 print_language_names ()
 789 {
 790   language *lang;
 791   char **name, **ext;
 792
 793   puts ("\nThese are the currently supported languages, along with the\n\
 794 default file names and dot suffixes:");
 795   for (lang = lang_names; lang->name != NULL; lang++)
 796     {
 797       printf ("  %-*s", 10, lang->name);
 798       if (lang->filenames != NULL)
 799         for (name = lang->filenames; *name != NULL; name++)
 800           printf (" %s", *name);
 801       if (lang->suffixes != NULL)
 802         for (ext = lang->suffixes; *ext != NULL; ext++)
 803           printf (" .%s", *ext);
 804       puts ("");
 805     }
 806   puts ("where `auto' means use default language for files based on file\n\
 807 name suffix, and `none' means only do regexp processing on files.\n\
 808 If no language is specified and no matching suffix is found,\n\
 809 the first line of the file is read for a sharp-bang (#!) sequence\n\
 810 followed by the name of an interpreter.  If no such sequence is found,\n\
 811 Fortran is tried first; if no tags are found, C is tried next.\n\
 812 When parsing any C file, a \"class\" or \"template\" keyword\n\
 813 switches to C++.");
 814   puts ("Compressed files are supported using gzip and bzip2.\n\
 815 \n\
 816 For detailed help on a given language use, for example,\n\
 817 etags --help --lang=ada.");
 818 }
 819
 820 #ifndef EMACS_NAME
 821 # define EMACS_NAME "standalone"
 822 #endif
 823 #ifndef VERSION
 824 # define VERSION "version"
 825 #endif
 826 static void
 827 print_version ()
 828 {
 829   printf ("%s (%s %s)\n", (CTAGS) ? "ctags" : "etags", EMACS_NAME, VERSION);
 830   puts ("Copyright (C) 2002 Free Software Foundation, Inc. and Ken Arnold");
 831   puts ("This program is distributed under the same terms as Emacs");
 832
 833   exit (GOOD);
 834 }
 835
 836 static void
 837 print_help (argbuffer)
 838      argument *argbuffer;
 839 {
 840   bool help_for_lang = FALSE;
 841
 842   for (; argbuffer->arg_type != at_end; argbuffer++)
 843     if (argbuffer->arg_type == at_language)
 844       {
 845         if (help_for_lang)
 846           puts ("");
 847         puts (argbuffer->lang->help);
 848         help_for_lang = TRUE;
 849       }
 850
 851   if (help_for_lang)
 852     exit (GOOD);
 853
 854   printf ("Usage: %s [options] [[regex-option ...] file-name] ...\n\
 855 \n\
 856 These are the options accepted by %s.\n", progname, progname);
 857 #ifdef LONG_OPTIONS
 858   puts ("You may use unambiguous abbreviations for the long option names.");
 859 #else
 860   puts ("Long option names do not work with this executable, as it is not\n\
 861 linked with GNU getopt.");
 862 #endif /* LONG_OPTIONS */
 863   puts ("  A - as file name means read names from stdin (one per line).\n\
 864 Absolute names are stored in the output file as they are.\n\
 865 Relative ones are stored relative to the output file's directory.\n");
 866
 867   if (!CTAGS)
 868     puts ("-a, --append\n\
 869         Append tag entries to existing tags file.");
 870
 871   puts ("--packages-only\n\
 872         For Ada files, only generate tags for packages.");
 873
 874   if (CTAGS)
 875     puts ("-B, --backward-search\n\
 876         Write the search commands for the tag entries using '?', the\n\
 877         backward-search command instead of '/', the forward-search command.");
 878
 879   /* This option is mostly obsolete, because etags can now automatically
 880      detect C++.  Retained for backward compatibility and for debugging and
 881      experimentation.  In principle, we could want to tag as C++ even
 882      before any "class" or "template" keyword.
 883   puts ("-C, --c++\n\
 884         Treat files whose name suffix defaults to C language as C++ files.");
 885   */
 886
 887   puts ("--declarations\n\
 888         In C and derived languages, create tags for function declarations,");
 889   if (CTAGS)
 890     puts ("\tand create tags for extern variables if --globals is used.");
 891   else
 892     puts
 893       ("\tand create tags for extern variables unless --no-globals is used.");
 894
 895   if (CTAGS)
 896     puts ("-d, --defines\n\
 897         Create tag entries for C #define constants and enum constants, too.");
 898   else
 899     puts ("-D, --no-defines\n\
 900         Don't create tag entries for C #define constants and enum constants.\n\
 901         This makes the tags file smaller.");
 902
 903   if (!CTAGS)
 904     puts ("-i FILE, --include=FILE\n\
 905         Include a note in tag file indicating that, when searching for\n\
 906         a tag, one should also consult the tags file FILE after\n\
 907         checking the current file.");
 908
 909   puts ("-l LANG, --language=LANG\n\
 910         Force the following files to be considered as written in the\n\
 911         named language up to the next --language=LANG option.");
 912
 913   if (CTAGS)
 914     puts ("--globals\n\
 915         Create tag entries for global variables in some languages.");
 916   else
 917     puts ("--no-globals\n\
 918         Do not create tag entries for global variables in some\n\
 919         languages.  This makes the tags file smaller.");
 920   puts ("--members\n\
 921         Create tag entries for members of structures in some languages.");
 922
 923 #ifdef ETAGS_REGEXPS
 924   puts ("-r REGEXP, --regex=REGEXP or --regex=@regexfile\n\
 925         Make a tag for each line matching a regular expression pattern\n\
 926         in the following files.  {LANGUAGE}REGEXP uses REGEXP for LANGUAGE\n\
 927         files only.  REGEXFILE is a file containing one REGEXP per line.\n\
 928         REGEXP takes the form /TAGREGEXP/TAGNAME/MODS, where TAGNAME/ is\n\
 929         optional.  The TAGREGEXP pattern is anchored (as if preceded by ^).");
 930   puts ("       If TAGNAME/ is present, the tags created are named.\n\
 931         For example Tcl named tags can be created with:\n\
 932           --regex=\"/proc[ \\t]+\\([^ \\t]+\\)/\\1/.\".\n\
 933         MODS are optional one-letter modifiers: `i' means to ignore case,\n\
 934         `m' means to allow multi-line matches, `s' implies `m' and\n\
 935         causes dot to match any character, including newline.");
 936   puts ("-R, --no-regex\n\
 937         Don't create tags from regexps for the following files.");
 938 #endif /* ETAGS_REGEXPS */
 939   puts ("-I, --ignore-indentation\n\
 940         In C and C++ do not assume that a closing brace in the first\n\
 941         column is the final brace of a function or structure definition.");
 942   puts ("-o FILE, --output=FILE\n\
 943         Write the tags to FILE.");
 944   puts ("--parse-stdin=NAME\n\
 945         Read from standard input and record tags as belonging to file NAME.");
 946
 947   if (CTAGS)
 948     {
 949       puts ("-t, --typedefs\n\
 950         Generate tag entries for C and Ada typedefs.");
 951       puts ("-T, --typedefs-and-c++\n\
 952         Generate tag entries for C typedefs, C struct/enum/union tags,\n\
 953         and C++ member functions.");
 954     }
 955
 956   if (CTAGS)
 957     puts ("-u, --update\n\
 958         Update the tag entries for the given files, leaving tag\n\
 959         entries for other files in place.  Currently, this is\n\
 960         implemented by deleting the existing entries for the given\n\
 961         files and then rewriting the new entries at the end of the\n\
 962         tags file.  It is often faster to simply rebuild the entire\n\
 963         tag file than to use this.");
 964
 965   if (CTAGS)
 966     {
 967       puts ("-v, --vgrind\n\
 968         Generates an index of items intended for human consumption,\n\
 969         similar to the output of vgrind.  The index is sorted, and\n\
 970         gives the page number of each item.");
 971       puts ("-w, --no-warn\n\
 972         Suppress warning messages about entries defined in multiple\n\
 973         files.");
 974       puts ("-x, --cxref\n\
 975         Like --vgrind, but in the style of cxref, rather than vgrind.\n\
 976         The output uses line numbers instead of page numbers, but\n\
 977         beyond that the differences are cosmetic; try both to see\n\
 978         which you like.");
 979     }
 980
 981   puts ("-V, --version\n\
 982         Print the version of the program.\n\
 983 -h, --help\n\
 984         Print this help message.\n\
 985         Followed by one or more `--language' options prints detailed\n\
 986         help about tag generation for the specified languages.");
 987
 988   print_language_names ();
 989
 990   puts ("");
 991   puts ("Report bugs to bug-gnu-emacs@gnu.org");
 992
 993   exit (GOOD);
 994 }
 995
 996 \f
 997 #ifdef VMS                      /* VMS specific functions */
 998
 999 #define EOS     '\0'
1000
1001 /* This is a BUG!  ANY arbitrary limit is a BUG!
1002    Won't someone please fix this?  */
1003 #define MAX_FILE_SPEC_LEN       255
1004 typedef struct  {
1005   short   curlen;
1006   char    body[MAX_FILE_SPEC_LEN + 1];
1007 } vspec;
1008
1009 /*
1010  v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names
1011  returning in each successive call the next file name matching the input
1012  spec. The function expects that each in_spec passed
1013  to it will be processed to completion; in particular, up to and
1014  including the call following that in which the last matching name
1015  is returned, the function ignores the value of in_spec, and will
1016  only start processing a new spec with the following call.
1017  If an error occurs, on return out_spec contains the value
1018  of in_spec when the error occurred.
1019
1020  With each successive file name returned in out_spec, the
1021  function's return value is one. When there are no more matching
1022  names the function returns zero. If on the first call no file
1023  matches in_spec, or there is any other error, -1 is returned.
1024 */
1025
1026 #include        <rmsdef.h>
1027 #include        <descrip.h>
1028 #define         OUTSIZE MAX_FILE_SPEC_LEN
1029 static short
1030 fn_exp (out, in)
1031      vspec *out;
1032      char *in;
1033 {
1034   static long context = 0;
1035   static struct dsc$descriptor_s o;
1036   static struct dsc$descriptor_s i;
1037   static bool pass1 = TRUE;
1038   long status;
1039   short retval;
1040
1041   if (pass1)
1042     {
1043       pass1 = FALSE;
1044       o.dsc$a_pointer = (char *) out;
1045       o.dsc$w_length = (short)OUTSIZE;
1046       i.dsc$a_pointer = in;
1047       i.dsc$w_length = (short)strlen(in);
1048       i.dsc$b_dtype = DSC$K_DTYPE_T;
1049       i.dsc$b_class = DSC$K_CLASS_S;
1050       o.dsc$b_dtype = DSC$K_DTYPE_VT;
1051       o.dsc$b_class = DSC$K_CLASS_VS;
1052     }
1053   if ((status = lib$find_file(&i, &o, &context, 0, 0)) == RMS$_NORMAL)
1054     {
1055       out->body[out->curlen] = EOS;
1056       return 1;
1057     }
1058   else if (status == RMS$_NMF)
1059     retval = 0;
1060   else
1061     {
1062       strcpy(out->body, in);
1063       retval = -1;
1064     }
1065   lib$find_file_end(&context);
1066   pass1 = TRUE;
1067   return retval;
1068 }
1069
1070 /*
1071   v1.01 nmm 19-Aug-85 gfnames - return in successive calls the
1072   name of each file specified by the provided arg expanding wildcards.
1073 */
1074 static char *
1075 gfnames (arg, p_error)
1076      char *arg;
1077      bool *p_error;
1078 {
1079   static vspec filename = {MAX_FILE_SPEC_LEN, "\0"};
1080
1081   switch (fn_exp (&filename, arg))
1082     {
1083     case 1:
1084       *p_error = FALSE;
1085       return filename.body;
1086     case 0:
1087       *p_error = FALSE;
1088       return NULL;
1089     default:
1090       *p_error = TRUE;
1091       return filename.body;
1092     }
1093 }
1094
1095 #ifndef OLD  /* Newer versions of VMS do provide `system'.  */
1096 system (cmd)
1097      char *cmd;
1098 {
1099   error ("%s", "system() function not implemented under VMS");
1100 }
1101 #endif
1102
1103 #define VERSION_DELIM   ';'
1104 char *massage_name (s)
1105      char *s;
1106 {
1107   char *start = s;
1108
1109   for ( ; *s; s++)
1110     if (*s == VERSION_DELIM)
1111       {
1112         *s = EOS;
1113         break;
1114       }
1115     else
1116       *s = lowcase (*s);
1117   return start;
1118 }
1119 #endif /* VMS */
1120
1121 \f
1122 int
1123 main (argc, argv)
1124      int argc;
1125      char *argv[];
1126 {
1127   int i;
1128   unsigned int nincluded_files;
1129   char **included_files;
1130   argument *argbuffer;
1131   int current_arg, file_count;
1132   linebuffer filename_lb;
1133   bool help_asked = FALSE;
1134 #ifdef VMS
1135   bool got_err;
1136 #endif
1137  char *optstring;
1138  int opt;
1139
1140
1141 #ifdef DOS_NT
1142   _fmode = O_BINARY;   /* all of files are treated as binary files */
1143 #endif /* DOS_NT */
1144
1145   progname = argv[0];
1146   nincluded_files = 0;
1147   included_files = xnew (argc, char *);
1148   current_arg = 0;
1149   file_count = 0;
1150
1151   /* Allocate enough no matter what happens.  Overkill, but each one
1152      is small. */
1153   argbuffer = xnew (argc, argument);
1154
1155   /*
1156    * If etags, always find typedefs and structure tags.  Why not?
1157    * Also default to find macro constants, enum constants and
1158    * global variables.
1159    */
1160   if (!CTAGS)
1161     {
1162       typedefs = typedefs_or_cplusplus = constantypedefs = TRUE;
1163       globals = TRUE;
1164     }
1165
1166   optstring = "-";
1167 #ifdef ETAGS_REGEXPS
1168   optstring = "-r:Rc:";
1169 #endif /* ETAGS_REGEXPS */
1170 #ifndef LONG_OPTIONS
1171   optstring = optstring + 1;
1172 #endif /* LONG_OPTIONS */
1173   optstring = concat (optstring,
1174                       "Cf:Il:o:SVhH",
1175                       (CTAGS) ? "BxdtTuvw" : "aDi:");
1176
1177   while ((opt = getopt_long (argc, argv, optstring, longopts, 0)) != EOF)
1178     switch (opt)
1179       {
1180       case 0:
1181         /* If getopt returns 0, then it has already processed a
1182            long-named option.  We should do nothing.  */
1183         break;
1184
1185       case 1:
1186         /* This means that a file name has been seen.  Record it. */
1187         argbuffer[current_arg].arg_type = at_filename;
1188         argbuffer[current_arg].what     = optarg;
1189         ++current_arg;
1190         ++file_count;
1191         break;
1192
1193       case STDIN:
1194         /* Parse standard input.  Idea by Vivek <vivek@etla.org>. */
1195         argbuffer[current_arg].arg_type = at_stdin;
1196         argbuffer[current_arg].what     = optarg;
1197         ++current_arg;
1198         ++file_count;
1199         if (parsing_stdin)
1200           fatal ("cannot parse standard input more than once", (char *)NULL);
1201         parsing_stdin = TRUE;
1202         break;
1203
1204         /* Common options. */
1205       case 'C': cplusplus = TRUE;               break;
1206       case 'f':         /* for compatibility with old makefiles */
1207       case 'o':
1208         if (tagfile)
1209           {
1210             error ("-o option may only be given once.", (char *)NULL);
1211             suggest_asking_for_help ();
1212             /* NOTREACHED */
1213           }
1214         tagfile = optarg;
1215         break;
1216       case 'I':
1217       case 'S':         /* for backward compatibility */
1218         ignoreindent = TRUE;
1219         break;
1220       case 'l':
1221         {
1222           language *lang = get_language_from_langname (optarg);
1223           if (lang != NULL)
1224             {
1225               argbuffer[current_arg].lang = lang;
1226               argbuffer[current_arg].arg_type = at_language;
1227               ++current_arg;
1228             }
1229         }
1230         break;
1231       case 'c':
1232         /* Backward compatibility: support obsolete --ignore-case-regexp. */
1233         optarg = concat (optarg, "i", ""); /* memory leak here */
1234         /* FALLTHRU */
1235       case 'r':
1236         argbuffer[current_arg].arg_type = at_regexp;
1237         argbuffer[current_arg].what = optarg;
1238         ++current_arg;
1239         break;
1240       case 'R':
1241         argbuffer[current_arg].arg_type = at_regexp;
1242         argbuffer[current_arg].what = NULL;
1243         ++current_arg;
1244         break;
1245       case 'V':
1246         print_version ();
1247         break;
1248       case 'h':
1249       case 'H':
1250         help_asked = TRUE;
1251         break;
1252
1253         /* Etags options */
1254       case 'a': append_to_tagfile = TRUE;                       break;
1255       case 'D': constantypedefs = FALSE;                        break;
1256       case 'i': included_files[nincluded_files++] = optarg;     break;
1257
1258         /* Ctags options. */
1259       case 'B': searchar = '?';                                 break;
1260       case 'd': constantypedefs = TRUE;                         break;
1261       case 't': typedefs = TRUE;                                break;
1262       case 'T': typedefs = typedefs_or_cplusplus = TRUE;        break;
1263       case 'u': update = TRUE;                                  break;
1264       case 'v': vgrind_style = TRUE;                      /*FALLTHRU*/
1265       case 'x': cxref_style = TRUE;                             break;
1266       case 'w': no_warnings = TRUE;                             break;
1267       default:
1268         suggest_asking_for_help ();
1269         /* NOTREACHED */
1270       }
1271
1272   for (; optind < argc; optind++)
1273     {
1274       argbuffer[current_arg].arg_type = at_filename;
1275       argbuffer[current_arg].what = argv[optind];
1276       ++current_arg;
1277       ++file_count;
1278     }
1279
1280   argbuffer[current_arg].arg_type = at_end;
1281
1282   if (help_asked)
1283     print_help (argbuffer);
1284     /* NOTREACHED */
1285
1286   if (nincluded_files == 0 && file_count == 0)
1287     {
1288       error ("no input files specified.", (char *)NULL);
1289       suggest_asking_for_help ();
1290       /* NOTREACHED */
1291     }
1292
1293   if (tagfile == NULL)
1294     tagfile = CTAGS ? "tags" : "TAGS";
1295   cwd = etags_getcwd ();        /* the current working directory */
1296   if (cwd[strlen (cwd) - 1] != '/')
1297     {
1298       char *oldcwd = cwd;
1299       cwd = concat (oldcwd, "/", "");
1300       free (oldcwd);
1301     }
1302   if (streq (tagfile, "-"))
1303     tagfiledir = cwd;
1304   else
1305     tagfiledir = absolute_dirname (tagfile, cwd);
1306
1307   init ();                      /* set up boolean "functions" */
1308
1309   linebuffer_init (&lb);
1310   linebuffer_init (&filename_lb);
1311   linebuffer_init (&filebuf);
1312   linebuffer_init (&token_name);
1313
1314   if (!CTAGS)
1315     {
1316       if (streq (tagfile, "-"))
1317         {
1318           tagf = stdout;
1319 #ifdef DOS_NT
1320           /* Switch redirected `stdout' to binary mode (setting `_fmode'
1321              doesn't take effect until after `stdout' is already open). */
1322           if (!isatty (fileno (stdout)))
1323             setmode (fileno (stdout), O_BINARY);
1324 #endif /* DOS_NT */
1325         }
1326       else
1327         tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1328       if (tagf == NULL)
1329         pfatal (tagfile);
1330     }
1331
1332   /*
1333    * Loop through files finding functions.
1334    */
1335   for (i = 0; i < current_arg; i++)
1336     {
1337       static language *lang;    /* non-NULL if language is forced */
1338       char *this_file;
1339
1340       switch (argbuffer[i].arg_type)
1341         {
1342         case at_language:
1343           lang = argbuffer[i].lang;
1344           break;
1345 #ifdef ETAGS_REGEXPS
1346         case at_regexp:
1347           analyse_regex (argbuffer[i].what);
1348           break;
1349 #endif
1350         case at_filename:
1351 #ifdef VMS
1352           while ((this_file = gfnames (argbuffer[i].what, &got_err)) != NULL)
1353             {
1354               if (got_err)
1355                 {
1356                   error ("can't find file %s\n", this_file);
1357                   argc--, argv++;
1358                 }
1359               else
1360                 {
1361                   this_file = massage_name (this_file);
1362                 }
1363 #else
1364               this_file = argbuffer[i].what;
1365 #endif
1366               /* Input file named "-" means read file names from stdin
1367                  (one per line) and use them. */
1368               if (streq (this_file, "-"))
1369                 {
1370                   if (parsing_stdin)
1371                     fatal ("cannot parse standard input AND read file names from it",
1372                            (char *)NULL);
1373                   while (readline_internal (&filename_lb, stdin) > 0)
1374                     process_file_name (filename_lb.buffer, lang);
1375                 }
1376               else
1377                 process_file_name (this_file, lang);
1378 #ifdef VMS
1379             }
1380 #endif
1381           break;
1382         case at_stdin:
1383           this_file = argbuffer[i].what;
1384           process_file (stdin, this_file, lang);
1385           break;
1386         }
1387     }
1388
1389 #ifdef ETAGS_REGEXPS
1390   free_regexps ();
1391 #endif /* ETAGS_REGEXPS */
1392   free (lb.buffer);
1393   free (filebuf.buffer);
1394   free (token_name.buffer);
1395
1396   if (!CTAGS || cxref_style)
1397     {
1398       put_entries (nodehead);   /* write the remainig tags (ETAGS) */
1399       free_tree (nodehead);
1400       nodehead = NULL;
1401       if (!CTAGS)
1402         {
1403           fdesc *fdp;
1404
1405           /* Output file entries that have no tags. */
1406           for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
1407             if (!fdp->written)
1408               fprintf (tagf, "\f\n%s,0\n", fdp->taggedfname);
1409
1410           while (nincluded_files-- > 0)
1411             fprintf (tagf, "\f\n%s,include\n", *included_files++);
1412         }
1413
1414       if (fclose (tagf) == EOF)
1415         pfatal (tagfile);
1416       exit (GOOD);
1417     }
1418
1419   if (update)
1420     {
1421       char cmd[BUFSIZ];
1422       for (i = 0; i < current_arg; ++i)
1423         {
1424           switch (argbuffer[i].arg_type)
1425             {
1426             case at_filename:
1427             case at_stdin:
1428               break;
1429             default:
1430               continue;         /* the for loop */
1431             }
1432           sprintf (cmd,
1433                    "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS",
1434                    tagfile, argbuffer[i].what, tagfile);
1435           if (system (cmd) != GOOD)
1436             fatal ("failed to execute shell command", (char *)NULL);
1437         }
1438       append_to_tagfile = TRUE;
1439     }
1440
1441   tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1442   if (tagf == NULL)
1443     pfatal (tagfile);
1444   put_entries (nodehead);       /* write all the tags (CTAGS) */
1445   free_tree (nodehead);
1446   nodehead = NULL;
1447   if (fclose (tagf) == EOF)
1448     pfatal (tagfile);
1449
1450   if (update)
1451     {
1452       char cmd[2*BUFSIZ+10];
1453       sprintf (cmd, "sort -o %.*s %.*s", BUFSIZ, tagfile, BUFSIZ, tagfile);
1454       exit (system (cmd));
1455     }
1456   return GOOD;
1457 }
1458
1459
1460 /*
1461  * Return a compressor given the file name.  If EXTPTR is non-zero,
1462  * return a pointer into FILE where the compressor-specific
1463  * extension begins.  If no compressor is found, NULL is returned
1464  * and EXTPTR is not significant.
1465  * Idea by Vladimir Alexiev <vladimir@cs.ualberta.ca> (1998)
1466  */
1467 static compressor *
1468 get_compressor_from_suffix (file, extptr)
1469      char *file;
1470      char **extptr;
1471 {
1472   compressor *compr;
1473   char *slash, *suffix;
1474
1475   /* This relies on FN to be after canonicalize_filename,
1476      so we don't need to consider backslashes on DOS_NT.  */
1477   slash = etags_strrchr (file, '/');
1478   suffix = etags_strrchr (file, '.');
1479   if (suffix == NULL || suffix < slash)
1480     return NULL;
1481   if (extptr != NULL)
1482     *extptr = suffix;
1483   suffix += 1;
1484   /* Let those poor souls who live with DOS 8+3 file name limits get
1485      some solace by treating foo.cgz as if it were foo.c.gz, etc.
1486      Only the first do loop is run if not MSDOS */
1487   do
1488     {
1489       for (compr = compressors; compr->suffix != NULL; compr++)
1490         if (streq (compr->suffix, suffix))
1491           return compr;
1492       if (!MSDOS)
1493         break;                  /* do it only once: not really a loop */
1494       if (extptr != NULL)
1495         *extptr = ++suffix;
1496     } while (*suffix != '\0');
1497   return NULL;
1498 }
1499
1500
1501
1502 /*
1503  * Return a language given the name.
1504  */
1505 static language *
1506 get_language_from_langname (name)
1507      const char *name;
1508 {
1509   language *lang;
1510
1511   if (name == NULL)
1512     error ("empty language name", (char *)NULL);
1513   else
1514     {
1515       for (lang = lang_names; lang->name != NULL; lang++)
1516         if (streq (name, lang->name))
1517           return lang;
1518       error ("unknown language \"%s\"", name);
1519     }
1520
1521   return NULL;
1522 }
1523
1524
1525 /*
1526  * Return a language given the interpreter name.
1527  */
1528 static language *
1529 get_language_from_interpreter (interpreter)
1530      char *interpreter;
1531 {
1532   language *lang;
1533   char **iname;
1534
1535   if (interpreter == NULL)
1536     return NULL;
1537   for (lang = lang_names; lang->name != NULL; lang++)
1538     if (lang->interpreters != NULL)
1539       for (iname = lang->interpreters; *iname != NULL; iname++)
1540         if (streq (*iname, interpreter))
1541             return lang;
1542
1543   return NULL;
1544 }
1545
1546
1547
1548 /*
1549  * Return a language given the file name.
1550  */
1551 static language *
1552 get_language_from_filename (file, case_sensitive)
1553      char *file;
1554      bool case_sensitive;
1555 {
1556   language *lang;
1557   char **name, **ext, *suffix;
1558
1559   /* Try whole file name first. */
1560   for (lang = lang_names; lang->name != NULL; lang++)
1561     if (lang->filenames != NULL)
1562       for (name = lang->filenames; *name != NULL; name++)
1563         if ((case_sensitive)
1564             ? streq (*name, file)
1565             : strcaseeq (*name, file))
1566           return lang;
1567
1568   /* If not found, try suffix after last dot. */
1569   suffix = etags_strrchr (file, '.');
1570   if (suffix == NULL)
1571     return NULL;
1572   suffix += 1;
1573   for (lang = lang_names; lang->name != NULL; lang++)
1574     if (lang->suffixes != NULL)
1575       for (ext = lang->suffixes; *ext != NULL; ext++)
1576         if ((case_sensitive)
1577             ? streq (*ext, suffix)
1578             : strcaseeq (*ext, suffix))
1579           return lang;
1580   return NULL;
1581 }
1582
1583 \f
1584 /*
1585  * This routine is called on each file argument.
1586  */
1587 static void
1588 process_file_name (file, lang)
1589      char *file;
1590      language *lang;
1591 {
1592   struct stat stat_buf;
1593   FILE *inf;
1594   fdesc *fdp;
1595   compressor *compr;
1596   char *compressed_name, *uncompressed_name;
1597   char *ext, *real_name;
1598   int retval;
1599
1600   canonicalize_filename (file);
1601   if (streq (file, tagfile) && !streq (tagfile, "-"))
1602     {
1603       error ("skipping inclusion of %s in self.", file);
1604       return;
1605     }
1606   if ((compr = get_compressor_from_suffix (file, &ext)) == NULL)
1607     {
1608       compressed_name = NULL;
1609       real_name = uncompressed_name = savestr (file);
1610     }
1611   else
1612     {
1613       real_name = compressed_name = savestr (file);
1614       uncompressed_name = savenstr (file, ext - file);
1615     }
1616
1617   /* If the canonicalized uncompressed name
1618      has already been dealt with, skip it silently. */
1619   for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
1620     {
1621       assert (fdp->infname != NULL);
1622       if (streq (uncompressed_name, fdp->infname))
1623         goto cleanup;
1624     }
1625
1626   if (stat (real_name, &stat_buf) != 0)
1627     {
1628       /* Reset real_name and try with a different name. */
1629       real_name = NULL;
1630       if (compressed_name != NULL) /* try with the given suffix */
1631         {
1632           if (stat (uncompressed_name, &stat_buf) == 0)
1633             real_name = uncompressed_name;
1634         }
1635       else                      /* try all possible suffixes */
1636         {
1637           for (compr = compressors; compr->suffix != NULL; compr++)
1638             {
1639               compressed_name = concat (file, ".", compr->suffix);
1640               if (stat (compressed_name, &stat_buf) != 0)
1641                 {
1642                   if (MSDOS)
1643                     {
1644                       char *suf = compressed_name + strlen (file);
1645                       size_t suflen = strlen (compr->suffix) + 1;
1646                       for ( ; suf[1]; suf++, suflen--)
1647                         {
1648                           memmove (suf, suf + 1, suflen);
1649                           if (stat (compressed_name, &stat_buf) == 0)
1650                             {
1651                               real_name = compressed_name;
1652                               break;
1653                             }
1654                         }
1655                       if (real_name != NULL)
1656                         break;
1657                     } /* MSDOS */
1658                   free (compressed_name);
1659                   compressed_name = NULL;
1660                 }
1661               else
1662                 {
1663                   real_name = compressed_name;
1664                   break;
1665                 }
1666             }
1667         }
1668       if (real_name == NULL)
1669         {
1670           perror (file);
1671           goto cleanup;
1672         }
1673     } /* try with a different name */
1674
1675   if (!S_ISREG (stat_buf.st_mode))
1676     {
1677       error ("skipping %s: it is not a regular file.", real_name);
1678       goto cleanup;
1679     }
1680   if (real_name == compressed_name)
1681     {
1682       char *cmd = concat (compr->command, " ", real_name);
1683       inf = (FILE *) popen (cmd, "r");
1684       free (cmd);
1685     }
1686   else
1687     inf = fopen (real_name, "r");
1688   if (inf == NULL)
1689     {
1690       perror (real_name);
1691       goto cleanup;
1692     }
1693
1694   process_file (inf, uncompressed_name, lang);
1695
1696   if (real_name == compressed_name)
1697     retval = pclose (inf);
1698   else
1699     retval = fclose (inf);
1700   if (retval < 0)
1701     pfatal (file);
1702
1703  cleanup:
1704   if (compressed_name) free (compressed_name);
1705   if (uncompressed_name) free (uncompressed_name);
1706   last_node = NULL;
1707   curfdp = NULL;
1708   return;
1709 }
1710
1711 static void
1712 process_file (fh, fn, lang)
1713      FILE *fh;
1714      char *fn;
1715      language *lang;
1716 {
1717   static const fdesc emptyfdesc;
1718   fdesc *fdp;
1719
1720   /* Create a new input file description entry. */
1721   fdp = xnew (1, fdesc);
1722   *fdp = emptyfdesc;
1723   fdp->next = fdhead;
1724   fdp->infname = savestr (fn);
1725   fdp->lang = lang;
1726   fdp->infabsname = absolute_filename (fn, cwd);
1727   fdp->infabsdir = absolute_dirname (fn, cwd);
1728   if (filename_is_absolute (fn))
1729     {
1730       /* An absolute file name.  Canonicalize it. */
1731       fdp->taggedfname = absolute_filename (fn, NULL);
1732     }
1733   else
1734     {
1735       /* A file name relative to cwd.  Make it relative
1736          to the directory of the tags file. */
1737       fdp->taggedfname = relative_filename (fn, tagfiledir);
1738     }
1739   fdp->usecharno = TRUE;        /* use char position when making tags */
1740   fdp->prop = NULL;
1741   fdp->written = FALSE;         /* not written on tags file yet */
1742
1743   fdhead = fdp;
1744   curfdp = fdhead;              /* the current file description */
1745
1746   find_entries (fh);
1747
1748   /* If not Ctags, and if this is not metasource and if it contained no #line
1749      directives, we can write the tags and free all nodes pointing to
1750      curfdp. */
1751   if (!CTAGS
1752       && curfdp->usecharno      /* no #line directives in this file */
1753       && !curfdp->lang->metasource)
1754     {
1755       node *np, *prev;
1756
1757       /* Look for the head of the sublist relative to this file.  See add_node
1758          for the structure of the node tree. */
1759       prev = NULL;
1760       for (np = nodehead; np != NULL; prev = np, np = np->left)
1761         if (np->fdp == curfdp)
1762           break;
1763
1764       /* If we generated tags for this file, write and delete them. */
1765       if (np != NULL)
1766         {
1767           /* This is the head of the last sublist, if any.  The following
1768              instructions depend on this being true. */
1769           assert (np->left == NULL);
1770
1771           assert (fdhead == curfdp);
1772           assert (last_node->fdp == curfdp);
1773           put_entries (np);     /* write tags for file curfdp->taggedfname */
1774           free_tree (np);       /* remove the written nodes */
1775           if (prev == NULL)
1776             nodehead = NULL;    /* no nodes left */
1777           else
1778             prev->left = NULL;  /* delete the pointer to the sublist */
1779         }
1780     }
1781 }
1782
1783 /*
1784  * This routine sets up the boolean pseudo-functions which work
1785  * by setting boolean flags dependent upon the corresponding character.
1786  * Every char which is NOT in that string is not a white char.  Therefore,
1787  * all of the array "_wht" is set to FALSE, and then the elements
1788  * subscripted by the chars in "white" are set to TRUE.  Thus "_wht"
1789  * of a char is TRUE if it is the string "white", else FALSE.
1790  */
1791 static void
1792 init ()
1793 {
1794   register char *sp;
1795   register int i;
1796
1797   for (i = 0; i < CHARS; i++)
1798     iswhite(i) = notinname(i) = begtoken(i) = intoken(i) = endtoken(i) = FALSE;
1799   for (sp = white; *sp != '\0'; sp++) iswhite (*sp) = TRUE;
1800   for (sp = nonam; *sp != '\0'; sp++) notinname (*sp) = TRUE;
1801   notinname('\0') = notinname('\n');
1802   for (sp = begtk; *sp != '\0'; sp++) begtoken (*sp) = TRUE;
1803   begtoken('\0') = begtoken('\n');
1804   for (sp = midtk; *sp != '\0'; sp++) intoken (*sp) = TRUE;
1805   intoken('\0') = intoken('\n');
1806   for (sp = endtk; *sp != '\0'; sp++) endtoken (*sp) = TRUE;
1807   endtoken('\0') = endtoken('\n');
1808 }
1809
1810 /*
1811  * This routine opens the specified file and calls the function
1812  * which finds the function and type definitions.
1813  */
1814 static void
1815 find_entries (inf)
1816      FILE *inf;
1817 {
1818   char *cp;
1819   language *lang = curfdp->lang;
1820   Lang_function *parser = NULL;
1821
1822   /* If user specified a language, use it. */
1823   if (lang != NULL && lang->function != NULL)
1824     {
1825       parser = lang->function;
1826     }
1827
1828   /* Else try to guess the language given the file name. */
1829   if (parser == NULL)
1830     {
1831       lang = get_language_from_filename (curfdp->infname, TRUE);
1832       if (lang != NULL && lang->function != NULL)
1833         {
1834           curfdp->lang = lang;
1835           parser = lang->function;
1836         }
1837     }
1838
1839   /* Else look for sharp-bang as the first two characters. */
1840   if (parser == NULL
1841       && readline_internal (&lb, inf) > 0
1842       && lb.len >= 2
1843       && lb.buffer[0] == '#'
1844       && lb.buffer[1] == '!')
1845     {
1846       char *lp;
1847
1848       /* Set lp to point at the first char after the last slash in the
1849          line or, if no slashes, at the first nonblank.  Then set cp to
1850          the first successive blank and terminate the string. */
1851       lp = etags_strrchr (lb.buffer+2, '/');
1852       if (lp != NULL)
1853         lp += 1;
1854       else
1855         lp = skip_spaces (lb.buffer + 2);
1856       cp = skip_non_spaces (lp);
1857       *cp = '\0';
1858
1859       if (strlen (lp) > 0)
1860         {
1861           lang = get_language_from_interpreter (lp);
1862           if (lang != NULL && lang->function != NULL)
1863             {
1864               curfdp->lang = lang;
1865               parser = lang->function;
1866             }
1867         }
1868     }
1869
1870   /* We rewind here, even if inf may be a pipe.  We fail if the
1871      length of the first line is longer than the pipe block size,
1872      which is unlikely. */
1873   rewind (inf);
1874
1875   /* Else try to guess the language given the case insensitive file name. */
1876   if (parser == NULL)
1877     {
1878       lang = get_language_from_filename (curfdp->infname, FALSE);
1879       if (lang != NULL && lang->function != NULL)
1880         {
1881           curfdp->lang = lang;
1882           parser = lang->function;
1883         }
1884     }
1885
1886   /* Else try Fortran or C. */
1887   if (parser == NULL)
1888     {
1889       node *old_last_node = last_node;
1890
1891       curfdp->lang = get_language_from_langname ("fortran");
1892       find_entries (inf);
1893
1894       if (old_last_node == last_node)
1895         /* No Fortran entries found.  Try C. */
1896         {
1897           /* We do not tag if rewind fails.
1898              Only the file name will be recorded in the tags file. */
1899           rewind (inf);
1900           curfdp->lang = get_language_from_langname (cplusplus ? "c++" : "c");
1901           find_entries (inf);
1902         }
1903       return;
1904     }
1905
1906   if (!no_line_directive
1907       && curfdp->lang != NULL && curfdp->lang->metasource)
1908     /* It may be that this is a bingo.y file, and we already parsed a bingo.c
1909        file, or anyway we parsed a file that is automatically generated from
1910        this one.  If this is the case, the bingo.c file contained #line
1911        directives that generated tags pointing to this file.  Let's delete
1912        them all before parsing this file, which is the real source. */
1913     {
1914       fdesc **fdpp = &fdhead;
1915       while (*fdpp != NULL)
1916         if (*fdpp != curfdp
1917             && streq ((*fdpp)->taggedfname, curfdp->taggedfname))
1918           /* We found one of those!  We must delete both the file description
1919              and all tags referring to it. */
1920           {
1921             fdesc *badfdp = *fdpp;
1922
1923             /* Delete the tags referring to badfdp->taggedfname
1924                that were obtained from badfdp->infname. */
1925             invalidate_nodes (badfdp, &nodehead);
1926
1927             *fdpp = badfdp->next; /* remove the bad description from the list */
1928             free_fdesc (badfdp);
1929           }
1930         else
1931           fdpp = &(*fdpp)->next; /* advance the list pointer */
1932     }
1933
1934   assert (parser != NULL);
1935
1936   /* Generic initialisations before reading from file. */
1937   linebuffer_setlen (&filebuf, 0); /* reset the file buffer */
1938
1939   /* Generic initialisations before parsing file with readline. */
1940   lineno = 0;                  /* reset global line number */
1941   charno = 0;                  /* reset global char number */
1942   linecharno = 0;              /* reset global char number of line start */
1943
1944   parser (inf);
1945
1946 #ifdef ETAGS_REGEXPS
1947   regex_tag_multiline ();
1948 #endif /* ETAGS_REGEXPS */
1949 }
1950
1951 \f
1952 /*
1953  * Check whether an implicitly named tag should be created,
1954  * then call `pfnote'.
1955  * NAME is a string that is internally copied by this function.
1956  *
1957  * TAGS format specification
1958  * Idea by Sam Kendall <kendall@mv.mv.com> (1997)
1959  * The following is explained in some more detail in etc/ETAGS.EBNF.
1960  *
1961  * make_tag creates tags with "implicit tag names" (unnamed tags)
1962  * if the following are all true, assuming NONAM=" \f\t\n\r()=,;":
1963  *  1. NAME does not contain any of the characters in NONAM;
1964  *  2. LINESTART contains name as either a rightmost, or rightmost but
1965  *     one character, substring;
1966  *  3. the character, if any, immediately before NAME in LINESTART must
1967  *     be a character in NONAM;
1968  *  4. the character, if any, immediately after NAME in LINESTART must
1969  *     also be a character in NONAM.
1970  *
1971  * The implementation uses the notinname() macro, which recognises the
1972  * characters stored in the string `nonam'.
1973  * etags.el needs to use the same characters that are in NONAM.
1974  */
1975 static void
1976 make_tag (name, namelen, is_func, linestart, linelen, lno, cno)
1977      char *name;                /* tag name, or NULL if unnamed */
1978      int namelen;               /* tag length */
1979      bool is_func;              /* tag is a function */
1980      char *linestart;           /* start of the line where tag is */
1981      int linelen;               /* length of the line where tag is */
1982      int lno;                   /* line number */
1983      long cno;                  /* character number */
1984 {
1985   bool named = (name != NULL && namelen > 0);
1986
1987   if (!CTAGS && named)          /* maybe set named to false */
1988     /* Let's try to make an implicit tag name, that is, create an unnamed tag
1989        such that etags.el can guess a name from it. */
1990     {
1991       int i;
1992       register char *cp = name;
1993
1994       for (i = 0; i < namelen; i++)
1995         if (notinname (*cp++))
1996           break;
1997       if (i == namelen)                         /* rule #1 */
1998         {
1999           cp = linestart + linelen - namelen;
2000           if (notinname (linestart[linelen-1]))
2001             cp -= 1;                            /* rule #4 */
2002           if (cp >= linestart                   /* rule #2 */
2003               && (cp == linestart
2004                   || notinname (cp[-1]))        /* rule #3 */
2005               && strneq (name, cp, namelen))    /* rule #2 */
2006             named = FALSE;      /* use implicit tag name */
2007         }
2008     }
2009
2010   if (named)
2011     name = savenstr (name, namelen);
2012   else
2013     name = NULL;
2014   pfnote (name, is_func, linestart, linelen, lno, cno);
2015 }
2016
2017 /* Record a tag. */
2018 static void
2019 pfnote (name, is_func, linestart, linelen, lno, cno)
2020      char *name;                /* tag name, or NULL if unnamed */
2021      bool is_func;              /* tag is a function */
2022      char *linestart;           /* start of the line where tag is */
2023      int linelen;               /* length of the line where tag is */
2024      int lno;                   /* line number */
2025      long cno;                  /* character number */
2026 {
2027   register node *np;
2028
2029   assert (name == NULL || name[0] != '\0');
2030   if (CTAGS && name == NULL)
2031     return;
2032
2033   np = xnew (1, node);
2034
2035   /* If ctags mode, change name "main" to M<thisfilename>. */
2036   if (CTAGS && !cxref_style && streq (name, "main"))
2037     {
2038       register char *fp = etags_strrchr (curfdp->taggedfname, '/');
2039       np->name = concat ("M", fp == NULL ? curfdp->taggedfname : fp + 1, "");
2040       fp = etags_strrchr (np->name, '.');
2041       if (fp != NULL && fp[1] != '\0' && fp[2] == '\0')
2042         fp[0] = '\0';
2043     }
2044   else
2045     np->name = name;
2046   np->valid = TRUE;
2047   np->been_warned = FALSE;
2048   np->fdp = curfdp;
2049   np->is_func = is_func;
2050   np->lno = lno;
2051   if (np->fdp->usecharno)
2052     /* Our char numbers are 0-base, because of C language tradition?
2053        ctags compatibility?  old versions compatibility?   I don't know.
2054        Anyway, since emacs's are 1-base we expect etags.el to take care
2055        of the difference.  If we wanted to have 1-based numbers, we would
2056        uncomment the +1 below. */
2057     np->cno = cno /* + 1 */ ;
2058   else
2059     np->cno = invalidcharno;
2060   np->left = np->right = NULL;
2061   if (CTAGS && !cxref_style)
2062     {
2063       if (strlen (linestart) < 50)
2064         np->regex = concat (linestart, "$", "");
2065       else
2066         np->regex = savenstr (linestart, 50);
2067     }
2068   else
2069     np->regex = savenstr (linestart, linelen);
2070
2071   add_node (np, &nodehead);
2072 }
2073
2074 /*
2075  * free_tree ()
2076  *      recurse on left children, iterate on right children.
2077  */
2078 static void
2079 free_tree (np)
2080      register node *np;
2081 {
2082   while (np)
2083     {
2084       register node *node_right = np->right;
2085       free_tree (np->left);
2086       if (np->name != NULL)
2087         free (np->name);
2088       free (np->regex);
2089       free (np);
2090       np = node_right;
2091     }
2092 }
2093
2094 /*
2095  * free_fdesc ()
2096  *      delete a file description
2097  */
2098 static void
2099 free_fdesc (fdp)
2100      register fdesc *fdp;
2101 {
2102   if (fdp->infname != NULL) free (fdp->infname);
2103   if (fdp->infabsname != NULL) free (fdp->infabsname);
2104   if (fdp->infabsdir != NULL) free (fdp->infabsdir);
2105   if (fdp->taggedfname != NULL) free (fdp->taggedfname);
2106   if (fdp->prop != NULL) free (fdp->prop);
2107   free (fdp);
2108 }
2109
2110 /*
2111  * add_node ()
2112  *      Adds a node to the tree of nodes.  In etags mode, sort by file
2113  *      name.  In ctags mode, sort by tag name.  Make no attempt at
2114  *      balancing.
2115  *
2116  *      add_node is the only function allowed to add nodes, so it can
2117  *      maintain state.
2118  */
2119 static void
2120 add_node (np, cur_node_p)
2121      node *np, **cur_node_p;
2122 {
2123   register int dif;
2124   register node *cur_node = *cur_node_p;
2125
2126   if (cur_node == NULL)
2127     {
2128       *cur_node_p = np;
2129       last_node = np;
2130       return;
2131     }
2132
2133   if (!CTAGS)
2134     /* Etags Mode */
2135     {
2136       /* For each file name, tags are in a linked sublist on the right
2137          pointer.  The first tags of different files are a linked list
2138          on the left pointer.  last_node points to the end of the last
2139          used sublist. */
2140       if (last_node != NULL && last_node->fdp == np->fdp)
2141         {
2142           /* Let's use the same sublist as the last added node. */
2143           assert (last_node->right == NULL);
2144           last_node->right = np;
2145           last_node = np;
2146         }
2147       else if (cur_node->fdp == np->fdp)
2148         {
2149           /* Scanning the list we found the head of a sublist which is
2150              good for us.  Let's scan this sublist. */
2151           add_node (np, &cur_node->right);
2152         }
2153       else
2154         /* The head of this sublist is not good for us.  Let's try the
2155            next one. */
2156         add_node (np, &cur_node->left);
2157     } /* if ETAGS mode */
2158
2159   else
2160     {
2161       /* Ctags Mode */
2162       dif = strcmp (np->name, cur_node->name);
2163
2164       /*
2165        * If this tag name matches an existing one, then
2166        * do not add the node, but maybe print a warning.
2167        */
2168       if (!dif)
2169         {
2170           if (np->fdp == cur_node->fdp)
2171             {
2172               if (!no_warnings)
2173                 {
2174                   fprintf (stderr, "Duplicate entry in file %s, line %d: %s\n",
2175                            np->fdp->infname, lineno, np->name);
2176                   fprintf (stderr, "Second entry ignored\n");
2177                 }
2178             }
2179           else if (!cur_node->been_warned && !no_warnings)
2180             {
2181               fprintf
2182                 (stderr,
2183                  "Duplicate entry in files %s and %s: %s (Warning only)\n",
2184                  np->fdp->infname, cur_node->fdp->infname, np->name);
2185               cur_node->been_warned = TRUE;
2186             }
2187           return;
2188         }
2189
2190       /* Actually add the node */
2191       add_node (np, dif < 0 ? &cur_node->left : &cur_node->right);
2192     } /* if CTAGS mode */
2193 }
2194
2195 /*
2196  * invalidate_nodes ()
2197  *      Scan the node tree and invalidate all nodes pointing to the
2198  *      given file description (CTAGS case) or free them (ETAGS case).
2199  */
2200 static void
2201 invalidate_nodes (badfdp, npp)
2202      fdesc *badfdp;
2203      node **npp;
2204 {
2205   node *np = *npp;
2206
2207   if (np == NULL)
2208     return;
2209
2210   if (CTAGS)
2211     {
2212       if (np->left != NULL)
2213         invalidate_nodes (badfdp, &np->left);
2214       if (np->fdp == badfdp)
2215         np->valid = FALSE;
2216       if (np->right != NULL)
2217         invalidate_nodes (badfdp, &np->right);
2218     }
2219   else
2220     {
2221       assert (np->fdp != NULL);
2222       if (np->fdp == badfdp)
2223         {
2224           *npp = np->left;      /* detach the sublist from the list */
2225           np->left = NULL;      /* isolate it */
2226           free_tree (np);       /* free it */
2227           invalidate_nodes (badfdp, npp);
2228         }
2229       else
2230         invalidate_nodes (badfdp, &np->left);
2231     }
2232 }
2233
2234 \f
2235 static int total_size_of_entries __P((node *));
2236 static int number_len __P((long));
2237
2238 /* Length of a non-negative number's decimal representation. */
2239 static int
2240 number_len (num)
2241      long num;
2242 {
2243   int len = 1;
2244   while ((num /= 10) > 0)
2245     len += 1;
2246   return len;
2247 }
2248
2249 /*
2250  * Return total number of characters that put_entries will output for
2251  * the nodes in the linked list at the right of the specified node.
2252  * This count is irrelevant with etags.el since emacs 19.34 at least,
2253  * but is still supplied for backward compatibility.
2254  */
2255 static int
2256 total_size_of_entries (np)
2257      register node *np;
2258 {
2259   register int total = 0;
2260
2261   for (; np != NULL; np = np->right)
2262     if (np->valid)
2263       {
2264         total += strlen (np->regex) + 1;                /* pat\177 */
2265         if (np->name != NULL)
2266           total += strlen (np->name) + 1;               /* name\001 */
2267         total += number_len ((long) np->lno) + 1;       /* lno, */
2268         if (np->cno != invalidcharno)                   /* cno */
2269           total += number_len (np->cno);
2270         total += 1;                                     /* newline */
2271       }
2272
2273   return total;
2274 }
2275
2276 static void
2277 put_entries (np)
2278      register node *np;
2279 {
2280   register char *sp;
2281   static fdesc *fdp = NULL;
2282
2283   if (np == NULL)
2284     return;
2285
2286   /* Output subentries that precede this one */
2287   if (CTAGS)
2288     put_entries (np->left);
2289
2290   /* Output this entry */
2291   if (np->valid)
2292     {
2293       if (!CTAGS)
2294         {
2295           /* Etags mode */
2296           if (fdp != np->fdp)
2297             {
2298               fdp = np->fdp;
2299               fprintf (tagf, "\f\n%s,%d\n",
2300                        fdp->taggedfname, total_size_of_entries (np));
2301               fdp->written = TRUE;
2302             }
2303           fputs (np->regex, tagf);
2304           fputc ('\177', tagf);
2305           if (np->name != NULL)
2306             {
2307               fputs (np->name, tagf);
2308               fputc ('\001', tagf);
2309             }
2310           fprintf (tagf, "%d,", np->lno);
2311           if (np->cno != invalidcharno)
2312             fprintf (tagf, "%ld", np->cno);
2313           fputs ("\n", tagf);
2314         }
2315       else
2316         {
2317           /* Ctags mode */
2318           if (np->name == NULL)
2319             error ("internal error: NULL name in ctags mode.", (char *)NULL);
2320
2321           if (cxref_style)
2322             {
2323               if (vgrind_style)
2324                 fprintf (stdout, "%s %s %d\n",
2325                          np->name, np->fdp->taggedfname, (np->lno + 63) / 64);
2326               else
2327                 fprintf (stdout, "%-16s %3d %-16s %s\n",
2328                          np->name, np->lno, np->fdp->taggedfname, np->regex);
2329             }
2330           else
2331             {
2332               fprintf (tagf, "%s\t%s\t", np->name, np->fdp->taggedfname);
2333
2334               if (np->is_func)
2335                 {               /* function or #define macro with args */
2336                   putc (searchar, tagf);
2337                   putc ('^', tagf);
2338
2339                   for (sp = np->regex; *sp; sp++)
2340                     {
2341                       if (*sp == '\\' || *sp == searchar)
2342                         putc ('\\', tagf);
2343                       putc (*sp, tagf);
2344                     }
2345                   putc (searchar, tagf);
2346                 }
2347               else
2348                 {               /* anything else; text pattern inadequate */
2349                   fprintf (tagf, "%d", np->lno);
2350                 }
2351               putc ('\n', tagf);
2352             }
2353         }
2354     } /* if this node contains a valid tag */
2355
2356   /* Output subentries that follow this one */
2357   put_entries (np->right);
2358   if (!CTAGS)
2359     put_entries (np->left);
2360 }
2361
2362 \f
2363 /* C extensions. */
2364 #define C_EXT   0x00fff         /* C extensions */
2365 #define C_PLAIN 0x00000         /* C */
2366 #define C_PLPL  0x00001         /* C++ */
2367 #define C_STAR  0x00003         /* C* */
2368 #define C_JAVA  0x00005         /* JAVA */
2369 #define C_AUTO  0x01000         /* C, but switch to C++ if `class' is met */
2370 #define YACC    0x10000         /* yacc file */
2371
2372 /*
2373  * The C symbol tables.
2374  */
2375 enum sym_type
2376 {
2377   st_none,
2378   st_C_objprot, st_C_objimpl, st_C_objend,
2379   st_C_gnumacro,
2380   st_C_ignore,
2381   st_C_javastruct,
2382   st_C_operator,
2383   st_C_class, st_C_template,
2384   st_C_struct, st_C_extern, st_C_enum, st_C_define, st_C_typedef, st_C_typespec
2385 };
2386
2387 static unsigned int hash __P((const char *, unsigned int));
2388 static struct C_stab_entry * in_word_set __P((const char *, unsigned int));
2389 static enum sym_type C_symtype __P((char *, int, int));
2390
2391 /* Feed stuff between (but not including) %[ and %] lines to:
2392       gperf -c -k 1,3 -o -p -r -t
2393 %[
2394 struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
2395 %%
2396 if,             0,      st_C_ignore
2397 for,            0,      st_C_ignore
2398 while,          0,      st_C_ignore
2399 switch,         0,      st_C_ignore
2400 return,         0,      st_C_ignore
2401 @interface,     0,      st_C_objprot
2402 @protocol,      0,      st_C_objprot
2403 @implementation,0,      st_C_objimpl
2404 @end,           0,      st_C_objend
2405 import,         C_JAVA, st_C_ignore
2406 package,        C_JAVA, st_C_ignore
2407 friend,         C_PLPL, st_C_ignore
2408 extends,        C_JAVA, st_C_javastruct
2409 implements,     C_JAVA, st_C_javastruct
2410 interface,      C_JAVA, st_C_struct
2411 class,          0,      st_C_class
2412 namespace,      C_PLPL, st_C_struct
2413 domain,         C_STAR, st_C_struct
2414 union,          0,      st_C_struct
2415 struct,         0,      st_C_struct
2416 extern,         0,      st_C_extern
2417 enum,           0,      st_C_enum
2418 typedef,        0,      st_C_typedef
2419 define,         0,      st_C_define
2420 operator,       C_PLPL, st_C_operator
2421 template,       0,      st_C_template
2422 bool,           C_PLPL, st_C_typespec
2423 long,           0,      st_C_typespec
2424 short,          0,      st_C_typespec
2425 int,            0,      st_C_typespec
2426 char,           0,      st_C_typespec
2427 float,          0,      st_C_typespec
2428 double,         0,      st_C_typespec
2429 signed,         0,      st_C_typespec
2430 unsigned,       0,      st_C_typespec
2431 auto,           0,      st_C_typespec
2432 void,           0,      st_C_typespec
2433 static,         0,      st_C_typespec
2434 const,          0,      st_C_typespec
2435 volatile,       0,      st_C_typespec
2436 explicit,       C_PLPL, st_C_typespec
2437 mutable,        C_PLPL, st_C_typespec
2438 typename,       C_PLPL, st_C_typespec
2439 # DEFUN used in emacs, the next three used in glibc (SYSCALL only for mach).
2440 DEFUN,          0,      st_C_gnumacro
2441 SYSCALL,        0,      st_C_gnumacro
2442 ENTRY,          0,      st_C_gnumacro
2443 PSEUDO,         0,      st_C_gnumacro
2444 # These are defined inside C functions, so currently they are not met.
2445 # EXFUN used in glibc, DEFVAR_* in emacs.
2446 #EXFUN,         0,      st_C_gnumacro
2447 #DEFVAR_,       0,      st_C_gnumacro
2448 %]
2449 and replace lines between %< and %> with its output,
2450 then make in_word_set and C_stab_entry static. */
2451 /*%<*/
2452 /* C code produced by gperf version 2.7.1 (19981006 egcs) */
2453 /* Command-line: gperf -c -k 1,3 -o -p -r -t  */
2454 struct C_stab_entry { char *name; int c_ext; enum sym_type type; };
2455
2456 #define TOTAL_KEYWORDS 47
2457 #define MIN_WORD_LENGTH 2
2458 #define MAX_WORD_LENGTH 15
2459 #define MIN_HASH_VALUE 18
2460 #define MAX_HASH_VALUE 138
2461 /* maximum key range = 121, duplicates = 0 */
2462
2463 #ifdef __GNUC__
2464 __inline
2465 #endif
2466 static unsigned int
2467 hash (str, len)
2468      register const char *str;
2469      register unsigned int len;
2470 {
2471   static unsigned char asso_values[] =
2472     {
2473       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2474       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2475       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2476       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2477       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2478       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2479       139, 139, 139, 139,  63, 139, 139, 139,  33,  44,
2480        62, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2481        42, 139, 139,  12,  32, 139, 139, 139, 139, 139,
2482       139, 139, 139, 139, 139, 139, 139,  34,  59,  37,
2483        24,  58,  33,   3, 139,  16, 139, 139,  42,  60,
2484        18,  11,  39, 139,  23,  57,   4,  63,   6,  20,
2485       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2486       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2487       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2488       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2489       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2490       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2491       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2492       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2493       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2494       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2495       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2496       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2497       139, 139, 139, 139, 139, 139, 139, 139, 139, 139,
2498       139, 139, 139, 139, 139, 139
2499     };
2500   register int hval = len;
2501
2502   switch (hval)
2503     {
2504       default:
2505       case 3:
2506         hval += asso_values[(unsigned char)str[2]];
2507       case 2:
2508       case 1:
2509         hval += asso_values[(unsigned char)str[0]];
2510         break;
2511     }
2512   return hval;
2513 }
2514
2515 #ifdef __GNUC__
2516 __inline
2517 #endif
2518 static struct C_stab_entry *
2519 in_word_set (str, len)
2520      register const char *str;
2521      register unsigned int len;
2522 {
2523   static struct C_stab_entry wordlist[] =
2524     {
2525       {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2526       {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2527       {"if",            0,      st_C_ignore},
2528       {""}, {""}, {""}, {""},
2529       {"int",           0,      st_C_typespec},
2530       {""}, {""},
2531       {"void",          0,      st_C_typespec},
2532       {""}, {""},
2533       {"interface",     C_JAVA, st_C_struct},
2534       {""},
2535       {"SYSCALL",       0,      st_C_gnumacro},
2536       {""},
2537       {"return",                0,      st_C_ignore},
2538       {""}, {""}, {""}, {""}, {""}, {""}, {""},
2539       {"while",         0,      st_C_ignore},
2540       {"auto",          0,      st_C_typespec},
2541       {""}, {""}, {""}, {""}, {""}, {""},
2542       {"float",         0,      st_C_typespec},
2543       {"typedef",       0,      st_C_typedef},
2544       {"typename",      C_PLPL, st_C_typespec},
2545       {""}, {""}, {""},
2546       {"friend",                C_PLPL, st_C_ignore},
2547       {"volatile",      0,      st_C_typespec},
2548       {""}, {""},
2549       {"for",           0,      st_C_ignore},
2550       {"const",         0,      st_C_typespec},
2551       {"import",                C_JAVA, st_C_ignore},
2552       {""},
2553       {"define",        0,      st_C_define},
2554       {"long",          0,      st_C_typespec},
2555       {"implements",    C_JAVA, st_C_javastruct},
2556       {"signed",        0,      st_C_typespec},
2557       {""},
2558       {"extern",        0,      st_C_extern},
2559       {"extends",       C_JAVA, st_C_javastruct},
2560       {""},
2561       {"mutable",       C_PLPL, st_C_typespec},
2562       {"template",      0,      st_C_template},
2563       {"short",         0,      st_C_typespec},
2564       {"bool",          C_PLPL, st_C_typespec},
2565       {"char",          0,      st_C_typespec},
2566       {"class",         0,      st_C_class},
2567       {"operator",      C_PLPL, st_C_operator},
2568       {""},
2569       {"switch",                0,      st_C_ignore},
2570       {""},
2571       {"ENTRY",         0,      st_C_gnumacro},
2572       {""},
2573       {"package",       C_JAVA, st_C_ignore},
2574       {"union",         0,      st_C_struct},
2575       {"@end",          0,      st_C_objend},
2576       {"struct",        0,      st_C_struct},
2577       {"namespace",     C_PLPL, st_C_struct},
2578       {""}, {""},
2579       {"domain",        C_STAR, st_C_struct},
2580       {"@interface",    0,      st_C_objprot},
2581       {"PSEUDO",                0,      st_C_gnumacro},
2582       {"double",        0,      st_C_typespec},
2583       {""},
2584       {"@protocol",     0,      st_C_objprot},
2585       {""},
2586       {"static",        0,      st_C_typespec},
2587       {""}, {""},
2588       {"DEFUN",         0,      st_C_gnumacro},
2589       {""}, {""}, {""}, {""},
2590       {"explicit",      C_PLPL, st_C_typespec},
2591       {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2592       {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2593       {""},
2594       {"enum",          0,      st_C_enum},
2595       {""}, {""},
2596       {"unsigned",      0,      st_C_typespec},
2597       {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""}, {""},
2598       {"@implementation",0,     st_C_objimpl}
2599     };
2600
2601   if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH)
2602     {
2603       register int key = hash (str, len);
2604
2605       if (key <= MAX_HASH_VALUE && key >= 0)
2606         {
2607           register const char *s = wordlist[key].name;
2608
2609           if (*str == *s && !strncmp (str + 1, s + 1, len - 1))
2610             return &wordlist[key];
2611         }
2612     }
2613   return 0;
2614 }
2615 /*%>*/
2616
2617 static enum sym_type
2618 C_symtype (str, len, c_ext)
2619      char *str;
2620      int len;
2621      int c_ext;
2622 {
2623   register struct C_stab_entry *se = in_word_set (str, len);
2624
2625   if (se == NULL || (se->c_ext && !(c_ext & se->c_ext)))
2626     return st_none;
2627   return se->type;
2628 }
2629
2630 \f
2631 /*
2632  * C functions and variables are recognized using a simple
2633  * finite automaton.  fvdef is its state variable.
2634  */
2635 static enum
2636 {
2637   fvnone,                       /* nothing seen */
2638   fdefunkey,                    /* Emacs DEFUN keyword seen */
2639   fdefunname,                   /* Emacs DEFUN name seen */
2640   foperator,                    /* func: operator keyword seen (cplpl) */
2641   fvnameseen,                   /* function or variable name seen */
2642   fstartlist,                   /* func: just after open parenthesis */
2643   finlist,                      /* func: in parameter list */
2644   flistseen,                    /* func: after parameter list */
2645   fignore,                      /* func: before open brace */
2646   vignore                       /* var-like: ignore until ';' */
2647 } fvdef;
2648
2649 static bool fvextern;           /* func or var: extern keyword seen; */
2650
2651 /*
2652  * typedefs are recognized using a simple finite automaton.
2653  * typdef is its state variable.
2654  */
2655 static enum
2656 {
2657   tnone,                        /* nothing seen */
2658   tkeyseen,                     /* typedef keyword seen */
2659   ttypeseen,                    /* defined type seen */
2660   tinbody,                      /* inside typedef body */
2661   tend,                         /* just before typedef tag */
2662   tignore                       /* junk after typedef tag */
2663 } typdef;
2664
2665 /*
2666  * struct-like structures (enum, struct and union) are recognized
2667  * using another simple finite automaton.  `structdef' is its state
2668  * variable.
2669  */
2670 static enum
2671 {
2672   snone,                        /* nothing seen yet,
2673                                    or in struct body if cblev > 0 */
2674   skeyseen,                     /* struct-like keyword seen */
2675   stagseen,                     /* struct-like tag seen */
2676   sintemplate,                  /* inside template (ignore) */
2677   scolonseen                    /* colon seen after struct-like tag */
2678 } structdef;
2679
2680 /*
2681  * When objdef is different from onone, objtag is the name of the class.
2682  */
2683 static char *objtag = "<uninited>";
2684
2685 /*
2686  * Yet another little state machine to deal with preprocessor lines.
2687  */
2688 static enum
2689 {
2690   dnone,                        /* nothing seen */
2691   dsharpseen,                   /* '#' seen as first char on line */
2692   ddefineseen,                  /* '#' and 'define' seen */
2693   dignorerest                   /* ignore rest of line */
2694 } definedef;
2695
2696 /*
2697  * State machine for Objective C protocols and implementations.
2698  * Idea by Tom R.Hageman <tom@basil.icce.rug.nl> (1995)
2699  */
2700 static enum
2701 {
2702   onone,                        /* nothing seen */
2703   oprotocol,                    /* @interface or @protocol seen */
2704   oimplementation,              /* @implementations seen */
2705   otagseen,                     /* class name seen */
2706   oparenseen,                   /* parenthesis before category seen */
2707   ocatseen,                     /* category name seen */
2708   oinbody,                      /* in @implementation body */
2709   omethodsign,                  /* in @implementation body, after +/- */
2710   omethodtag,                   /* after method name */
2711   omethodcolon,                 /* after method colon */
2712   omethodparm,                  /* after method parameter */
2713   oignore                       /* wait for @end */
2714 } objdef;
2715
2716
2717 /*
2718  * Use this structure to keep info about the token read, and how it
2719  * should be tagged.  Used by the make_C_tag function to build a tag.
2720  */
2721 static struct tok
2722 {
2723   char *line;                   /* string containing the token */
2724   int offset;                   /* where the token starts in LINE */
2725   int length;                   /* token length */
2726   /*
2727     The previous members can be used to pass strings around for generic
2728     purposes.  The following ones specifically refer to creating tags.  In this
2729     case the token contained here is the pattern that will be used to create a
2730     tag.
2731   */
2732   bool valid;                   /* do not create a tag; the token should be
2733                                    invalidated whenever a state machine is
2734                                    reset prematurely */
2735   bool named;                   /* create a named tag */
2736   int lineno;                   /* source line number of tag */
2737   long linepos;                 /* source char number of tag */
2738 } token;                        /* latest token read */
2739
2740 /*
2741  * Variables and functions for dealing with nested structures.
2742  * Idea by Mykola Dzyuba <mdzyuba@yahoo.com> (2001)
2743  */
2744 static void pushclass_above __P((int, char *, int));
2745 static void popclass_above __P((int));
2746 static void write_classname __P((linebuffer *, char *qualifier));
2747
2748 static struct {
2749   char **cname;                 /* nested class names */
2750   int *cblev;                   /* nested class curly brace level */
2751   int nl;                       /* class nesting level (elements used) */
2752   int size;                     /* length of the array */
2753 } cstack;                       /* stack for nested declaration tags */
2754 /* Current struct nesting depth (namespace, class, struct, union, enum). */
2755 #define nestlev         (cstack.nl)
2756 /* After struct keyword or in struct body, not inside a nested function. */
2757 #define instruct        (structdef == snone && nestlev > 0                      \
2758                          && cblev == cstack.cblev[nestlev-1] + 1)
2759
2760 static void
2761 pushclass_above (cblev, str, len)
2762      int cblev;
2763      char *str;
2764      int len;
2765 {
2766   int nl;
2767
2768   popclass_above (cblev);
2769   nl = cstack.nl;
2770   if (nl >= cstack.size)
2771     {
2772       int size = cstack.size *= 2;
2773       xrnew (cstack.cname, size, char *);
2774       xrnew (cstack.cblev, size, int);
2775     }
2776   assert (nl == 0 || cstack.cblev[nl-1] < cblev);
2777   cstack.cname[nl] = (str == NULL) ? NULL : savenstr (str, len);
2778   cstack.cblev[nl] = cblev;
2779   cstack.nl = nl + 1;
2780 }
2781
2782 static void
2783 popclass_above (cblev)
2784      int cblev;
2785 {
2786   int nl;
2787
2788   for (nl = cstack.nl - 1;
2789        nl >= 0 && cstack.cblev[nl] >= cblev;
2790        nl--)
2791     {
2792       if (cstack.cname[nl] != NULL)
2793         free (cstack.cname[nl]);
2794       cstack.nl = nl;
2795     }
2796 }
2797
2798 static void
2799 write_classname (cn, qualifier)
2800      linebuffer *cn;
2801      char *qualifier;
2802 {
2803   int i, len;
2804   int qlen = strlen (qualifier);
2805
2806   if (cstack.nl == 0 || cstack.cname[0] == NULL)
2807     {
2808       len = 0;
2809       cn->len = 0;
2810       cn->buffer[0] = '\0';
2811     }
2812   else
2813     {
2814       len = strlen (cstack.cname[0]);
2815       linebuffer_setlen (cn, len);
2816       strcpy (cn->buffer, cstack.cname[0]);
2817     }
2818   for (i = 1; i < cstack.nl; i++)
2819     {
2820       char *s;
2821       int slen;
2822
2823       s = cstack.cname[i];
2824       if (s == NULL)
2825         continue;
2826       slen = strlen (s);
2827       len += slen + qlen;
2828       linebuffer_setlen (cn, len);
2829       strncat (cn->buffer, qualifier, qlen);
2830       strncat (cn->buffer, s, slen);
2831     }
2832 }
2833
2834 \f
2835 static bool consider_token __P((char *, int, int, int *, int, int, bool *));
2836 static void make_C_tag __P((bool));
2837
2838 /*
2839  * consider_token ()
2840  *      checks to see if the current token is at the start of a
2841  *      function or variable, or corresponds to a typedef, or
2842  *      is a struct/union/enum tag, or #define, or an enum constant.
2843  *
2844  *      *IS_FUNC gets TRUE iff the token is a function or #define macro
2845  *      with args.  C_EXTP points to which language we are looking at.
2846  *
2847  * Globals
2848  *      fvdef                   IN OUT
2849  *      structdef               IN OUT
2850  *      definedef               IN OUT
2851  *      typdef                  IN OUT
2852  *      objdef                  IN OUT
2853  */
2854
2855 static bool
2856 consider_token (str, len, c, c_extp, cblev, parlev, is_func_or_var)
2857      register char *str;        /* IN: token pointer */
2858      register int len;          /* IN: token length */
2859      register int c;            /* IN: first char after the token */
2860      int *c_extp;               /* IN, OUT: C extensions mask */
2861      int cblev;                 /* IN: curly brace level */
2862      int parlev;                /* IN: parenthesis level */
2863      bool *is_func_or_var;      /* OUT: function or variable found */
2864 {
2865   /* When structdef is stagseen, scolonseen, or snone with cblev > 0,
2866      structtype is the type of the preceding struct-like keyword, and
2867      structcblev is the curly brace level where it has been seen. */
2868   static enum sym_type structtype;
2869   static int structcblev;
2870   static enum sym_type toktype;
2871
2872
2873   toktype = C_symtype (str, len, *c_extp);
2874
2875   /*
2876    * Advance the definedef state machine.
2877    */
2878   switch (definedef)
2879     {
2880     case dnone:
2881       /* We're not on a preprocessor line. */
2882       if (toktype == st_C_gnumacro)
2883         {
2884           fvdef = fdefunkey;
2885           return FALSE;
2886         }
2887       break;
2888     case dsharpseen:
2889       if (toktype == st_C_define)
2890         {
2891           definedef = ddefineseen;
2892         }
2893       else
2894         {
2895           definedef = dignorerest;
2896         }
2897       return FALSE;
2898     case ddefineseen:
2899       /*
2900        * Make a tag for any macro, unless it is a constant
2901        * and constantypedefs is FALSE.
2902        */
2903       definedef = dignorerest;
2904       *is_func_or_var = (c == '(');
2905       if (!*is_func_or_var && !constantypedefs)
2906         return FALSE;
2907       else
2908         return TRUE;
2909     case dignorerest:
2910       return FALSE;
2911     default:
2912       error ("internal error: definedef value.", (char *)NULL);
2913     }
2914
2915   /*
2916    * Now typedefs
2917    */
2918   switch (typdef)
2919     {
2920     case tnone:
2921       if (toktype == st_C_typedef)
2922         {
2923           if (typedefs)
2924             typdef = tkeyseen;
2925           fvextern = FALSE;
2926           fvdef = fvnone;
2927           return FALSE;
2928         }
2929       break;
2930     case tkeyseen:
2931       switch (toktype)
2932         {
2933         case st_none:
2934         case st_C_typespec:
2935         case st_C_class:
2936         case st_C_struct:
2937         case st_C_enum:
2938           typdef = ttypeseen;
2939           break;
2940         }
2941       break;
2942     case ttypeseen:
2943       if (structdef == snone && fvdef == fvnone)
2944         {
2945           fvdef = fvnameseen;
2946           return TRUE;
2947         }
2948       break;
2949     case tend:
2950       switch (toktype)
2951         {
2952         case st_C_typespec:
2953         case st_C_class:
2954         case st_C_struct:
2955         case st_C_enum:
2956           return FALSE;
2957         }
2958       return TRUE;
2959     }
2960
2961   /*
2962    * This structdef business is NOT invoked when we are ctags and the
2963    * file is plain C.  This is because a struct tag may have the same
2964    * name as another tag, and this loses with ctags.
2965    */
2966   switch (toktype)
2967     {
2968     case st_C_javastruct:
2969       if (structdef == stagseen)
2970         structdef = scolonseen;
2971       return FALSE;
2972     case st_C_template:
2973     case st_C_class:
2974       if ((*c_extp & C_AUTO)    /* automatic detection of C++ language */
2975           && cblev == 0
2976           && definedef == dnone && structdef == snone
2977           && typdef == tnone && fvdef == fvnone)
2978         *c_extp = (*c_extp | C_PLPL) & ~C_AUTO;
2979       if (toktype == st_C_template)
2980         break;
2981       /* FALLTHRU */
2982     case st_C_struct:
2983     case st_C_enum:
2984       if (parlev == 0
2985           && fvdef != vignore
2986           && (typdef == tkeyseen
2987               || (typedefs_or_cplusplus && structdef == snone)))
2988         {
2989           structdef = skeyseen;
2990           structtype = toktype;
2991           structcblev = cblev;
2992         }
2993       return FALSE;
2994     }
2995
2996   if (structdef == skeyseen)
2997     {
2998       structdef = stagseen;
2999       return TRUE;
3000     }
3001
3002   if (typdef != tnone)
3003     definedef = dnone;
3004
3005   /* Detect Objective C constructs. */
3006   switch (objdef)
3007     {
3008     case onone:
3009       switch (toktype)
3010         {
3011         case st_C_objprot:
3012           objdef = oprotocol;
3013           return FALSE;
3014         case st_C_objimpl:
3015           objdef = oimplementation;
3016           return FALSE;
3017         }
3018       break;
3019     case oimplementation:
3020       /* Save the class tag for functions or variables defined inside. */
3021       objtag = savenstr (str, len);
3022       objdef = oinbody;
3023       return FALSE;
3024     case oprotocol:
3025       /* Save the class tag for categories. */
3026       objtag = savenstr (str, len);
3027       objdef = otagseen;
3028       *is_func_or_var = TRUE;
3029       return TRUE;
3030     case oparenseen:
3031       objdef = ocatseen;
3032       *is_func_or_var = TRUE;
3033       return TRUE;
3034     case oinbody:
3035       break;
3036     case omethodsign:
3037       if (parlev == 0)
3038         {
3039           objdef = omethodtag;
3040           linebuffer_setlen (&token_name, len);
3041           strncpy (token_name.buffer, str, len);
3042           token_name.buffer[len] = '\0';
3043           return TRUE;
3044         }
3045       return FALSE;
3046     case omethodcolon:
3047       if (parlev == 0)
3048         objdef = omethodparm;
3049       return FALSE;
3050     case omethodparm:
3051       if (parlev == 0)
3052         {
3053           objdef = omethodtag;
3054           linebuffer_setlen (&token_name, token_name.len + len);
3055           strncat (token_name.buffer, str, len);
3056           return TRUE;
3057         }
3058       return FALSE;
3059     case oignore:
3060       if (toktype == st_C_objend)
3061         {
3062           /* Memory leakage here: the string pointed by objtag is
3063              never released, because many tests would be needed to
3064              avoid breaking on incorrect input code.  The amount of
3065              memory leaked here is the sum of the lengths of the
3066              class tags.
3067           free (objtag); */
3068           objdef = onone;
3069         }
3070       return FALSE;
3071     }
3072
3073   /* A function, variable or enum constant? */
3074   switch (toktype)
3075     {
3076     case st_C_extern:
3077       fvextern = TRUE;
3078       /* FALLTHRU */
3079     case st_C_typespec:
3080       switch  (fvdef)
3081         {
3082         case finlist:
3083         case flistseen:
3084         case fignore:
3085         case vignore:
3086           break;
3087         default:
3088           fvdef = fvnone;
3089         }
3090       return FALSE;
3091     case st_C_ignore:
3092       fvextern = FALSE;
3093       fvdef = vignore;
3094       return FALSE;
3095     case st_C_operator:
3096       fvdef = foperator;
3097       *is_func_or_var = TRUE;
3098       return TRUE;
3099     case st_none:
3100       if (constantypedefs
3101           && structdef == snone
3102           && structtype == st_C_enum && cblev > structcblev)
3103         return TRUE;            /* enum constant */
3104       switch (fvdef)
3105         {
3106         case fdefunkey:
3107           if (cblev > 0)
3108             break;
3109           fvdef = fdefunname;   /* GNU macro */
3110           *is_func_or_var = TRUE;
3111           return TRUE;
3112         case fvnone:
3113           if ((strneq (str, "asm", 3) && endtoken (str[3]))
3114               || (strneq (str, "__asm__", 7) && endtoken (str[7])))
3115             {
3116               fvdef = vignore;
3117               return FALSE;
3118             }
3119           if (strneq (str+len-10, "::operator", 10))
3120             {
3121               if (*c_extp & C_AUTO) /* automatic detection of C++ */
3122                 *c_extp = (*c_extp | C_PLPL) & ~C_AUTO;
3123               fvdef = foperator;
3124               *is_func_or_var = TRUE;
3125               return TRUE;
3126             }
3127           if (cblev > 0 && !instruct)
3128             break;
3129           fvdef = fvnameseen;   /* function or variable */
3130           *is_func_or_var = TRUE;
3131           return TRUE;
3132         }
3133       break;
3134     }
3135
3136   return FALSE;
3137 }
3138
3139 \f
3140 /*
3141  * C_entries often keeps pointers to tokens or lines which are older than
3142  * the line currently read.  By keeping two line buffers, and switching
3143  * them at end of line, it is possible to use those pointers.
3144  */
3145 static struct
3146 {
3147   long linepos;
3148   linebuffer lb;
3149 } lbs[2];
3150
3151 #define current_lb_is_new (newndx == curndx)
3152 #define switch_line_buffers() (curndx = 1 - curndx)
3153
3154 #define curlb (lbs[curndx].lb)
3155 #define newlb (lbs[newndx].lb)
3156 #define curlinepos (lbs[curndx].linepos)
3157 #define newlinepos (lbs[newndx].linepos)
3158
3159 #define plainc ((c_ext & C_EXT) == C_PLAIN)
3160 #define cplpl (c_ext & C_PLPL)
3161 #define cjava ((c_ext & C_JAVA) == C_JAVA)
3162
3163 #define CNL_SAVE_DEFINEDEF()                                            \
3164 do {                                                                    \
3165   curlinepos = charno;                                                  \
3166   readline (&curlb, inf);                                               \
3167   lp = curlb.buffer;                                                    \
3168   quotednl = FALSE;                                                     \
3169   newndx = curndx;                                                      \
3170 } while (0)
3171
3172 #define CNL()                                                           \
3173 do {                                                                    \
3174   CNL_SAVE_DEFINEDEF();                                                 \
3175   if (savetoken.valid)                                                  \
3176     {                                                                   \
3177       token = savetoken;                                                \
3178       savetoken.valid = FALSE;                                          \
3179     }                                                                   \
3180   definedef = dnone;                                                    \
3181 } while (0)
3182
3183
3184 static void
3185 make_C_tag (isfun)
3186      bool isfun;
3187 {
3188   /* This function should never be called when token.valid is FALSE, but
3189      we must protect against invalid input or internal errors. */
3190   if (!DEBUG && !token.valid)
3191     return;
3192
3193   if (token.valid)
3194     make_tag (token_name.buffer, token_name.len, isfun, token.line,
3195               token.offset+token.length+1, token.lineno, token.linepos);
3196   else                          /* this case is optimised away if !DEBUG */
3197     make_tag (concat ("INVALID TOKEN:-->", token_name.buffer, ""),
3198               token_name.len + 17, isfun, token.line,
3199               token.offset+token.length+1, token.lineno, token.linepos);
3200
3201   token.valid = FALSE;
3202 }
3203
3204
3205 /*
3206  * C_entries ()
3207  *      This routine finds functions, variables, typedefs,
3208  *      #define's, enum constants and struct/union/enum definitions in
3209  *      C syntax and adds them to the list.
3210  */
3211 static void
3212 C_entries (c_ext, inf)
3213      int c_ext;                 /* extension of C */
3214      FILE *inf;                 /* input file */
3215 {
3216   register char c;              /* latest char read; '\0' for end of line */
3217   register char *lp;            /* pointer one beyond the character `c' */
3218   int curndx, newndx;           /* indices for current and new lb */
3219   register int tokoff;          /* offset in line of start of current token */
3220   register int toklen;          /* length of current token */
3221   char *qualifier;              /* string used to qualify names */
3222   int qlen;                     /* length of qualifier */
3223   int cblev;                    /* current curly brace level */
3224   int parlev;                   /* current parenthesis level */
3225   int typdefcblev;              /* cblev where a typedef struct body begun */
3226   bool incomm, inquote, inchar, quotednl, midtoken;
3227   bool yacc_rules;              /* in the rules part of a yacc file */
3228   struct tok savetoken;         /* token saved during preprocessor handling */
3229
3230
3231   linebuffer_init (&lbs[0].lb);
3232   linebuffer_init (&lbs[1].lb);
3233   if (cstack.size == 0)
3234     {
3235       cstack.size = (DEBUG) ? 1 : 4;
3236       cstack.nl = 0;
3237       cstack.cname = xnew (cstack.size, char *);
3238       cstack.cblev = xnew (cstack.size, int);
3239     }
3240
3241   tokoff = toklen = typdefcblev = 0; /* keep compiler quiet */
3242   curndx = newndx = 0;
3243   lp = curlb.buffer;
3244   *lp = 0;
3245
3246   fvdef = fvnone; fvextern = FALSE; typdef = tnone;
3247   structdef = snone; definedef = dnone; objdef = onone;
3248   yacc_rules = FALSE;
3249   midtoken = inquote = inchar = incomm = quotednl = FALSE;
3250   token.valid = savetoken.valid = FALSE;
3251   cblev = 0;
3252   parlev = 0;
3253   if (cjava)
3254     { qualifier = "."; qlen = 1; }
3255   else
3256     { qualifier = "::"; qlen = 2; }
3257
3258
3259   while (!feof (inf))
3260     {
3261       c = *lp++;
3262       if (c == '\\')
3263         {
3264           /* If we're at the end of the line, the next character is a
3265              '\0'; don't skip it, because it's the thing that tells us
3266              to read the next line.  */
3267           if (*lp == '\0')
3268             {
3269               quotednl = TRUE;
3270               continue;
3271             }
3272           lp++;
3273           c = ' ';
3274         }
3275       else if (incomm)
3276         {
3277           switch (c)
3278             {
3279             case '*':
3280               if (*lp == '/')
3281                 {
3282                   c = *lp++;
3283                   incomm = FALSE;
3284                 }
3285               break;
3286             case '\0':
3287               /* Newlines inside comments do not end macro definitions in
3288                  traditional cpp. */
3289               CNL_SAVE_DEFINEDEF ();
3290               break;
3291             }
3292           continue;
3293         }
3294       else if (inquote)
3295         {
3296           switch (c)
3297             {
3298             case '"':
3299               inquote = FALSE;
3300               break;
3301             case '\0':
3302               /* Newlines inside strings do not end macro definitions
3303                  in traditional cpp, even though compilers don't
3304                  usually accept them. */
3305               CNL_SAVE_DEFINEDEF ();
3306               break;
3307             }
3308           continue;
3309         }
3310       else if (inchar)
3311         {
3312           switch (c)
3313             {
3314             case '\0':
3315               /* Hmmm, something went wrong. */
3316               CNL ();
3317               /* FALLTHRU */
3318             case '\'':
3319               inchar = FALSE;
3320               break;
3321             }
3322           continue;
3323         }
3324       else
3325         switch (c)
3326           {
3327           case '"':
3328             inquote = TRUE;
3329             switch (fvdef)
3330               {
3331               case fdefunkey:
3332               case fstartlist:
3333               case finlist:
3334               case fignore:
3335               case vignore:
3336                 break;
3337               default:
3338                 fvextern = FALSE;
3339                 fvdef = fvnone;
3340               }
3341             continue;
3342           case '\'':
3343             inchar = TRUE;
3344             if (fvdef != finlist && fvdef != fignore && fvdef !=vignore)
3345               {
3346                 fvextern = FALSE;
3347                 fvdef = fvnone;
3348               }
3349             continue;
3350           case '/':
3351             if (*lp == '*')
3352               {
3353                 lp++;
3354                 incomm = TRUE;
3355                 continue;
3356               }
3357             else if (/* cplpl && */ *lp == '/')
3358               {
3359                 c = '\0';
3360                 break;
3361               }
3362             else
3363               break;
3364           case '%':
3365             if ((c_ext & YACC) && *lp == '%')
3366               {
3367                 /* Entering or exiting rules section in yacc file. */
3368                 lp++;
3369                 definedef = dnone; fvdef = fvnone; fvextern = FALSE;
3370                 typdef = tnone; structdef = snone;
3371                 midtoken = inquote = inchar = incomm = quotednl = FALSE;
3372                 cblev = 0;
3373                 yacc_rules = !yacc_rules;
3374                 continue;
3375               }
3376             else
3377               break;
3378           case '#':
3379             if (definedef == dnone)
3380               {
3381                 char *cp;
3382                 bool cpptoken = TRUE;
3383
3384                 /* Look back on this line.  If all blanks, or nonblanks
3385                    followed by an end of comment, this is a preprocessor
3386                    token. */
3387                 for (cp = newlb.buffer; cp < lp-1; cp++)
3388                   if (!iswhite (*cp))
3389                     {
3390                       if (*cp == '*' && *(cp+1) == '/')
3391                         {
3392                           cp++;
3393                           cpptoken = TRUE;
3394                         }
3395                       else
3396                         cpptoken = FALSE;
3397                     }
3398                 if (cpptoken)
3399                   definedef = dsharpseen;
3400               } /* if (definedef == dnone) */
3401
3402             continue;
3403           } /* switch (c) */
3404
3405
3406       /* Consider token only if some involved conditions are satisfied. */
3407       if (typdef != tignore
3408           && definedef != dignorerest
3409           && fvdef != finlist
3410           && structdef != sintemplate
3411           && (definedef != dnone
3412               || structdef != scolonseen))
3413         {
3414           if (midtoken)
3415             {
3416               if (endtoken (c))
3417                 {
3418                   if (c == ':' && *lp == ':' && begtoken (lp[1]))
3419                     /* This handles :: in the middle,
3420                        but not at the beginning of an identifier.
3421                        Also, space-separated :: is not recognised. */
3422                     {
3423                       if (c_ext & C_AUTO) /* automatic detection of C++ */
3424                         c_ext = (c_ext | C_PLPL) & ~C_AUTO;
3425                       lp += 2;
3426                       toklen += 2;
3427                       c = lp[-1];
3428                       goto still_in_token;
3429                     }
3430                   else
3431                     {
3432                       bool funorvar = FALSE;
3433
3434                       if (yacc_rules
3435                           || consider_token (newlb.buffer + tokoff, toklen, c,
3436                                              &c_ext, cblev, parlev, &funorvar))
3437                         {
3438                           if (fvdef == foperator)
3439                             {
3440                               char *oldlp = lp;
3441                               lp = skip_spaces (lp-1);
3442                               if (*lp != '\0')
3443                                 lp += 1;
3444                               while (*lp != '\0'
3445                                      && !iswhite (*lp) && *lp != '(')
3446                                 lp += 1;
3447                               c = *lp++;
3448                               toklen += lp - oldlp;
3449                             }
3450                           token.named = FALSE;
3451                           if (!plainc
3452                               && nestlev > 0 && definedef == dnone)
3453                             /* in struct body */
3454                             {
3455                               write_classname (&token_name, qualifier);
3456                               linebuffer_setlen (&token_name,
3457                                                  token_name.len+qlen+toklen);
3458                               strcat (token_name.buffer, qualifier);
3459                               strncat (token_name.buffer,
3460                                        newlb.buffer + tokoff, toklen);
3461                               token.named = TRUE;
3462                             }
3463                           else if (objdef == ocatseen)
3464                             /* Objective C category */
3465                             {
3466                               int len = strlen (objtag) + 2 + toklen;
3467                               linebuffer_setlen (&token_name, len);
3468                               strcpy (token_name.buffer, objtag);
3469                               strcat (token_name.buffer, "(");
3470                               strncat (token_name.buffer,
3471                                        newlb.buffer + tokoff, toklen);
3472                               strcat (token_name.buffer, ")");
3473                               token.named = TRUE;
3474                             }
3475                           else if (objdef == omethodtag
3476                                    || objdef == omethodparm)
3477                             /* Objective C method */
3478                             {
3479                               token.named = TRUE;
3480                             }
3481                           else if (fvdef == fdefunname)
3482                             /* GNU DEFUN and similar macros */
3483                             {
3484                               bool defun = (newlb.buffer[tokoff] == 'F');
3485                               int off = tokoff;
3486                               int len = toklen;
3487
3488                               /* Rewrite the tag so that emacs lisp DEFUNs
3489                                  can be found by their elisp name */
3490                               if (defun)
3491                                 {
3492                                   off += 1;
3493                                   len -= 1;
3494                                 }
3495                               len = toklen;
3496                               linebuffer_setlen (&token_name, len);
3497                               strncpy (token_name.buffer,
3498                                        newlb.buffer + off, len);
3499                               token_name.buffer[len] = '\0';
3500                               if (defun)
3501                                 while (--len >= 0)
3502                                   if (token_name.buffer[len] == '_')
3503                                     token_name.buffer[len] = '-';
3504                               token.named = defun;
3505                             }
3506                           else
3507                             {
3508                               linebuffer_setlen (&token_name, toklen);
3509                               strncpy (token_name.buffer,
3510                                        newlb.buffer + tokoff, toklen);
3511                               token_name.buffer[toklen] = '\0';
3512                               /* Name macros and members. */
3513                               token.named = (structdef == stagseen
3514                                              || typdef == ttypeseen
3515                                              || typdef == tend
3516                                              || (funorvar
3517                                                  && definedef == dignorerest)
3518                                              || (funorvar
3519                                                  && definedef == dnone
3520                                                  && structdef == snone
3521                                                  && cblev > 0));
3522                             }
3523                           token.lineno = lineno;
3524                           token.offset = tokoff;
3525                           token.length = toklen;
3526                           token.line = newlb.buffer;
3527                           token.linepos = newlinepos;
3528                           token.valid = TRUE;
3529
3530                           if (definedef == dnone
3531                               && (fvdef == fvnameseen
3532                                   || fvdef == foperator
3533                                   || structdef == stagseen
3534                                   || typdef == tend
3535                                   || typdef == ttypeseen
3536                                   || objdef != onone))
3537                             {
3538                               if (current_lb_is_new)
3539                                 switch_line_buffers ();
3540                             }
3541                           else if (definedef != dnone
3542                                    || fvdef == fdefunname
3543                                    || instruct)
3544                             make_C_tag (funorvar);
3545                         }
3546                       midtoken = FALSE;
3547                     }
3548                 } /* if (endtoken (c)) */
3549               else if (intoken (c))
3550                 still_in_token:
3551                 {
3552                   toklen++;
3553                   continue;
3554                 }
3555             } /* if (midtoken) */
3556           else if (begtoken (c))
3557             {
3558               switch (definedef)
3559                 {
3560                 case dnone:
3561                   switch (fvdef)
3562                     {
3563                     case fstartlist:
3564                       fvdef = finlist;
3565                       continue;
3566                     case flistseen:
3567                       if (plainc || declarations)
3568                         {
3569                           make_C_tag (TRUE); /* a function */
3570                           fvdef = fignore;
3571                         }
3572                       break;
3573                     case fvnameseen:
3574                       fvdef = fvnone;
3575                       break;
3576                     }
3577                   if (structdef == stagseen && !cjava)
3578                     {
3579                       popclass_above (cblev);
3580                       structdef = snone;
3581                     }
3582                   break;
3583                 case dsharpseen:
3584                   savetoken = token;
3585                   break;
3586                 }
3587               if (!yacc_rules || lp == newlb.buffer + 1)
3588                 {
3589                   tokoff = lp - 1 - newlb.buffer;
3590                   toklen = 1;
3591                   midtoken = TRUE;
3592                 }
3593               continue;
3594             } /* if (begtoken) */
3595         } /* if must look at token */
3596
3597
3598       /* Detect end of line, colon, comma, semicolon and various braces
3599          after having handled a token.*/
3600       switch (c)
3601         {
3602         case ':':
3603           if (yacc_rules && token.offset == 0 && token.valid)
3604             {
3605               make_C_tag (FALSE); /* a yacc function */
3606               break;
3607             }
3608           if (definedef != dnone)
3609             break;
3610           switch (objdef)
3611             {
3612             case  otagseen:
3613               objdef = oignore;
3614               make_C_tag (TRUE); /* an Objective C class */
3615               break;
3616             case omethodtag:
3617             case omethodparm:
3618               objdef = omethodcolon;
3619               linebuffer_setlen (&token_name, token_name.len + 1);
3620               strcat (token_name.buffer, ":");
3621               break;
3622             }
3623           if (structdef == stagseen)
3624             {
3625               structdef = scolonseen;
3626               break;
3627             }
3628           /* Should be useless, but may be work as a safety net. */
3629           if (cplpl && fvdef == flistseen)
3630             {
3631               make_C_tag (TRUE); /* a function */
3632               fvdef = fignore;
3633               break;
3634             }
3635           break;
3636         case ';':
3637           if (definedef != dnone)
3638             break;
3639           switch (typdef)
3640             {
3641             case tend:
3642             case ttypeseen:
3643               make_C_tag (FALSE); /* a typedef */
3644               typdef = tnone;
3645               fvdef = fvnone;
3646               break;
3647             case tnone:
3648             case tinbody:
3649             case tignore:
3650               switch (fvdef)
3651                 {
3652                 case fignore:
3653                   if (typdef == tignore || cplpl)
3654                     fvdef = fvnone;
3655                   break;
3656                 case fvnameseen:
3657                   if ((globals && cblev == 0 && (!fvextern || declarations))
3658                       || (members && instruct))
3659                     make_C_tag (FALSE); /* a variable */
3660                   fvextern = FALSE;
3661                   fvdef = fvnone;
3662                   token.valid = FALSE;
3663                   break;
3664                 case flistseen:
3665                   if (declarations
3666                       && (typdef == tnone || (typdef != tignore && instruct)))
3667                     make_C_tag (TRUE);  /* a function declaration */
3668                   /* FALLTHRU */
3669                 default:
3670                   fvextern = FALSE;
3671                   fvdef = fvnone;
3672                   if (declarations
3673                        && cplpl && structdef == stagseen)
3674                     make_C_tag (FALSE); /* forward declaration */
3675                   else
3676                     token.valid = FALSE;
3677                 } /* switch (fvdef) */
3678               /* FALLTHRU */
3679             default:
3680               if (!instruct)
3681                 typdef = tnone;
3682             }
3683           if (structdef == stagseen)
3684             structdef = snone;
3685           break;
3686         case ',':
3687           if (definedef != dnone)
3688             break;
3689           switch (objdef)
3690             {
3691             case omethodtag:
3692             case omethodparm:
3693               make_C_tag (TRUE); /* an Objective C method */
3694               objdef = oinbody;
3695               break;
3696             }
3697           switch (fvdef)
3698             {
3699             case fdefunkey:
3700             case foperator:
3701             case fstartlist:
3702             case finlist:
3703             case fignore:
3704             case vignore:
3705               break;
3706             case fdefunname:
3707               fvdef = fignore;
3708               break;
3709             case fvnameseen:    /* a variable */
3710               if ((globals && cblev == 0 && (!fvextern || declarations))
3711                   || (members && instruct))
3712                 make_C_tag (FALSE);
3713               break;
3714             case flistseen:     /* a function */
3715               if ((declarations && typdef == tnone && !instruct)
3716                   || (members && typdef != tignore && instruct))
3717                 {
3718                   make_C_tag (TRUE); /* a function declaration */
3719                   fvdef = fvnameseen;
3720                 }
3721               else if (!declarations)
3722                 fvdef = fvnone;
3723               token.valid = FALSE;
3724               break;
3725             default:
3726               fvdef = fvnone;
3727             }
3728           if (structdef == stagseen)
3729             structdef = snone;
3730           break;
3731         case '[':
3732           if (definedef != dnone)
3733             break;
3734           if (structdef == stagseen)
3735             structdef = snone;
3736           switch (typdef)
3737             {
3738             case ttypeseen:
3739             case tend:
3740               typdef = tignore;
3741               make_C_tag (FALSE);       /* a typedef */
3742               break;
3743             case tnone:
3744             case tinbody:
3745               switch (fvdef)
3746                 {
3747                 case foperator:
3748                 case finlist:
3749                 case fignore:
3750                 case vignore:
3751                   break;
3752                 case fvnameseen:
3753                   if ((members && cblev == 1)
3754                       || (globals && cblev == 0
3755                           && (!fvextern || declarations)))
3756                     make_C_tag (FALSE); /* a variable */
3757                   /* FALLTHRU */
3758                 default:
3759                   fvdef = fvnone;
3760                 }
3761               break;
3762             }
3763           break;
3764         case '(':
3765           if (definedef != dnone)
3766             break;
3767           if (objdef == otagseen && parlev == 0)
3768             objdef = oparenseen;
3769           switch (fvdef)
3770             {
3771             case fvnameseen:
3772               if (typdef == ttypeseen
3773                   && *lp != '*'
3774                   && !instruct)
3775                 {
3776                   /* This handles constructs like:
3777                      typedef void OperatorFun (int fun); */
3778                   make_C_tag (FALSE);
3779                   typdef = tignore;
3780                   fvdef = fignore;
3781                   break;
3782                 }
3783               /* FALLTHRU */
3784             case foperator:
3785               fvdef = fstartlist;
3786               break;
3787             case flistseen:
3788               fvdef = finlist;
3789               break;
3790             }
3791           parlev++;
3792           break;
3793         case ')':
3794           if (definedef != dnone)
3795             break;
3796           if (objdef == ocatseen && parlev == 1)
3797             {
3798               make_C_tag (TRUE); /* an Objective C category */
3799               objdef = oignore;
3800             }
3801           if (--parlev == 0)
3802             {
3803               switch (fvdef)
3804                 {
3805                 case fstartlist:
3806                 case finlist:
3807                   fvdef = flistseen;
3808                   break;
3809                 }
3810               if (!instruct
3811                   && (typdef == tend
3812                       || typdef == ttypeseen))
3813                 {
3814                   typdef = tignore;
3815                   make_C_tag (FALSE); /* a typedef */
3816                 }
3817             }
3818           else if (parlev < 0)  /* can happen due to ill-conceived #if's. */
3819             parlev = 0;
3820           break;
3821         case '{':
3822           if (definedef != dnone)
3823             break;
3824           if (typdef == ttypeseen)
3825             {
3826               /* Whenever typdef is set to tinbody (currently only
3827                  here), typdefcblev should be set to cblev. */
3828               typdef = tinbody;
3829               typdefcblev = cblev;
3830             }
3831           switch (fvdef)
3832             {
3833             case flistseen:
3834               make_C_tag (TRUE);    /* a function */
3835               /* FALLTHRU */
3836             case fignore:
3837               fvdef = fvnone;
3838               break;
3839             case fvnone:
3840               switch (objdef)
3841                 {
3842                 case otagseen:
3843                   make_C_tag (TRUE); /* an Objective C class */
3844                   objdef = oignore;
3845                   break;
3846                 case omethodtag:
3847                 case omethodparm:
3848                   make_C_tag (TRUE); /* an Objective C method */
3849                   objdef = oinbody;
3850                   break;
3851                 default:
3852                   /* Neutralize `extern "C" {' grot. */
3853                   if (cblev == 0 && structdef == snone && nestlev == 0
3854                       && typdef == tnone)
3855                     cblev = -1;
3856                 }
3857               break;
3858             }
3859           switch (structdef)
3860             {
3861             case skeyseen:         /* unnamed struct */
3862               pushclass_above (cblev, NULL, 0);
3863               structdef = snone;
3864               break;
3865             case stagseen:         /* named struct or enum */
3866             case scolonseen:       /* a class */
3867               pushclass_above (cblev, token.line+token.offset, token.length);
3868               structdef = snone;
3869               make_C_tag (FALSE);  /* a struct or enum */
3870               break;
3871             }
3872           cblev++;
3873           break;
3874         case '*':
3875           if (definedef != dnone)
3876             break;
3877           if (fvdef == fstartlist)
3878             {
3879               fvdef = fvnone;   /* avoid tagging `foo' in `foo (*bar()) ()' */
3880               token.valid = FALSE;
3881             }
3882           break;
3883         case '}':
3884           if (definedef != dnone)
3885             break;
3886           if (!ignoreindent && lp == newlb.buffer + 1)
3887             {
3888               if (cblev != 0)
3889                 token.valid = FALSE;
3890               cblev = 0;        /* reset curly brace level if first column */
3891               parlev = 0;       /* also reset paren level, just in case... */
3892             }
3893           else if (cblev > 0)
3894             cblev--;
3895           else
3896             token.valid = FALSE; /* something gone amiss, token unreliable */
3897           popclass_above (cblev);
3898           structdef = snone;
3899           /* Only if typdef == tinbody is typdefcblev significant. */
3900           if (typdef == tinbody && cblev <= typdefcblev)
3901             {
3902               assert (cblev == typdefcblev);
3903               typdef = tend;
3904             }
3905           break;
3906         case '=':
3907           if (definedef != dnone)
3908             break;
3909           switch (fvdef)
3910             {
3911             case foperator:
3912             case finlist:
3913             case fignore:
3914             case vignore:
3915               break;
3916             case fvnameseen:
3917               if ((members && cblev == 1)
3918                   || (globals && cblev == 0 && (!fvextern || declarations)))
3919                 make_C_tag (FALSE); /* a variable */
3920               /* FALLTHRU */
3921             default:
3922               fvdef = vignore;
3923             }
3924           break;
3925         case '<':
3926           if (cplpl && structdef == stagseen)
3927             {
3928               structdef = sintemplate;
3929               break;
3930             }
3931           goto resetfvdef;
3932         case '>':
3933           if (structdef == sintemplate)
3934             {
3935               structdef = stagseen;
3936               break;
3937             }
3938           goto resetfvdef;
3939         case '+':
3940         case '-':
3941           if (objdef == oinbody && cblev == 0)
3942             {
3943               objdef = omethodsign;
3944               break;
3945             }
3946           /* FALLTHRU */
3947         resetfvdef:
3948         case '#': case '~': case '&': case '%': case '/': case '|':
3949         case '^': case '!': case '.': case '?': case ']':
3950           if (definedef != dnone)
3951             break;
3952           /* These surely cannot follow a function tag in C. */
3953           switch (fvdef)
3954             {
3955             case foperator:
3956             case finlist:
3957             case fignore:
3958             case vignore:
3959               break;
3960             default:
3961               fvdef = fvnone;
3962             }
3963           break;
3964         case '\0':
3965           if (objdef == otagseen)
3966             {
3967               make_C_tag (TRUE); /* an Objective C class */
3968               objdef = oignore;
3969             }
3970           /* If a macro spans multiple lines don't reset its state. */
3971           if (quotednl)
3972             CNL_SAVE_DEFINEDEF ();
3973           else
3974             CNL ();
3975           break;
3976         } /* switch (c) */
3977
3978     } /* while not eof */
3979
3980   free (lbs[0].lb.buffer);
3981   free (lbs[1].lb.buffer);
3982 }
3983
3984 /*
3985  * Process either a C++ file or a C file depending on the setting
3986  * of a global flag.
3987  */
3988 static void
3989 default_C_entries (inf)
3990      FILE *inf;
3991 {
3992   C_entries (cplusplus ? C_PLPL : C_AUTO, inf);
3993 }
3994
3995 /* Always do plain C. */
3996 static void
3997 plain_C_entries (inf)
3998      FILE *inf;
3999 {
4000   C_entries (0, inf);
4001 }
4002
4003 /* Always do C++. */
4004 static void
4005 Cplusplus_entries (inf)
4006      FILE *inf;
4007 {
4008   C_entries (C_PLPL, inf);
4009 }
4010
4011 /* Always do Java. */
4012 static void
4013 Cjava_entries (inf)
4014      FILE *inf;
4015 {
4016   C_entries (C_JAVA, inf);
4017 }
4018
4019 /* Always do C*. */
4020 static void
4021 Cstar_entries (inf)
4022      FILE *inf;
4023 {
4024   C_entries (C_STAR, inf);
4025 }
4026
4027 /* Always do Yacc. */
4028 static void
4029 Yacc_entries (inf)
4030      FILE *inf;
4031 {
4032   C_entries (YACC, inf);
4033 }
4034
4035 \f
4036 /* Useful macros. */
4037 #define LOOP_ON_INPUT_LINES(file_pointer, line_buffer, char_pointer)    \
4038   for (;                        /* loop initialization */               \
4039        !feof (file_pointer)     /* loop test */                         \
4040        &&                       /* instructions at start of loop */     \
4041           (readline (&line_buffer, file_pointer),                       \
4042            char_pointer = line_buffer.buffer,                           \
4043            TRUE);                                                       \
4044       )
4045 #define LOOKING_AT(cp, keyword) /* keyword is a constant string */      \
4046   (strneq ((cp), keyword, sizeof(keyword)-1) /* cp points at keyword */ \
4047    && notinname ((cp)[sizeof(keyword)-1])       /* end of keyword */    \
4048    && ((cp) = skip_spaces((cp)+sizeof(keyword)-1))) /* skip spaces */
4049
4050 /*
4051  * Read a file, but do no processing.  This is used to do regexp
4052  * matching on files that have no language defined.
4053  */
4054 static void
4055 just_read_file (inf)
4056      FILE *inf;
4057 {
4058   register char *dummy;
4059
4060   LOOP_ON_INPUT_LINES (inf, lb, dummy)
4061     continue;
4062 }
4063
4064 \f
4065 /* Fortran parsing */
4066
4067 static void F_takeprec __P((void));
4068 static void F_getit __P((FILE *));
4069
4070 static void
4071 F_takeprec ()
4072 {
4073   dbp = skip_spaces (dbp);
4074   if (*dbp != '*')
4075     return;
4076   dbp++;
4077   dbp = skip_spaces (dbp);
4078   if (strneq (dbp, "(*)", 3))
4079     {
4080       dbp += 3;
4081       return;
4082     }
4083   if (!ISDIGIT (*dbp))
4084     {
4085       --dbp;                    /* force failure */
4086       return;
4087     }
4088   do
4089     dbp++;
4090   while (ISDIGIT (*dbp));
4091 }
4092
4093 static void
4094 F_getit (inf)
4095      FILE *inf;
4096 {
4097   register char *cp;
4098
4099   dbp = skip_spaces (dbp);
4100   if (*dbp == '\0')
4101     {
4102       readline (&lb, inf);
4103       dbp = lb.buffer;
4104       if (dbp[5] != '&')
4105         return;
4106       dbp += 6;
4107       dbp = skip_spaces (dbp);
4108     }
4109   if (!ISALPHA (*dbp) && *dbp != '_' && *dbp != '$')
4110     return;
4111   for (cp = dbp + 1; *cp != '\0' && intoken (*cp); cp++)
4112     continue;
4113   make_tag (dbp, cp-dbp, TRUE,
4114             lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4115 }
4116
4117
4118 static void
4119 Fortran_functions (inf)
4120      FILE *inf;
4121 {
4122   LOOP_ON_INPUT_LINES (inf, lb, dbp)
4123     {
4124       if (*dbp == '%')
4125         dbp++;                  /* Ratfor escape to fortran */
4126       dbp = skip_spaces (dbp);
4127       if (*dbp == '\0')
4128         continue;
4129       switch (lowcase (*dbp))
4130         {
4131         case 'i':
4132           if (nocase_tail ("integer"))
4133             F_takeprec ();
4134           break;
4135         case 'r':
4136           if (nocase_tail ("real"))
4137             F_takeprec ();
4138           break;
4139         case 'l':
4140           if (nocase_tail ("logical"))
4141             F_takeprec ();
4142           break;
4143         case 'c':
4144           if (nocase_tail ("complex") || nocase_tail ("character"))
4145             F_takeprec ();
4146           break;
4147         case 'd':
4148           if (nocase_tail ("double"))
4149             {
4150               dbp = skip_spaces (dbp);
4151               if (*dbp == '\0')
4152                 continue;
4153               if (nocase_tail ("precision"))
4154                 break;
4155               continue;
4156             }
4157           break;
4158         }
4159       dbp = skip_spaces (dbp);
4160       if (*dbp == '\0')
4161         continue;
4162       switch (lowcase (*dbp))
4163         {
4164         case 'f':
4165           if (nocase_tail ("function"))
4166             F_getit (inf);
4167           continue;
4168         case 's':
4169           if (nocase_tail ("subroutine"))
4170             F_getit (inf);
4171           continue;
4172         case 'e':
4173           if (nocase_tail ("entry"))
4174             F_getit (inf);
4175           continue;
4176         case 'b':
4177           if (nocase_tail ("blockdata") || nocase_tail ("block data"))
4178             {
4179               dbp = skip_spaces (dbp);
4180               if (*dbp == '\0') /* assume un-named */
4181                 make_tag ("blockdata", 9, TRUE,
4182                           lb.buffer, dbp - lb.buffer, lineno, linecharno);
4183               else
4184                 F_getit (inf);  /* look for name */
4185             }
4186           continue;
4187         }
4188     }
4189 }
4190
4191 \f
4192 /*
4193  * Ada parsing
4194  * Original code by
4195  * Philippe Waroquiers <philippe.waroquiers@eurocontrol.int> (1998)
4196  */
4197
4198 static void Ada_getit __P((FILE *, char *));
4199
4200 /* Once we are positioned after an "interesting" keyword, let's get
4201    the real tag value necessary. */
4202 static void
4203 Ada_getit (inf, name_qualifier)
4204      FILE *inf;
4205      char *name_qualifier;
4206 {
4207   register char *cp;
4208   char *name;
4209   char c;
4210
4211   while (!feof (inf))
4212     {
4213       dbp = skip_spaces (dbp);
4214       if (*dbp == '\0'
4215           || (dbp[0] == '-' && dbp[1] == '-'))
4216         {
4217           readline (&lb, inf);
4218           dbp = lb.buffer;
4219         }
4220       switch (lowcase(*dbp))
4221         {
4222         case 'b':
4223           if (nocase_tail ("body"))
4224             {
4225               /* Skipping body of   procedure body   or   package body or ....
4226                  resetting qualifier to body instead of spec. */
4227               name_qualifier = "/b";
4228               continue;
4229             }
4230           break;
4231         case 't':
4232           /* Skipping type of   task type   or   protected type ... */
4233           if (nocase_tail ("type"))
4234             continue;
4235           break;
4236         }
4237       if (*dbp == '"')
4238         {
4239           dbp += 1;
4240           for (cp = dbp; *cp != '\0' && *cp != '"'; cp++)
4241             continue;
4242         }
4243       else
4244         {
4245           dbp = skip_spaces (dbp);
4246           for (cp = dbp;
4247                (*cp != '\0'
4248                 && (ISALPHA (*cp) || ISDIGIT (*cp) || *cp == '_' || *cp == '.'));
4249                cp++)
4250             continue;
4251           if (cp == dbp)
4252             return;
4253         }
4254       c = *cp;
4255       *cp = '\0';
4256       name = concat (dbp, name_qualifier, "");
4257       *cp = c;
4258       make_tag (name, strlen (name), TRUE,
4259                 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4260       free (name);
4261       if (c == '"')
4262         dbp = cp + 1;
4263       return;
4264     }
4265 }
4266
4267 static void
4268 Ada_funcs (inf)
4269      FILE *inf;
4270 {
4271   bool inquote = FALSE;
4272   bool skip_till_semicolumn = FALSE;
4273
4274   LOOP_ON_INPUT_LINES (inf, lb, dbp)
4275     {
4276       while (*dbp != '\0')
4277         {
4278           /* Skip a string i.e. "abcd". */
4279           if (inquote || (*dbp == '"'))
4280             {
4281               dbp = etags_strchr ((inquote) ? dbp : dbp+1, '"');
4282               if (dbp != NULL)
4283                 {
4284                   inquote = FALSE;
4285                   dbp += 1;
4286                   continue;     /* advance char */
4287                 }
4288               else
4289                 {
4290                   inquote = TRUE;
4291                   break;        /* advance line */
4292                 }
4293             }
4294
4295           /* Skip comments. */
4296           if (dbp[0] == '-' && dbp[1] == '-')
4297             break;              /* advance line */
4298
4299           /* Skip character enclosed in single quote i.e. 'a'
4300              and skip single quote starting an attribute i.e. 'Image. */
4301           if (*dbp == '\'')
4302             {
4303               dbp++ ;
4304               if (*dbp != '\0')
4305                 dbp++;
4306               continue;
4307             }
4308
4309           if (skip_till_semicolumn)
4310             {
4311               if (*dbp == ';')
4312                 skip_till_semicolumn = FALSE;
4313               dbp++;
4314               continue;         /* advance char */
4315             }
4316
4317           /* Search for beginning of a token.  */
4318           if (!begtoken (*dbp))
4319             {
4320               dbp++;
4321               continue;         /* advance char */
4322             }
4323
4324           /* We are at the beginning of a token. */
4325           switch (lowcase(*dbp))
4326             {
4327             case 'f':
4328               if (!packages_only && nocase_tail ("function"))
4329                 Ada_getit (inf, "/f");
4330               else
4331                 break;          /* from switch */
4332               continue;         /* advance char */
4333             case 'p':
4334               if (!packages_only && nocase_tail ("procedure"))
4335                 Ada_getit (inf, "/p");
4336               else if (nocase_tail ("package"))
4337                 Ada_getit (inf, "/s");
4338               else if (nocase_tail ("protected")) /* protected type */
4339                 Ada_getit (inf, "/t");
4340               else
4341                 break;          /* from switch */
4342               continue;         /* advance char */
4343
4344             case 'u':
4345               if (typedefs && !packages_only && nocase_tail ("use"))
4346                 {
4347                   /* when tagging types, avoid tagging  use type Pack.Typename;
4348                      for this, we will skip everything till a ; */
4349                   skip_till_semicolumn = TRUE;
4350                   continue;     /* advance char */
4351                 }
4352
4353             case 't':
4354               if (!packages_only && nocase_tail ("task"))
4355                 Ada_getit (inf, "/k");
4356               else if (typedefs && !packages_only && nocase_tail ("type"))
4357                 {
4358                   Ada_getit (inf, "/t");
4359                   while (*dbp != '\0')
4360                     dbp += 1;
4361                 }
4362               else
4363                 break;          /* from switch */
4364               continue;         /* advance char */
4365             }
4366
4367           /* Look for the end of the token. */
4368           while (!endtoken (*dbp))
4369             dbp++;
4370
4371         } /* advance char */
4372     } /* advance line */
4373 }
4374
4375 \f
4376 /*
4377  * Unix and microcontroller assembly tag handling
4378  * Labels:  /^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]/
4379  * Idea by Bob Weiner, Motorola Inc. (1994)
4380  */
4381 static void
4382 Asm_labels (inf)
4383      FILE *inf;
4384 {
4385   register char *cp;
4386
4387   LOOP_ON_INPUT_LINES (inf, lb, cp)
4388     {
4389       /* If first char is alphabetic or one of [_.$], test for colon
4390          following identifier. */
4391       if (ISALPHA (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
4392         {
4393           /* Read past label. */
4394           cp++;
4395           while (ISALNUM (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
4396             cp++;
4397           if (*cp == ':' || iswhite (*cp))
4398             /* Found end of label, so copy it and add it to the table. */
4399             make_tag (lb.buffer, cp - lb.buffer, TRUE,
4400                       lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4401         }
4402     }
4403 }
4404
4405 \f
4406 /*
4407  * Perl support
4408  * Perl sub names: /^sub[ \t\n]+[^ \t\n{]+/
4409  * Perl variable names: /^(my|local).../
4410  * Original code by Bart Robinson <lomew@cs.utah.edu> (1995)
4411  * Additions by Michael Ernst <mernst@alum.mit.edu> (1997)
4412  * Ideas by Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> (2001)
4413  */
4414 static void
4415 Perl_functions (inf)
4416      FILE *inf;
4417 {
4418   char *package = savestr ("main"); /* current package name */
4419   register char *cp;
4420
4421   LOOP_ON_INPUT_LINES (inf, lb, cp)
4422     {
4423       skip_spaces(cp);
4424
4425       if (LOOKING_AT (cp, "package"))
4426         {
4427           free (package);
4428           get_tag (cp, &package);
4429         }
4430       else if (LOOKING_AT (cp, "sub"))
4431         {
4432           char *pos;
4433           char *sp = cp;
4434
4435           while (!notinname (*cp))
4436             cp++;
4437           if (cp == sp)
4438             continue;           /* nothing found */
4439           if ((pos = etags_strchr (sp, ':')) != NULL
4440               && pos < cp && pos[1] == ':')
4441             /* The name is already qualified. */
4442             make_tag (sp, cp - sp, TRUE,
4443                       lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4444           else
4445             /* Qualify it. */
4446             {
4447               char savechar, *name;
4448
4449               savechar = *cp;
4450               *cp = '\0';
4451               name = concat (package, "::", sp);
4452               *cp = savechar;
4453               make_tag (name, strlen(name), TRUE,
4454                         lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4455               free (name);
4456             }
4457         }
4458        else if (globals)        /* only if we are tagging global vars */
4459         {
4460           /* Skip a qualifier, if any. */
4461           bool qual = LOOKING_AT (cp, "my") || LOOKING_AT (cp, "local");
4462           /* After "my" or "local", but before any following paren or space. */
4463           char *varstart = cp;
4464
4465           if (qual              /* should this be removed?  If yes, how? */
4466               && (*cp == '$' || *cp == '@' || *cp == '%'))
4467             {
4468               varstart += 1;
4469               do
4470                 cp++;
4471               while (ISALNUM (*cp) || *cp == '_');
4472             }
4473           else if (qual)
4474             {
4475               /* Should be examining a variable list at this point;
4476                  could insist on seeing an open parenthesis. */
4477               while (*cp != '\0' && *cp != ';' && *cp != '=' &&  *cp != ')')
4478                 cp++;
4479             }
4480           else
4481             continue;
4482
4483           make_tag (varstart, cp - varstart, FALSE,
4484                     lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4485         }
4486     }
4487 }
4488
4489
4490 /*
4491  * Python support
4492  * Look for /^[\t]*def[ \t\n]+[^ \t\n(:]+/ or /^class[ \t\n]+[^ \t\n(:]+/
4493  * Idea by Eric S. Raymond <esr@thyrsus.com> (1997)
4494  * More ideas by seb bacon <seb@jamkit.com> (2002)
4495  */
4496 static void
4497 Python_functions (inf)
4498      FILE *inf;
4499 {
4500   register char *cp;
4501
4502   LOOP_ON_INPUT_LINES (inf, lb, cp)
4503     {
4504       cp = skip_spaces (cp);
4505       if (LOOKING_AT (cp, "def") || LOOKING_AT (cp, "class"))
4506         {
4507           char *name = cp;
4508           while (!notinname (*cp) && *cp != ':')
4509             cp++;
4510           make_tag (name, cp - name, TRUE,
4511                     lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4512         }
4513     }
4514 }
4515
4516 \f
4517 /*
4518  * PHP support
4519  * Look for:
4520  *  - /^[ \t]*function[ \t\n]+[^ \t\n(]+/
4521  *  - /^[ \t]*class[ \t\n]+[^ \t\n]+/
4522  *  - /^[ \t]*define\(\"[^\"]+/
4523  * Only with --members:
4524  *  - /^[ \t]*var[ \t\n]+\$[^ \t\n=;]/
4525  * Idea by Diez B. Roggisch (2001)
4526  */
4527 static void
4528 PHP_functions (inf)
4529      FILE *inf;
4530 {
4531   register char *cp, *name;
4532   bool search_identifier = FALSE;
4533
4534   LOOP_ON_INPUT_LINES (inf, lb, cp)
4535     {
4536       cp = skip_spaces (cp);
4537       name = cp;
4538       if (search_identifier
4539           && *cp != '\0')
4540         {
4541           while (!notinname (*cp))
4542             cp++;
4543           make_tag (name, cp - name, TRUE,
4544                     lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4545           search_identifier = FALSE;
4546         }
4547       else if (LOOKING_AT (cp, "function"))
4548         {
4549           if(*cp == '&')
4550             cp = skip_spaces (cp+1);
4551           if(*cp != '\0')
4552             {
4553               name = cp;
4554               while (!notinname (*cp))
4555                 cp++;
4556               make_tag (name, cp - name, TRUE,
4557                         lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4558             }
4559           else
4560             search_identifier = TRUE;
4561         }
4562       else if (LOOKING_AT (cp, "class"))
4563         {
4564           if (*cp != '\0')
4565             {
4566               name = cp;
4567               while (*cp != '\0' && !iswhite (*cp))
4568                 cp++;
4569               make_tag (name, cp - name, FALSE,
4570                         lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4571             }
4572           else
4573             search_identifier = TRUE;
4574         }
4575       else if (strneq (cp, "define", 6)
4576                && (cp = skip_spaces (cp+6))
4577                && *cp++ == '('
4578                && (*cp == '"' || *cp == '\''))
4579         {
4580           char quote = *cp++;
4581           name = cp;
4582           while (*cp != quote && *cp != '\0')
4583             cp++;
4584           make_tag (name, cp - name, FALSE,
4585                     lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4586         }
4587       else if (members
4588                && LOOKING_AT (cp, "var")
4589                && *cp == '$')
4590         {
4591           name = cp;
4592           while (!notinname(*cp))
4593             cp++;
4594           make_tag (name, cp - name, FALSE,
4595                     lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4596         }
4597     }
4598 }
4599
4600 \f
4601 /*
4602  * Cobol tag functions
4603  * We could look for anything that could be a paragraph name.
4604  * i.e. anything that starts in column 8 is one word and ends in a full stop.
4605  * Idea by Corny de Souza (1993)
4606  */
4607 static void
4608 Cobol_paragraphs (inf)
4609      FILE *inf;
4610 {
4611   register char *bp, *ep;
4612
4613   LOOP_ON_INPUT_LINES (inf, lb, bp)
4614     {
4615       if (lb.len < 9)
4616         continue;
4617       bp += 8;
4618
4619       /* If eoln, compiler option or comment ignore whole line. */
4620       if (bp[-1] != ' ' || !ISALNUM (bp[0]))
4621         continue;
4622
4623       for (ep = bp; ISALNUM (*ep) || *ep == '-'; ep++)
4624         continue;
4625       if (*ep++ == '.')
4626         make_tag (bp, ep - bp, TRUE,
4627                   lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
4628     }
4629 }
4630
4631 \f
4632 /*
4633  * Makefile support
4634  * Ideas by Assar Westerlund <assar@sics.se> (2001)
4635  */
4636 static void
4637 Makefile_targets (inf)
4638      FILE *inf;
4639 {
4640   register char *bp;
4641
4642   LOOP_ON_INPUT_LINES (inf, lb, bp)
4643     {
4644       if (*bp == '\t' || *bp == '#')
4645         continue;
4646       while (*bp != '\0' && *bp != '=' && *bp != ':')
4647         bp++;
4648       if (*bp == ':' || (globals && *bp == '='))
4649         make_tag (lb.buffer, bp - lb.buffer, TRUE,
4650                   lb.buffer, bp - lb.buffer + 1, lineno, linecharno);
4651     }
4652 }
4653
4654 \f
4655 /*
4656  * Pascal parsing
4657  * Original code by Mosur K. Mohan (1989)
4658  *
4659  *  Locates tags for procedures & functions.  Doesn't do any type- or
4660  *  var-definitions.  It does look for the keyword "extern" or
4661  *  "forward" immediately following the procedure statement; if found,
4662  *  the tag is skipped.
4663  */
4664 static void
4665 Pascal_functions (inf)
4666      FILE *inf;
4667 {
4668   linebuffer tline;             /* mostly copied from C_entries */
4669   long save_lcno;
4670   int save_lineno, namelen, taglen;
4671   char c, *name;
4672
4673   bool                          /* each of these flags is TRUE iff: */
4674     incomment,                  /* point is inside a comment */
4675     inquote,                    /* point is inside '..' string */
4676     get_tagname,                /* point is after PROCEDURE/FUNCTION
4677                                    keyword, so next item = potential tag */
4678     found_tag,                  /* point is after a potential tag */
4679     inparms,                    /* point is within parameter-list */
4680     verify_tag;                 /* point has passed the parm-list, so the
4681                                    next token will determine whether this
4682                                    is a FORWARD/EXTERN to be ignored, or
4683                                    whether it is a real tag */
4684
4685   save_lcno = save_lineno = namelen = taglen = 0; /* keep compiler quiet */
4686   name = NULL;                  /* keep compiler quiet */
4687   dbp = lb.buffer;
4688   *dbp = '\0';
4689   linebuffer_init (&tline);
4690
4691   incomment = inquote = FALSE;
4692   found_tag = FALSE;            /* have a proc name; check if extern */
4693   get_tagname = FALSE;          /* found "procedure" keyword         */
4694   inparms = FALSE;              /* found '(' after "proc"            */
4695   verify_tag = FALSE;           /* check if "extern" is ahead        */
4696
4697
4698   while (!feof (inf))           /* long main loop to get next char */
4699     {
4700       c = *dbp++;
4701       if (c == '\0')            /* if end of line */
4702         {
4703           readline (&lb, inf);
4704           dbp = lb.buffer;
4705           if (*dbp == '\0')
4706             continue;
4707           if (!((found_tag && verify_tag)
4708                 || get_tagname))
4709             c = *dbp++;         /* only if don't need *dbp pointing
4710                                    to the beginning of the name of
4711                                    the procedure or function */
4712         }
4713       if (incomment)
4714         {
4715           if (c == '}')         /* within { } comments */
4716             incomment = FALSE;
4717           else if (c == '*' && *dbp == ')') /* within (* *) comments */
4718             {
4719               dbp++;
4720               incomment = FALSE;
4721             }
4722           continue;
4723         }
4724       else if (inquote)
4725         {
4726           if (c == '\'')
4727             inquote = FALSE;
4728           continue;
4729         }
4730       else
4731         switch (c)
4732           {
4733           case '\'':
4734             inquote = TRUE;     /* found first quote */
4735             continue;
4736           case '{':             /* found open { comment */
4737             incomment = TRUE;
4738             continue;
4739           case '(':
4740             if (*dbp == '*')    /* found open (* comment */
4741               {
4742                 incomment = TRUE;
4743                 dbp++;
4744               }
4745             else if (found_tag) /* found '(' after tag, i.e., parm-list */
4746               inparms = TRUE;
4747             continue;
4748           case ')':             /* end of parms list */
4749             if (inparms)
4750               inparms = FALSE;
4751             continue;
4752           case ';':
4753             if (found_tag && !inparms) /* end of proc or fn stmt */
4754               {
4755                 verify_tag = TRUE;
4756                 break;
4757               }
4758             continue;
4759           }
4760       if (found_tag && verify_tag && (*dbp != ' '))
4761         {
4762           /* Check if this is an "extern" declaration. */
4763           if (*dbp == '\0')
4764             continue;
4765           if (lowcase (*dbp == 'e'))
4766             {
4767               if (nocase_tail ("extern")) /* superfluous, really! */
4768                 {
4769                   found_tag = FALSE;
4770                   verify_tag = FALSE;
4771                 }
4772             }
4773           else if (lowcase (*dbp) == 'f')
4774             {
4775               if (nocase_tail ("forward")) /* check for forward reference */
4776                 {
4777                   found_tag = FALSE;
4778                   verify_tag = FALSE;
4779                 }
4780             }
4781           if (found_tag && verify_tag) /* not external proc, so make tag */
4782             {
4783               found_tag = FALSE;
4784               verify_tag = FALSE;
4785               make_tag (name, namelen, TRUE,
4786                         tline.buffer, taglen, save_lineno, save_lcno);
4787               continue;
4788             }
4789         }
4790       if (get_tagname)          /* grab name of proc or fn */
4791         {
4792           char *cp;
4793
4794           if (*dbp == '\0')
4795             continue;
4796
4797           /* Find block name. */
4798           for (cp = dbp + 1; *cp != '\0' && !endtoken (*cp); cp++)
4799             continue;
4800
4801           /* Save all values for later tagging. */
4802           linebuffer_setlen (&tline, lb.len);
4803           strcpy (tline.buffer, lb.buffer);
4804           save_lineno = lineno;
4805           save_lcno = linecharno;
4806           name = tline.buffer + (dbp - lb.buffer);
4807           namelen = cp - dbp;
4808           taglen = cp - lb.buffer + 1;
4809
4810           dbp = cp;             /* set dbp to e-o-token */
4811           get_tagname = FALSE;
4812           found_tag = TRUE;
4813           continue;
4814
4815           /* And proceed to check for "extern". */
4816         }
4817       else if (!incomment && !inquote && !found_tag)
4818         {
4819           /* Check for proc/fn keywords. */
4820           switch (lowcase (c))
4821             {
4822             case 'p':
4823               if (nocase_tail ("rocedure")) /* c = 'p', dbp has advanced */
4824                 get_tagname = TRUE;
4825               continue;
4826             case 'f':
4827               if (nocase_tail ("unction"))
4828                 get_tagname = TRUE;
4829               continue;
4830             }
4831         }
4832     } /* while not eof */
4833
4834   free (tline.buffer);
4835 }
4836
4837 \f
4838 /*
4839  * Lisp tag functions
4840  *  look for (def or (DEF, quote or QUOTE
4841  */
4842
4843 static void L_getit __P((void));
4844
4845 static void
4846 L_getit ()
4847 {
4848   if (*dbp == '\'')             /* Skip prefix quote */
4849     dbp++;
4850   else if (*dbp == '(')
4851   {
4852     dbp++;
4853     /* Try to skip "(quote " */
4854     if (!LOOKING_AT (dbp, "quote") && !LOOKING_AT (dbp, "QUOTE"))
4855       /* Ok, then skip "(" before name in (defstruct (foo)) */
4856       dbp = skip_spaces (dbp);
4857   }
4858   get_tag (dbp, NULL);
4859 }
4860
4861 static void
4862 Lisp_functions (inf)
4863      FILE *inf;
4864 {
4865   LOOP_ON_INPUT_LINES (inf, lb, dbp)
4866     {
4867       if (dbp[0] != '(')
4868         continue;
4869
4870       if (strneq (dbp+1, "def", 3) || strneq (dbp+1, "DEF", 3))
4871         {
4872           dbp = skip_non_spaces (dbp);
4873           dbp = skip_spaces (dbp);
4874           L_getit ();
4875         }
4876       else
4877         {
4878           /* Check for (foo::defmumble name-defined ... */
4879           do
4880             dbp++;
4881           while (!notinname (*dbp) && *dbp != ':');
4882           if (*dbp == ':')
4883             {
4884               do
4885                 dbp++;
4886               while (*dbp == ':');
4887
4888               if (strneq (dbp, "def", 3) || strneq (dbp, "DEF", 3))
4889                 {
4890                   dbp = skip_non_spaces (dbp);
4891                   dbp = skip_spaces (dbp);
4892                   L_getit ();
4893                 }
4894             }
4895         }
4896     }
4897 }
4898
4899 \f
4900 /*
4901  * Postscript tag functions
4902  * Just look for lines where the first character is '/'
4903  * Also look at "defineps" for PSWrap
4904  * Ideas by:
4905  *   Richard Mlynarik <mly@adoc.xerox.com> (1997)
4906  *   Masatake Yamato <masata-y@is.aist-nara.ac.jp> (1999)
4907  */
4908 static void
4909 PS_functions (inf)
4910      FILE *inf;
4911 {
4912   register char *bp, *ep;
4913
4914   LOOP_ON_INPUT_LINES (inf, lb, bp)
4915     {
4916       if (bp[0] == '/')
4917         {
4918           for (ep = bp+1;
4919                *ep != '\0' && *ep != ' ' && *ep != '{';
4920                ep++)
4921             continue;
4922           make_tag (bp, ep - bp, TRUE,
4923                     lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
4924         }
4925       else if (LOOKING_AT (bp, "defineps"))
4926         get_tag (bp, NULL);
4927     }
4928 }
4929
4930 \f
4931 /*
4932  * Scheme tag functions
4933  * look for (def... xyzzy
4934  *          (def... (xyzzy
4935  *          (def ... ((...(xyzzy ....
4936  *          (set! xyzzy
4937  * Original code by Ken Haase (1985?)
4938  */
4939
4940 static void
4941 Scheme_functions (inf)
4942      FILE *inf;
4943 {
4944   register char *bp;
4945
4946   LOOP_ON_INPUT_LINES (inf, lb, bp)
4947     {
4948       if (strneq (bp, "(def", 4) || strneq (bp, "(DEF", 4))
4949         {
4950           bp = skip_non_spaces (bp+4);
4951           /* Skip over open parens and white space */
4952           while (notinname (*bp))
4953             bp++;
4954           get_tag (bp, NULL);
4955         }
4956       if (LOOKING_AT (bp, "(SET!") || LOOKING_AT (bp, "(set!"))
4957         get_tag (bp, NULL);
4958     }
4959 }
4960
4961 \f
4962 /* Find tags in TeX and LaTeX input files.  */
4963
4964 /* TEX_toktab is a table of TeX control sequences that define tags.
4965  * Each entry records one such control sequence.
4966  *
4967  * Original code from who knows whom.
4968  * Ideas by:
4969  *   Stefan Monnier (2002)
4970  */
4971
4972 static linebuffer *TEX_toktab = NULL; /* Table with tag tokens */
4973
4974 /* Default set of control sequences to put into TEX_toktab.
4975    The value of environment var TEXTAGS is prepended to this.  */
4976 static char *TEX_defenv = "\
4977 :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem\
4978 :part:appendix:entry:index:def\
4979 :newcommand:renewcommand:newenvironment:renewenvironment";
4980
4981 static void TEX_mode __P((FILE *));
4982 static void TEX_decode_env __P((char *, char *));
4983
4984 static char TEX_esc = '\\';
4985 static char TEX_opgrp = '{';
4986 static char TEX_clgrp = '}';
4987
4988 /*
4989  * TeX/LaTeX scanning loop.
4990  */
4991 static void
4992 TeX_commands (inf)
4993      FILE *inf;
4994 {
4995   char *cp;
4996   linebuffer *key;
4997
4998   /* Select either \ or ! as escape character.  */
4999   TEX_mode (inf);
5000
5001   /* Initialize token table once from environment. */
5002   if (TEX_toktab == NULL)
5003     TEX_decode_env ("TEXTAGS", TEX_defenv);
5004
5005   LOOP_ON_INPUT_LINES (inf, lb, cp)
5006     {
5007       /* Look at each TEX keyword in line. */
5008       for (;;)
5009         {
5010           /* Look for a TEX escape. */
5011           while (*cp++ != TEX_esc)
5012             if (cp[-1] == '\0' || cp[-1] == '%')
5013               goto tex_next_line;
5014
5015           for (key = TEX_toktab; key->buffer != NULL; key++)
5016             if (strneq (cp, key->buffer, key->len))
5017               {
5018                 register char *p;
5019                 int namelen, linelen;
5020                 bool opgrp = FALSE;
5021
5022                 cp = skip_spaces (cp + key->len);
5023                 if (*cp == TEX_opgrp)
5024                   {
5025                     opgrp = TRUE;
5026                     cp++;
5027                   }
5028                 for (p = cp;
5029                      (!iswhite (*p) && *p != '#' &&
5030                       *p != TEX_opgrp && *p != TEX_clgrp);
5031                      p++)
5032                   continue;
5033                 namelen = p - cp;
5034                 linelen = lb.len;
5035                 if (!opgrp || *p == TEX_clgrp)
5036                   {
5037                     while (*p != '\0' && *p != TEX_opgrp && *p != TEX_clgrp)
5038                       *p++;
5039                     linelen = p - lb.buffer + 1;
5040                   }
5041                 make_tag (cp, namelen, TRUE,
5042                           lb.buffer, linelen, lineno, linecharno);
5043                 goto tex_next_line; /* We only tag a line once */
5044               }
5045         }
5046     tex_next_line:
5047       ;
5048     }
5049 }
5050
5051 #define TEX_LESC '\\'
5052 #define TEX_SESC '!'
5053
5054 /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
5055    chars accordingly. */
5056 static void
5057 TEX_mode (inf)
5058      FILE *inf;
5059 {
5060   int c;
5061
5062   while ((c = getc (inf)) != EOF)
5063     {
5064       /* Skip to next line if we hit the TeX comment char. */
5065       if (c == '%')
5066         while (c != '\n')
5067           c = getc (inf);
5068       else if (c == TEX_LESC || c == TEX_SESC )
5069         break;
5070     }
5071
5072   if (c == TEX_LESC)
5073     {
5074       TEX_esc = TEX_LESC;
5075       TEX_opgrp = '{';
5076       TEX_clgrp = '}';
5077     }
5078   else
5079     {
5080       TEX_esc = TEX_SESC;
5081       TEX_opgrp = '<';
5082       TEX_clgrp = '>';
5083     }
5084   /* If the input file is compressed, inf is a pipe, and rewind may fail.
5085      No attempt is made to correct the situation. */
5086   rewind (inf);
5087 }
5088
5089 /* Read environment and prepend it to the default string.
5090    Build token table. */
5091 static void
5092 TEX_decode_env (evarname, defenv)
5093      char *evarname;
5094      char *defenv;
5095 {
5096   register char *env, *p;
5097   int i, len;
5098
5099   /* Append default string to environment. */
5100   env = getenv (evarname);
5101   if (!env)
5102     env = defenv;
5103   else
5104     {
5105       char *oldenv = env;
5106       env = concat (oldenv, defenv, "");
5107     }
5108
5109   /* Allocate a token table */
5110   for (len = 1, p = env; p;)
5111     if ((p = etags_strchr (p, ':')) && *++p != '\0')
5112       len++;
5113   TEX_toktab = xnew (len, linebuffer);
5114
5115   /* Unpack environment string into token table. Be careful about */
5116   /* zero-length strings (leading ':', "::" and trailing ':') */
5117   for (i = 0; *env != '\0';)
5118     {
5119       p = etags_strchr (env, ':');
5120       if (!p)                   /* End of environment string. */
5121         p = env + strlen (env);
5122       if (p - env > 0)
5123         {                       /* Only non-zero strings. */
5124           TEX_toktab[i].buffer = savenstr (env, p - env);
5125           TEX_toktab[i].len = p - env;
5126           i++;
5127         }
5128       if (*p)
5129         env = p + 1;
5130       else
5131         {
5132           TEX_toktab[i].buffer = NULL; /* Mark end of table. */
5133           TEX_toktab[i].len = 0;
5134           break;
5135         }
5136     }
5137 }
5138
5139 \f
5140 /* Texinfo support.  Dave Love, Mar. 2000.  */
5141 static void
5142 Texinfo_nodes (inf)
5143      FILE * inf;
5144 {
5145   char *cp, *start;
5146   LOOP_ON_INPUT_LINES (inf, lb, cp)
5147     if (LOOKING_AT (cp, "@node"))
5148       {
5149         start = cp;
5150         while (*cp != '\0' && *cp != ',')
5151           cp++;
5152         make_tag (start, cp - start, TRUE,
5153                   lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
5154       }
5155 }
5156
5157 \f
5158 /* Similar to LOOKING_AT but does not use notinname, does not skip */
5159 #define LOOKING_AT_NOCASE(cp, kw)       /* kw is a constant string */   \
5160   (strncaseeq ((cp), kw, sizeof(kw)-1)  /* cp points at kw */           \
5161    && ((cp) += sizeof(kw)-1))           /* skip spaces */
5162
5163 /*
5164  * HTML support.
5165  * Contents of <title>, <h1>, <h2>, <h3> are tags.
5166  * Contents of <a name=xxx> are tags with name xxx.
5167  *
5168  * Francesco Potortì, 2002.
5169  */
5170 static void
5171 HTML_labels (inf)
5172      FILE * inf;
5173 {
5174   bool getnext = FALSE;         /* next text outside of HTML tags is a tag */
5175   bool skiptag = FALSE;         /* skip to the end of the current HTML tag */
5176   bool intag = FALSE;           /* inside an html tag, looking for ID= */
5177   bool inanchor = FALSE;        /* when INTAG, is an anchor, look for NAME= */
5178   char *end;
5179
5180
5181   linebuffer_setlen (&token_name, 0); /* no name in buffer */
5182
5183   LOOP_ON_INPUT_LINES (inf, lb, dbp)
5184     for (;;)                    /* loop on the same line */
5185       {
5186         if (skiptag)            /* skip HTML tag */
5187           {
5188             while (*dbp != '\0' && *dbp != '>')
5189               dbp++;
5190             if (*dbp == '>')
5191               {
5192                 dbp += 1;
5193                 skiptag = FALSE;
5194                 continue;       /* look on the same line */
5195               }
5196             break;              /* go to next line */
5197           }
5198
5199         else if (intag) /* look for "name=" or "id=" */
5200           {
5201             while (*dbp != '\0' && *dbp != '>'
5202                    && lowcase (*dbp) != 'n' && lowcase (*dbp) != 'i')
5203               dbp++;
5204             if (*dbp == '\0')
5205               break;            /* go to next line */
5206             if (*dbp == '>')
5207               {
5208                 dbp += 1;
5209                 intag = FALSE;
5210                 continue;       /* look on the same line */
5211               }
5212             if ((inanchor && LOOKING_AT_NOCASE (dbp, "name="))
5213                 || LOOKING_AT_NOCASE (dbp, "id="))
5214               {
5215                 bool quoted = (dbp[0] == '"');
5216
5217                 if (quoted)
5218                   for (end = ++dbp; *end != '\0' && *end != '"'; end++)
5219                     continue;
5220                 else
5221                   for (end = dbp; *end != '\0' && intoken (*end); end++)
5222                     continue;
5223                 linebuffer_setlen (&token_name, end - dbp);
5224                 strncpy (token_name.buffer, dbp, end - dbp);
5225                 token_name.buffer[end - dbp] = '\0';
5226
5227                 dbp = end;
5228                 intag = FALSE;  /* we found what we looked for */
5229                 skiptag = TRUE; /* skip to the end of the tag */
5230                 getnext = TRUE; /* then grab the text */
5231                 continue;       /* look on the same line */
5232               }
5233             dbp += 1;
5234           }
5235
5236         else if (getnext)       /* grab next tokens and tag them */
5237           {
5238             dbp = skip_spaces (dbp);
5239             if (*dbp == '\0')
5240               break;            /* go to next line */
5241             if (*dbp == '<')
5242               {
5243                 intag = TRUE;
5244                 inanchor = (lowcase (dbp[1]) == 'a' && !intoken (dbp[2]));
5245                 continue;       /* look on the same line */
5246               }
5247
5248             for (end = dbp + 1; *end != '\0' && *end != '<'; end++)
5249               continue;
5250             make_tag (token_name.buffer, token_name.len, TRUE,
5251                       dbp, end - dbp, lineno, linecharno);
5252             linebuffer_setlen (&token_name, 0); /* no name in buffer */
5253             getnext = FALSE;
5254             break;              /* go to next line */
5255           }
5256
5257         else                    /* look for an interesting HTML tag */
5258           {
5259             while (*dbp != '\0' && *dbp != '<')
5260               dbp++;
5261             if (*dbp == '\0')
5262               break;            /* go to next line */
5263             intag = TRUE;
5264             if (lowcase (dbp[1]) == 'a' && !intoken (dbp[2]))
5265               {
5266                 inanchor = TRUE;
5267                 continue;       /* look on the same line */
5268               }
5269             else if (LOOKING_AT_NOCASE (dbp, "<title>")
5270                      || LOOKING_AT_NOCASE (dbp, "<h1>")
5271                      || LOOKING_AT_NOCASE (dbp, "<h2>")
5272                      || LOOKING_AT_NOCASE (dbp, "<h3>"))
5273               {
5274                 intag = FALSE;
5275                 getnext = TRUE;
5276                 continue;       /* look on the same line */
5277               }
5278             dbp += 1;
5279           }
5280       }
5281 }
5282
5283 \f
5284 /*
5285  * Prolog support
5286  *
5287  * Assumes that the predicate or rule starts at column 0.
5288  * Only the first clause of a predicate or rule is added.
5289  * Original code by Sunichirou Sugou (1989)
5290  * Rewritten by Anders Lindgren (1996)
5291  */
5292 static int prolog_pr __P((char *, char *));
5293 static void prolog_skip_comment __P((linebuffer *, FILE *));
5294 static int prolog_atom __P((char *, int));
5295
5296 static void
5297 Prolog_functions (inf)
5298      FILE *inf;
5299 {
5300   char *cp, *last;
5301   int len;
5302   int allocated;
5303
5304   allocated = 0;
5305   len = 0;
5306   last = NULL;
5307
5308   LOOP_ON_INPUT_LINES (inf, lb, cp)
5309     {
5310       if (cp[0] == '\0')        /* Empty line */
5311         continue;
5312       else if (iswhite (cp[0])) /* Not a predicate */
5313         continue;
5314       else if (cp[0] == '/' && cp[1] == '*')    /* comment. */
5315         prolog_skip_comment (&lb, inf);
5316       else if ((len = prolog_pr (cp, last)) > 0)
5317         {
5318           /* Predicate or rule.  Store the function name so that we
5319              only generate a tag for the first clause.  */
5320           if (last == NULL)
5321             last = xnew(len + 1, char);
5322           else if (len + 1 > allocated)
5323             xrnew (last, len + 1, char);
5324           allocated = len + 1;
5325           strncpy (last, cp, len);
5326           last[len] = '\0';
5327         }
5328     }
5329 }
5330
5331
5332 static void
5333 prolog_skip_comment (plb, inf)
5334      linebuffer *plb;
5335      FILE *inf;
5336 {
5337   char *cp;
5338
5339   do
5340     {
5341       for (cp = plb->buffer; *cp != '\0'; cp++)
5342         if (cp[0] == '*' && cp[1] == '/')
5343           return;
5344       readline (plb, inf);
5345     }
5346   while (!feof(inf));
5347 }
5348
5349 /*
5350  * A predicate or rule definition is added if it matches:
5351  *     <beginning of line><Prolog Atom><whitespace>(
5352  * or  <beginning of line><Prolog Atom><whitespace>:-
5353  *
5354  * It is added to the tags database if it doesn't match the
5355  * name of the previous clause header.
5356  *
5357  * Return the size of the name of the predicate or rule, or 0 if no
5358  * header was found.
5359  */
5360 static int
5361 prolog_pr (s, last)
5362      char *s;
5363      char *last;                /* Name of last clause. */
5364 {
5365   int pos;
5366   int len;
5367
5368   pos = prolog_atom (s, 0);
5369   if (pos < 1)
5370     return 0;
5371
5372   len = pos;
5373   pos = skip_spaces (s + pos) - s;
5374
5375   if ((s[pos] == '.'
5376        || (s[pos] == '(' && (pos += 1))
5377        || (s[pos] == ':' && s[pos + 1] == '-' && (pos += 2)))
5378       && (last == NULL          /* save only the first clause */
5379           || len != strlen (last)
5380           || !strneq (s, last, len)))
5381         {
5382           make_tag (s, len, TRUE, s, pos, lineno, linecharno);
5383           return len;
5384         }
5385   else
5386     return 0;
5387 }
5388
5389 /*
5390  * Consume a Prolog atom.
5391  * Return the number of bytes consumed, or -1 if there was an error.
5392  *
5393  * A prolog atom, in this context, could be one of:
5394  * - An alphanumeric sequence, starting with a lower case letter.
5395  * - A quoted arbitrary string. Single quotes can escape themselves.
5396  *   Backslash quotes everything.
5397  */
5398 static int
5399 prolog_atom (s, pos)
5400      char *s;
5401      int pos;
5402 {
5403   int origpos;
5404
5405   origpos = pos;
5406
5407   if (ISLOWER(s[pos]) || (s[pos] == '_'))
5408     {
5409       /* The atom is unquoted. */
5410       pos++;
5411       while (ISALNUM(s[pos]) || (s[pos] == '_'))
5412         {
5413           pos++;
5414         }
5415       return pos - origpos;
5416     }
5417   else if (s[pos] == '\'')
5418     {
5419       pos++;
5420
5421       for (;;)
5422         {
5423           if (s[pos] == '\'')
5424             {
5425               pos++;
5426               if (s[pos] != '\'')
5427                 break;
5428               pos++;            /* A double quote */
5429             }
5430           else if (s[pos] == '\0')
5431             /* Multiline quoted atoms are ignored. */
5432             return -1;
5433           else if (s[pos] == '\\')
5434             {
5435               if (s[pos+1] == '\0')
5436                 return -1;
5437               pos += 2;
5438             }
5439           else
5440             pos++;
5441         }
5442       return pos - origpos;
5443     }
5444   else
5445     return -1;
5446 }
5447
5448 \f
5449 /*
5450  * Support for Erlang
5451  *
5452  * Generates tags for functions, defines, and records.
5453  * Assumes that Erlang functions start at column 0.
5454  * Original code by Anders Lindgren (1996)
5455  */
5456 static int erlang_func __P((char *, char *));
5457 static void erlang_attribute __P((char *));
5458 static int erlang_atom __P((char *));
5459
5460 static void
5461 Erlang_functions (inf)
5462      FILE *inf;
5463 {
5464   char *cp, *last;
5465   int len;
5466   int allocated;
5467
5468   allocated = 0;
5469   len = 0;
5470   last = NULL;
5471
5472   LOOP_ON_INPUT_LINES (inf, lb, cp)
5473     {
5474       if (cp[0] == '\0')        /* Empty line */
5475         continue;
5476       else if (iswhite (cp[0])) /* Not function nor attribute */
5477         continue;
5478       else if (cp[0] == '%')    /* comment */
5479         continue;
5480       else if (cp[0] == '"')    /* Sometimes, strings start in column one */
5481         continue;
5482       else if (cp[0] == '-')    /* attribute, e.g. "-define" */
5483         {
5484           erlang_attribute (cp);
5485           last = NULL;
5486         }
5487       else if ((len = erlang_func (cp, last)) > 0)
5488         {
5489           /*
5490            * Function.  Store the function name so that we only
5491            * generates a tag for the first clause.
5492            */
5493           if (last == NULL)
5494             last = xnew (len + 1, char);
5495           else if (len + 1 > allocated)
5496             xrnew (last, len + 1, char);
5497           allocated = len + 1;
5498           strncpy (last, cp, len);
5499           last[len] = '\0';
5500         }
5501     }
5502 }
5503
5504
5505 /*
5506  * A function definition is added if it matches:
5507  *     <beginning of line><Erlang Atom><whitespace>(
5508  *
5509  * It is added to the tags database if it doesn't match the
5510  * name of the previous clause header.
5511  *
5512  * Return the size of the name of the function, or 0 if no function
5513  * was found.
5514  */
5515 static int
5516 erlang_func (s, last)
5517      char *s;
5518      char *last;                /* Name of last clause. */
5519 {
5520   int pos;
5521   int len;
5522
5523   pos = erlang_atom (s);
5524   if (pos < 1)
5525     return 0;
5526
5527   len = pos;
5528   pos = skip_spaces (s + pos) - s;
5529
5530   /* Save only the first clause. */
5531   if (s[pos++] == '('
5532       && (last == NULL
5533           || len != (int)strlen (last)
5534           || !strneq (s, last, len)))
5535         {
5536           make_tag (s, len, TRUE, s, pos, lineno, linecharno);
5537           return len;
5538         }
5539
5540   return 0;
5541 }
5542
5543
5544 /*
5545  * Handle attributes.  Currently, tags are generated for defines
5546  * and records.
5547  *
5548  * They are on the form:
5549  * -define(foo, bar).
5550  * -define(Foo(M, N), M+N).
5551  * -record(graph, {vtab = notable, cyclic = true}).
5552  */
5553 static void
5554 erlang_attribute (s)
5555      char *s;
5556 {
5557   char *cp = s;
5558
5559   if ((LOOKING_AT (cp, "-define") || LOOKING_AT (cp, "-record"))
5560       && *cp++ == '(')
5561     {
5562       int len = erlang_atom (skip_spaces (cp));
5563       if (len > 0)
5564         make_tag (cp, len, TRUE, s, cp + len - s, lineno, linecharno);
5565     }
5566   return;
5567 }
5568
5569
5570 /*
5571  * Consume an Erlang atom (or variable).
5572  * Return the number of bytes consumed, or -1 if there was an error.
5573  */
5574 static int
5575 erlang_atom (s)
5576      char *s;
5577 {
5578   int pos = 0;
5579
5580   if (ISALPHA (s[pos]) || s[pos] == '_')
5581     {
5582       /* The atom is unquoted. */
5583       do
5584         pos++;
5585       while (ISALNUM (s[pos]) || s[pos] == '_');
5586     }
5587   else if (s[pos] == '\'')
5588     {
5589       for (pos++; s[pos] != '\''; pos++)
5590         if (s[pos] == '\0'      /* multiline quoted atoms are ignored */
5591             || (s[pos] == '\\' && s[++pos] == '\0'))
5592           return 0;
5593       pos++;
5594     }
5595
5596   return pos;
5597 }
5598
5599 \f
5600 #ifdef ETAGS_REGEXPS
5601
5602 static char *scan_separators __P((char *));
5603 static void add_regex __P((char *, language *));
5604 static char *substitute __P((char *, char *, struct re_registers *));
5605
5606 /*
5607  * Take a string like "/blah/" and turn it into "blah", verifying
5608  * that the first and last characters are the same, and handling
5609  * quoted separator characters.  Actually, stops on the occurrence of
5610  * an unquoted separator.  Also process \t, \n, etc. and turn into
5611  * appropriate characters. Works in place.  Null terminates name string.
5612  * Returns pointer to terminating separator, or NULL for
5613  * unterminated regexps.
5614  */
5615 static char *
5616 scan_separators (name)
5617      char *name;
5618 {
5619   char sep = name[0];
5620   char *copyto = name;
5621   bool quoted = FALSE;
5622
5623   for (++name; *name != '\0'; ++name)
5624     {
5625       if (quoted)
5626         {
5627           switch (*name)
5628             {
5629             case 'a': *copyto++ = '\007'; break; /* BEL (bell)           */
5630             case 'b': *copyto++ = '\b'; break;   /* BS (back space)      */
5631             case 'd': *copyto++ = 0177; break;   /* DEL (delete)         */
5632             case 'e': *copyto++ = 033; break;    /* ESC (delete)         */
5633             case 'f': *copyto++ = '\f'; break;   /* FF (form feed)       */
5634             case 'n': *copyto++ = '\n'; break;   /* NL (new line)        */
5635             case 'r': *copyto++ = '\r'; break;   /* CR (carriage return) */
5636             case 't': *copyto++ = '\t'; break;   /* TAB (horizontal tab) */
5637             case 'v': *copyto++ = '\v'; break;   /* VT (vertical tab)    */
5638             default:
5639               if (*name == sep)
5640                 *copyto++ = sep;
5641               else
5642                 {
5643                   /* Something else is quoted, so preserve the quote. */
5644                   *copyto++ = '\\';
5645                   *copyto++ = *name;
5646                 }
5647               break;
5648             }
5649           quoted = FALSE;
5650         }
5651       else if (*name == '\\')
5652         quoted = TRUE;
5653       else if (*name == sep)
5654         break;
5655       else
5656         *copyto++ = *name;
5657     }
5658   if (*name != sep)
5659     name = NULL;                /* signal unterminated regexp */
5660
5661   /* Terminate copied string. */
5662   *copyto = '\0';
5663   return name;
5664 }
5665
5666 /* Look at the argument of --regex or --no-regex and do the right
5667    thing.  Same for each line of a regexp file. */
5668 static void
5669 analyse_regex (regex_arg)
5670      char *regex_arg;
5671 {
5672   if (regex_arg == NULL)
5673     {
5674       free_regexps ();          /* --no-regex: remove existing regexps */
5675       return;
5676     }
5677
5678   /* A real --regexp option or a line in a regexp file. */
5679   switch (regex_arg[0])
5680     {
5681       /* Comments in regexp file or null arg to --regex. */
5682     case '\0':
5683     case ' ':
5684     case '\t':
5685       break;
5686
5687       /* Read a regex file.  This is recursive and may result in a
5688          loop, which will stop when the file descriptors are exhausted. */
5689     case '@':
5690       {
5691         FILE *regexfp;
5692         linebuffer regexbuf;
5693         char *regexfile = regex_arg + 1;
5694
5695         /* regexfile is a file containing regexps, one per line. */
5696         regexfp = fopen (regexfile, "r");
5697         if (regexfp == NULL)
5698           {
5699             pfatal (regexfile);
5700             return;
5701           }
5702         linebuffer_init (&regexbuf);
5703         while (readline_internal (&regexbuf, regexfp) > 0)
5704           analyse_regex (regexbuf.buffer);
5705         free (regexbuf.buffer);
5706         fclose (regexfp);
5707       }
5708       break;
5709
5710       /* Regexp to be used for a specific language only. */
5711     case '{':
5712       {
5713         language *lang;
5714         char *lang_name = regex_arg + 1;
5715         char *cp;
5716
5717         for (cp = lang_name; *cp != '}'; cp++)
5718           if (*cp == '\0')
5719             {
5720               error ("unterminated language name in regex: %s", regex_arg);
5721               return;
5722             }
5723         *cp++ = '\0';
5724         lang = get_language_from_langname (lang_name);
5725         if (lang == NULL)
5726           return;
5727         add_regex (cp, lang);
5728       }
5729       break;
5730
5731       /* Regexp to be used for any language. */
5732     default:
5733       add_regex (regex_arg, NULL);
5734       break;
5735     }
5736 }
5737
5738 /* Separate the regexp pattern, compile it,
5739    and care for optional name and modifiers. */
5740 static void
5741 add_regex (regexp_pattern, lang)
5742      char *regexp_pattern;
5743      language *lang;
5744 {
5745   static struct re_pattern_buffer zeropattern;
5746   char sep, *pat, *name, *modifiers;
5747   const char *err;
5748   struct re_pattern_buffer *patbuf;
5749   regexp *rp;
5750   bool
5751     force_explicit_name = TRUE, /* do not use implicit tag names */
5752     ignore_case = FALSE,        /* case is significant */
5753     multi_line = FALSE,         /* matches are done one line at a time */
5754     single_line = FALSE;        /* dot does not match newline */
5755
5756
5757   if (strlen(regexp_pattern) < 3)
5758     {
5759       error ("null regexp", (char *)NULL);
5760       return;
5761     }
5762   sep = regexp_pattern[0];
5763   name = scan_separators (regexp_pattern);
5764   if (name == NULL)
5765     {
5766       error ("%s: unterminated regexp", regexp_pattern);
5767       return;
5768     }
5769   if (name[1] == sep)
5770     {
5771       error ("null name for regexp \"%s\"", regexp_pattern);
5772       return;
5773     }
5774   modifiers = scan_separators (name);
5775   if (modifiers == NULL)        /* no terminating separator --> no name */
5776     {
5777       modifiers = name;
5778       name = "";
5779     }
5780   else
5781     modifiers += 1;             /* skip separator */
5782
5783   /* Parse regex modifiers. */
5784   for (; modifiers[0] != '\0'; modifiers++)
5785     switch (modifiers[0])
5786       {
5787       case 'N':
5788         if (modifiers == name)
5789           error ("forcing explicit tag name but no name, ignoring", NULL);
5790         force_explicit_name = TRUE;
5791         break;
5792       case 'i':
5793         ignore_case = TRUE;
5794         break;
5795       case 's':
5796         single_line = TRUE;
5797         /* FALLTHRU */
5798       case 'm':
5799         multi_line = TRUE;
5800         need_filebuf = TRUE;
5801         break;
5802       default:
5803         {
5804           char wrongmod [2];
5805           wrongmod[0] = modifiers[0];
5806           wrongmod[1] = '\0';
5807           error ("invalid regexp modifier `%s', ignoring", wrongmod);
5808         }
5809         break;
5810       }
5811
5812   patbuf = xnew (1, struct re_pattern_buffer);
5813   *patbuf = zeropattern;
5814   if (ignore_case)
5815     {
5816       static char lc_trans[CHARS];
5817       int i;
5818       for (i = 0; i < CHARS; i++)
5819         lc_trans[i] = lowcase (i);
5820       patbuf->translate = lc_trans;     /* translation table to fold case  */
5821     }
5822
5823   if (multi_line)
5824     pat = concat ("^", regexp_pattern, ""); /* anchor to beginning of line */
5825   else
5826     pat = regexp_pattern;
5827
5828   if (single_line)
5829     re_set_syntax (RE_SYNTAX_EMACS | RE_DOT_NEWLINE);
5830   else
5831     re_set_syntax (RE_SYNTAX_EMACS);
5832
5833   err = re_compile_pattern (pat, strlen (regexp_pattern), patbuf);
5834   if (multi_line)
5835     free (pat);
5836   if (err != NULL)
5837     {
5838       error ("%s while compiling pattern", err);
5839       return;
5840     }
5841
5842   rp = p_head;
5843   p_head = xnew (1, regexp);
5844   p_head->pattern = savestr (regexp_pattern);
5845   p_head->p_next = rp;
5846   p_head->lang = lang;
5847   p_head->pat = patbuf;
5848   p_head->name = savestr (name);
5849   p_head->error_signaled = FALSE;
5850   p_head->force_explicit_name = force_explicit_name;
5851   p_head->ignore_case = ignore_case;
5852   p_head->multi_line = multi_line;
5853 }
5854
5855 /*
5856  * Do the substitutions indicated by the regular expression and
5857  * arguments.
5858  */
5859 static char *
5860 substitute (in, out, regs)
5861      char *in, *out;
5862      struct re_registers *regs;
5863 {
5864   char *result, *t;
5865   int size, dig, diglen;
5866
5867   result = NULL;
5868   size = strlen (out);
5869
5870   /* Pass 1: figure out how much to allocate by finding all \N strings. */
5871   if (out[size - 1] == '\\')
5872     fatal ("pattern error in \"%s\"", out);
5873   for (t = etags_strchr (out, '\\');
5874        t != NULL;
5875        t = etags_strchr (t + 2, '\\'))
5876     if (ISDIGIT (t[1]))
5877       {
5878         dig = t[1] - '0';
5879         diglen = regs->end[dig] - regs->start[dig];
5880         size += diglen - 2;
5881       }
5882     else
5883       size -= 1;
5884
5885   /* Allocate space and do the substitutions. */
5886   assert (size >= 0);
5887   result = xnew (size + 1, char);
5888
5889   for (t = result; *out != '\0'; out++)
5890     if (*out == '\\' && ISDIGIT (*++out))
5891       {
5892         dig = *out - '0';
5893         diglen = regs->end[dig] - regs->start[dig];
5894         strncpy (t, in + regs->start[dig], diglen);
5895         t += diglen;
5896       }
5897     else
5898       *t++ = *out;
5899   *t = '\0';
5900
5901   assert (t <= result + size);
5902   assert (t - result == (int)strlen (result));
5903
5904   return result;
5905 }
5906
5907 /* Deallocate all regexps. */
5908 static void
5909 free_regexps ()
5910 {
5911   regexp *rp;
5912   while (p_head != NULL)
5913     {
5914       rp = p_head->p_next;
5915       free (p_head->pattern);
5916       free (p_head->name);
5917       free (p_head);
5918       p_head = rp;
5919     }
5920   return;
5921 }
5922
5923 /*
5924  * Reads the whole file as a single string from `filebuf' and looks for
5925  * multi-line regular expressions, creating tags on matches.
5926  * readline already dealt with normal regexps.
5927  *
5928  * Idea by Ben Wing <ben@666.com> (2002).
5929  */
5930 static void
5931 regex_tag_multiline ()
5932 {
5933   char *buffer = filebuf.buffer;
5934   regexp *rp;
5935   char *name;
5936
5937   for (rp = p_head; rp != NULL; rp = rp->p_next)
5938     {
5939       int match = 0;
5940
5941       if (!rp->multi_line)
5942         continue;               /* skip normal regexps */
5943
5944       /* Generic initialisations before parsing file from memory. */
5945       lineno = 1;               /* reset global line number */
5946       charno = 0;               /* reset global char number */
5947       linecharno = 0;           /* reset global char number of line start */
5948
5949       /* Only use generic regexps or those for the current language. */
5950       if (rp->lang != NULL && rp->lang != curfdp->lang)
5951         continue;
5952
5953       while (match >= 0 && match < filebuf.len)
5954         {
5955           match = re_search (rp->pat, buffer, filebuf.len, charno,
5956                              filebuf.len - match, &rp->regs);
5957           switch (match)
5958             {
5959             case -2:
5960               /* Some error. */
5961               if (!rp->error_signaled)
5962                 {
5963                   error ("regexp stack overflow while matching \"%s\"",
5964                          rp->pattern);
5965                   rp->error_signaled = TRUE;
5966                 }
5967               break;
5968             case -1:
5969               /* No match. */
5970               break;
5971             default:
5972               if (match == rp->regs.end[0])
5973                 {
5974                   if (!rp->error_signaled)
5975                     {
5976                       error ("regexp matches the empty string: \"%s\"",
5977                              rp->pattern);
5978                       rp->error_signaled = TRUE;
5979                     }
5980                   match = -3;   /* exit from while loop */
5981                   break;
5982                 }
5983
5984               /* Match occurred.  Construct a tag. */
5985               while (charno < rp->regs.end[0])
5986                 if (buffer[charno++] == '\n')
5987                   lineno++, linecharno = charno;
5988               name = rp->name;
5989               if (name[0] == '\0')
5990                 name = NULL;
5991               else /* make a named tag */
5992                 name = substitute (buffer, rp->name, &rp->regs);
5993               if (rp->force_explicit_name)
5994                 /* Force explicit tag name, if a name is there. */
5995                 pfnote (name, TRUE, buffer + linecharno,
5996                         charno - linecharno + 1, lineno, linecharno);
5997               else
5998                 make_tag (name, strlen (name), TRUE, buffer + linecharno,
5999                           charno - linecharno + 1, lineno, linecharno);
6000               break;
6001             }
6002         }
6003     }
6004 }
6005
6006 #endif /* ETAGS_REGEXPS */
6007
6008 \f
6009 static bool
6010 nocase_tail (cp)
6011      char *cp;
6012 {
6013   register int len = 0;
6014
6015   while (*cp != '\0' && lowcase (*cp) == lowcase (dbp[len]))
6016     cp++, len++;
6017   if (*cp == '\0' && !intoken (dbp[len]))
6018     {
6019       dbp += len;
6020       return TRUE;
6021     }
6022   return FALSE;
6023 }
6024
6025 static void
6026 get_tag (bp, namepp)
6027      register char *bp;
6028      char **namepp;
6029 {
6030   register char *cp = bp;
6031
6032   if (*bp != '\0')
6033     {
6034       /* Go till you get to white space or a syntactic break */
6035       for (cp = bp + 1; !notinname (*cp); cp++)
6036         continue;
6037       make_tag (bp, cp - bp, TRUE,
6038                 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
6039     }
6040
6041   if (namepp != NULL)
6042     *namepp = savenstr (bp, cp - bp);
6043 }
6044
6045 /*
6046  * Read a line of text from `stream' into `lbp', excluding the
6047  * newline or CR-NL, if any.  Return the number of characters read from
6048  * `stream', which is the length of the line including the newline.
6049  *
6050  * On DOS or Windows we do not count the CR character, if any before the
6051  * NL, in the returned length; this mirrors the behavior of Emacs on those
6052  * platforms (for text files, it translates CR-NL to NL as it reads in the
6053  * file).
6054  *
6055  * If multi-line regular expressions are requested, each line read is
6056  * appended to `filebuf'.
6057  */
6058 static long
6059 readline_internal (lbp, stream)
6060      linebuffer *lbp;
6061      register FILE *stream;
6062 {
6063   char *buffer = lbp->buffer;
6064   register char *p = lbp->buffer;
6065   register char *pend;
6066   int chars_deleted;
6067
6068   pend = p + lbp->size;         /* Separate to avoid 386/IX compiler bug.  */
6069
6070   for (;;)
6071     {
6072       register int c = getc (stream);
6073       if (p == pend)
6074         {
6075           /* We're at the end of linebuffer: expand it. */
6076           lbp->size *= 2;
6077           xrnew (buffer, lbp->size, char);
6078           p += buffer - lbp->buffer;
6079           pend = buffer + lbp->size;
6080           lbp->buffer = buffer;
6081         }
6082       if (c == EOF)
6083         {
6084           *p = '\0';
6085           chars_deleted = 0;
6086           break;
6087         }
6088       if (c == '\n')
6089         {
6090           if (p > buffer && p[-1] == '\r')
6091             {
6092               p -= 1;
6093 #ifdef DOS_NT
6094              /* Assume CRLF->LF translation will be performed by Emacs
6095                 when loading this file, so CRs won't appear in the buffer.
6096                 It would be cleaner to compensate within Emacs;
6097                 however, Emacs does not know how many CRs were deleted
6098                 before any given point in the file.  */
6099               chars_deleted = 1;
6100 #else
6101               chars_deleted = 2;
6102 #endif
6103             }
6104           else
6105             {
6106               chars_deleted = 1;
6107             }
6108           *p = '\0';
6109           break;
6110         }
6111       *p++ = c;
6112     }
6113   lbp->len = p - buffer;
6114
6115   if (need_filebuf              /* we need filebuf for multi-line regexps */
6116       && chars_deleted > 0)     /* not at EOF */
6117     {
6118       while (filebuf.size <= filebuf.len + lbp->len + 1) /* +1 for \n */
6119         {
6120           /* Expand filebuf. */
6121           filebuf.size *= 2;
6122           xrnew (filebuf.buffer, filebuf.size, char);
6123         }
6124       strncpy (filebuf.buffer + filebuf.len, lbp->buffer, lbp->len);
6125       filebuf.len += lbp->len;
6126       filebuf.buffer[filebuf.len++] = '\n';
6127       filebuf.buffer[filebuf.len] = '\0';
6128     }
6129
6130   return lbp->len + chars_deleted;
6131 }
6132
6133 /*
6134  * Like readline_internal, above, but in addition try to match the
6135  * input line against relevant regular expressions and manage #line
6136  * directives.
6137  */
6138 static void
6139 readline (lbp, stream)
6140      linebuffer *lbp;
6141      FILE *stream;
6142 {
6143   long result;
6144
6145   linecharno = charno;          /* update global char number of line start */
6146   result = readline_internal (lbp, stream); /* read line */
6147   lineno += 1;                  /* increment global line number */
6148   charno += result;             /* increment global char number */
6149
6150   /* Honour #line directives. */
6151   if (!no_line_directive)
6152     {
6153       static bool discard_until_line_directive;
6154
6155       /* Check whether this is a #line directive. */
6156       if (result > 12 && strneq (lbp->buffer, "#line ", 6))
6157         {
6158           int start, lno;
6159
6160           if (DEBUG) start = 0; /* shut up the compiler */
6161           if (sscanf (lbp->buffer, "#line %d \"%n", &lno, &start) == 1)
6162             {
6163               char *endp = lbp->buffer + start;
6164
6165               assert (start > 0);
6166               while ((endp = etags_strchr (endp, '"')) != NULL
6167                      && endp[-1] == '\\')
6168                 endp++;
6169               if (endp != NULL)
6170                 /* Ok, this is a real #line directive.  Let's deal with it. */
6171                 {
6172                   char *taggedabsname;  /* absolute name of original file */
6173                   char *taggedfname;    /* name of original file as given */
6174                   char *name;           /* temp var */
6175
6176                   discard_until_line_directive = FALSE; /* found it */
6177                   name = lbp->buffer + start;
6178                   *endp = '\0';
6179                   canonicalize_filename (name); /* for DOS */
6180                   taggedabsname = absolute_filename (name, curfdp->infabsdir);
6181                   if (filename_is_absolute (name)
6182                       || filename_is_absolute (curfdp->infname))
6183                     taggedfname = savestr (taggedabsname);
6184                   else
6185                     taggedfname = relative_filename (taggedabsname,tagfiledir);
6186
6187                   if (streq (curfdp->taggedfname, taggedfname))
6188                     /* The #line directive is only a line number change.  We
6189                        deal with this afterwards. */
6190                     free (taggedfname);
6191                   else
6192                     /* The tags following this #line directive should be
6193                        attributed to taggedfname.  In order to do this, set
6194                        curfdp accordingly. */
6195                     {
6196                       fdesc *fdp; /* file description pointer */
6197
6198                       /* Go look for a file description already set up for the
6199                          file indicated in the #line directive.  If there is
6200                          one, use it from now until the next #line
6201                          directive. */
6202                       for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
6203                         if (streq (fdp->infname, curfdp->infname)
6204                             && streq (fdp->taggedfname, taggedfname))
6205                           /* If we remove the second test above (after the &&)
6206                              then all entries pertaining to the same file are
6207                              coalesced in the tags file.  If we use it, then
6208                              entries pertaining to the same file but generated
6209                              from different files (via #line directives) will
6210                              go into separate sections in the tags file.  These
6211                              alternatives look equivalent.  The first one
6212                              destroys some apparently useless information. */
6213                           {
6214                             curfdp = fdp;
6215                             free (taggedfname);
6216                             break;
6217                           }
6218                       /* Else, if we already tagged the real file, skip all
6219                          input lines until the next #line directive. */
6220                       if (fdp == NULL) /* not found */
6221                         for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
6222                           if (streq (fdp->infabsname, taggedabsname))
6223                             {
6224                               discard_until_line_directive = TRUE;
6225                               free (taggedfname);
6226                               break;
6227                             }
6228                       /* Else create a new file description and use that from
6229                          now on, until the next #line directive. */
6230                       if (fdp == NULL) /* not found */
6231                         {
6232                           fdp = fdhead;
6233                           fdhead = xnew (1, fdesc);
6234                           *fdhead = *curfdp; /* copy curr. file description */
6235                           fdhead->next = fdp;
6236                           fdhead->infname = savestr (curfdp->infname);
6237                           fdhead->infabsname = savestr (curfdp->infabsname);
6238                           fdhead->infabsdir = savestr (curfdp->infabsdir);
6239                           fdhead->taggedfname = taggedfname;
6240                           fdhead->usecharno = FALSE;
6241                           fdhead->prop = NULL;
6242                           fdhead->written = FALSE;
6243                           curfdp = fdhead;
6244                         }
6245                     }
6246                   free (taggedabsname);
6247                   lineno = lno - 1;
6248                   readline (lbp, stream);
6249                   return;
6250                 } /* if a real #line directive */
6251             } /* if #line is followed by a a number */
6252         } /* if line begins with "#line " */
6253
6254       /* If we are here, no #line directive was found. */
6255       if (discard_until_line_directive)
6256         {
6257           if (result > 0)
6258             {
6259               /* Do a tail recursion on ourselves, thus discarding the contents
6260                  of the line buffer. */
6261               readline (lbp, stream);
6262               return;
6263             }
6264           /* End of file. */
6265           discard_until_line_directive = FALSE;
6266           return;
6267         }
6268     } /* if #line directives should be considered */
6269
6270 #ifdef ETAGS_REGEXPS
6271   {
6272     int match;
6273     regexp *rp;
6274     char *name;
6275
6276     /* Match against relevant regexps. */
6277     if (lbp->len > 0)
6278       for (rp = p_head; rp != NULL; rp = rp->p_next)
6279         {
6280           /* Only use generic regexps or those for the current language.
6281              Also do not use multiline regexps, which is the job of
6282              regex_tag_multiline. */
6283           if ((rp->lang != NULL && rp->lang != fdhead->lang)
6284               || rp->multi_line)
6285             continue;
6286
6287           match = re_match (rp->pat, lbp->buffer, lbp->len, 0, &rp->regs);
6288           switch (match)
6289             {
6290             case -2:
6291               /* Some error. */
6292               if (!rp->error_signaled)
6293                 {
6294                   error ("regexp stack overflow while matching \"%s\"",
6295                          rp->pattern);
6296                   rp->error_signaled = TRUE;
6297                 }
6298               break;
6299             case -1:
6300               /* No match. */
6301               break;
6302             case 0:
6303               /* Empty string matched. */
6304               if (!rp->error_signaled)
6305                 {
6306                   error ("regexp matches the empty string: \"%s\"", rp->pattern);
6307                   rp->error_signaled = TRUE;
6308                 }
6309               break;
6310             default:
6311               /* Match occurred.  Construct a tag. */
6312               name = rp->name;
6313               if (name[0] == '\0')
6314                 name = NULL;
6315               else /* make a named tag */
6316                 name = substitute (lbp->buffer, rp->name, &rp->regs);
6317               if (rp->force_explicit_name)
6318                 /* Force explicit tag name, if a name is there. */
6319                 pfnote (name, TRUE, lbp->buffer, match, lineno, linecharno);
6320               else
6321                 make_tag (name, strlen (name), TRUE,
6322                           lbp->buffer, match, lineno, linecharno);
6323               break;
6324             }
6325         }
6326   }
6327 #endif /* ETAGS_REGEXPS */
6328 }
6329
6330 \f
6331 /*
6332  * Return a pointer to a space of size strlen(cp)+1 allocated
6333  * with xnew where the string CP has been copied.
6334  */
6335 static char *
6336 savestr (cp)
6337      char *cp;
6338 {
6339   return savenstr (cp, strlen (cp));
6340 }
6341
6342 /*
6343  * Return a pointer to a space of size LEN+1 allocated with xnew where
6344  * the string CP has been copied for at most the first LEN characters.
6345  */
6346 static char *
6347 savenstr (cp, len)
6348      char *cp;
6349      int len;
6350 {
6351   register char *dp;
6352
6353   dp = xnew (len + 1, char);
6354   strncpy (dp, cp, len);
6355   dp[len] = '\0';
6356   return dp;
6357 }
6358
6359 /*
6360  * Return the ptr in sp at which the character c last
6361  * appears; NULL if not found
6362  *
6363  * Identical to POSIX strrchr, included for portability.
6364  */
6365 static char *
6366 etags_strrchr (sp, c)
6367      register const char *sp;
6368      register int c;
6369 {
6370   register const char *r;
6371
6372   r = NULL;
6373   do
6374     {
6375       if (*sp == c)
6376         r = sp;
6377   } while (*sp++);
6378   return (char *)r;
6379 }
6380
6381 /*
6382  * Return the ptr in sp at which the character c first
6383  * appears; NULL if not found
6384  *
6385  * Identical to POSIX strchr, included for portability.
6386  */
6387 static char *
6388 etags_strchr (sp, c)
6389      register const char *sp;
6390      register int c;
6391 {
6392   do
6393     {
6394       if (*sp == c)
6395         return (char *)sp;
6396     } while (*sp++);
6397   return NULL;
6398 }
6399
6400 /*
6401  * Compare two strings, ignoring case for alphabetic characters.
6402  *
6403  * Same as BSD's strcasecmp, included for portability.
6404  */
6405 static int
6406 etags_strcasecmp (s1, s2)
6407      register const char *s1;
6408      register const char *s2;
6409 {
6410   while (*s1 != '\0'
6411          && (ISALPHA (*s1) && ISALPHA (*s2)
6412              ? lowcase (*s1) == lowcase (*s2)
6413              : *s1 == *s2))
6414     s1++, s2++;
6415
6416   return (ISALPHA (*s1) && ISALPHA (*s2)
6417           ? lowcase (*s1) - lowcase (*s2)
6418           : *s1 - *s2);
6419 }
6420
6421 /*
6422  * Compare two strings, ignoring case for alphabetic characters.
6423  * Stop after a given number of characters
6424  *
6425  * Same as BSD's strncasecmp, included for portability.
6426  */
6427 static int
6428 etags_strncasecmp (s1, s2, n)
6429      register const char *s1;
6430      register const char *s2;
6431      register int n;
6432 {
6433   while (*s1 != '\0' && n-- > 0
6434          && (ISALPHA (*s1) && ISALPHA (*s2)
6435              ? lowcase (*s1) == lowcase (*s2)
6436              : *s1 == *s2))
6437     s1++, s2++;
6438
6439   if (n < 0)
6440     return 0;
6441   else
6442     return (ISALPHA (*s1) && ISALPHA (*s2)
6443             ? lowcase (*s1) - lowcase (*s2)
6444             : *s1 - *s2);
6445 }
6446
6447 /* Skip spaces, return new pointer. */
6448 static char *
6449 skip_spaces (cp)
6450      char *cp;
6451 {
6452   while (iswhite (*cp))
6453     cp++;
6454   return cp;
6455 }
6456
6457 /* Skip non spaces, return new pointer. */
6458 static char *
6459 skip_non_spaces (cp)
6460      char *cp;
6461 {
6462   while (*cp != '\0' && !iswhite (*cp))
6463     cp++;
6464   return cp;
6465 }
6466
6467 /* Print error message and exit.  */
6468 void
6469 fatal (s1, s2)
6470      char *s1, *s2;
6471 {
6472   error (s1, s2);
6473   exit (BAD);
6474 }
6475
6476 static void
6477 pfatal (s1)
6478      char *s1;
6479 {
6480   perror (s1);
6481   exit (BAD);
6482 }
6483
6484 static void
6485 suggest_asking_for_help ()
6486 {
6487   fprintf (stderr, "\tTry `%s %s' for a complete list of options.\n",
6488            progname,
6489 #ifdef LONG_OPTIONS
6490            "--help"
6491 #else
6492            "-h"
6493 #endif
6494            );
6495   exit (BAD);
6496 }
6497
6498 /* Print error message.  `s1' is printf control string, `s2' is arg for it. */
6499 static void
6500 error (s1, s2)
6501      const char *s1, *s2;
6502 {
6503   fprintf (stderr, "%s: ", progname);
6504   fprintf (stderr, s1, s2);
6505   fprintf (stderr, "\n");
6506 }
6507
6508 /* Return a newly-allocated string whose contents
6509    concatenate those of s1, s2, s3.  */
6510 static char *
6511 concat (s1, s2, s3)
6512      char *s1, *s2, *s3;
6513 {
6514   int len1 = strlen (s1), len2 = strlen (s2), len3 = strlen (s3);
6515   char *result = xnew (len1 + len2 + len3 + 1, char);
6516
6517   strcpy (result, s1);
6518   strcpy (result + len1, s2);
6519   strcpy (result + len1 + len2, s3);
6520   result[len1 + len2 + len3] = '\0';
6521
6522   return result;
6523 }
6524
6525 \f
6526 /* Does the same work as the system V getcwd, but does not need to
6527    guess the buffer size in advance. */
6528 static char *
6529 etags_getcwd ()
6530 {
6531 #ifdef HAVE_GETCWD
6532   int bufsize = 200;
6533   char *path = xnew (bufsize, char);
6534
6535   while (getcwd (path, bufsize) == NULL)
6536     {
6537       if (errno != ERANGE)
6538         pfatal ("getcwd");
6539       bufsize *= 2;
6540       free (path);
6541       path = xnew (bufsize, char);
6542     }
6543
6544   canonicalize_filename (path);
6545   return path;
6546
6547 #else /* not HAVE_GETCWD */
6548 #if MSDOS
6549
6550   char *p, path[MAXPATHLEN + 1]; /* Fixed size is safe on MSDOS.  */
6551
6552   getwd (path);
6553
6554   for (p = path; *p != '\0'; p++)
6555     if (*p == '\\')
6556       *p = '/';
6557     else
6558       *p = lowcase (*p);
6559
6560   return strdup (path);
6561 #else /* not MSDOS */
6562   linebuffer path;
6563   FILE *pipe;
6564
6565   linebuffer_init (&path);
6566   pipe = (FILE *) popen ("pwd 2>/dev/null", "r");
6567   if (pipe == NULL || readline_internal (&path, pipe) == 0)
6568     pfatal ("pwd");
6569   pclose (pipe);
6570
6571   return path.buffer;
6572 #endif /* not MSDOS */
6573 #endif /* not HAVE_GETCWD */
6574 }
6575
6576 /* Return a newly allocated string containing the file name of FILE
6577    relative to the absolute directory DIR (which should end with a slash). */
6578 static char *
6579 relative_filename (file, dir)
6580      char *file, *dir;
6581 {
6582   char *fp, *dp, *afn, *res;
6583   int i;
6584
6585   /* Find the common root of file and dir (with a trailing slash). */
6586   afn = absolute_filename (file, cwd);
6587   fp = afn;
6588   dp = dir;
6589   while (*fp++ == *dp++)
6590     continue;
6591   fp--, dp--;                   /* back to the first differing char */
6592 #ifdef DOS_NT
6593   if (fp == afn && afn[0] != '/') /* cannot build a relative name */
6594     return afn;
6595 #endif
6596   do                            /* look at the equal chars until '/' */
6597     fp--, dp--;
6598   while (*fp != '/');
6599
6600   /* Build a sequence of "../" strings for the resulting relative file name. */
6601   i = 0;
6602   while ((dp = etags_strchr (dp + 1, '/')) != NULL)
6603     i += 1;
6604   res = xnew (3*i + strlen (fp + 1) + 1, char);
6605   res[0] = '\0';
6606   while (i-- > 0)
6607     strcat (res, "../");
6608
6609   /* Add the file name relative to the common root of file and dir. */
6610   strcat (res, fp + 1);
6611   free (afn);
6612
6613   return res;
6614 }
6615
6616 /* Return a newly allocated string containing the absolute file name
6617    of FILE given DIR (which should end with a slash). */
6618 static char *
6619 absolute_filename (file, dir)
6620      char *file, *dir;
6621 {
6622   char *slashp, *cp, *res;
6623
6624   if (filename_is_absolute (file))
6625     res = savestr (file);
6626 #ifdef DOS_NT
6627   /* We don't support non-absolute file names with a drive
6628      letter, like `d:NAME' (it's too much hassle).  */
6629   else if (file[1] == ':')
6630     fatal ("%s: relative file names with drive letters not supported", file);
6631 #endif
6632   else
6633     res = concat (dir, file, "");
6634
6635   /* Delete the "/dirname/.." and "/." substrings. */
6636   slashp = etags_strchr (res, '/');
6637   while (slashp != NULL && slashp[0] != '\0')
6638     {
6639       if (slashp[1] == '.')
6640         {
6641           if (slashp[2] == '.'
6642               && (slashp[3] == '/' || slashp[3] == '\0'))
6643             {
6644               cp = slashp;
6645               do
6646                 cp--;
6647               while (cp >= res && !filename_is_absolute (cp));
6648               if (cp < res)
6649                 cp = slashp;    /* the absolute name begins with "/.." */
6650 #ifdef DOS_NT
6651               /* Under MSDOS and NT we get `d:/NAME' as absolute
6652                  file name, so the luser could say `d:/../NAME'.
6653                  We silently treat this as `d:/NAME'.  */
6654               else if (cp[0] != '/')
6655                 cp = slashp;
6656 #endif
6657               strcpy (cp, slashp + 3);
6658               slashp = cp;
6659               continue;
6660             }
6661           else if (slashp[2] == '/' || slashp[2] == '\0')
6662             {
6663               strcpy (slashp, slashp + 2);
6664               continue;
6665             }
6666         }
6667
6668       slashp = etags_strchr (slashp + 1, '/');
6669     }
6670
6671   if (res[0] == '\0')
6672     return savestr ("/");
6673   else
6674     return res;
6675 }
6676
6677 /* Return a newly allocated string containing the absolute
6678    file name of dir where FILE resides given DIR (which should
6679    end with a slash). */
6680 static char *
6681 absolute_dirname (file, dir)
6682      char *file, *dir;
6683 {
6684   char *slashp, *res;
6685   char save;
6686
6687   canonicalize_filename (file);
6688   slashp = etags_strrchr (file, '/');
6689   if (slashp == NULL)
6690     return savestr (dir);
6691   save = slashp[1];
6692   slashp[1] = '\0';
6693   res = absolute_filename (file, dir);
6694   slashp[1] = save;
6695
6696   return res;
6697 }
6698
6699 /* Whether the argument string is an absolute file name.  The argument
6700    string must have been canonicalized with canonicalize_filename. */
6701 static bool
6702 filename_is_absolute (fn)
6703      char *fn;
6704 {
6705   return (fn[0] == '/'
6706 #ifdef DOS_NT
6707           || (ISALPHA(fn[0]) && fn[1] == ':' && fn[2] == '/')
6708 #endif
6709           );
6710 }
6711
6712 /* Translate backslashes into slashes.  Works in place. */
6713 static void
6714 canonicalize_filename (fn)
6715      register char *fn;
6716 {
6717 #ifdef DOS_NT
6718   /* Canonicalize drive letter case.  */
6719   if (fn[0] != '\0' && fn[1] == ':' && ISLOWER (fn[0]))
6720     fn[0] = upcase (fn[0]);
6721   /* Convert backslashes to slashes.  */
6722   for (; *fn != '\0'; fn++)
6723     if (*fn == '\\')
6724       *fn = '/';
6725 #else
6726   /* No action. */
6727   fn = NULL;                    /* shut up the compiler */
6728 #endif
6729 }
6730
6731 \f
6732 /* Initialize a linebuffer for use */
6733 static void
6734 linebuffer_init (lbp)
6735      linebuffer *lbp;
6736 {
6737   lbp->size = (DEBUG) ? 3 : 200;
6738   lbp->buffer = xnew (lbp->size, char);
6739   lbp->buffer[0] = '\0';
6740   lbp->len = 0;
6741 }
6742
6743 /* Set the minimum size of a string contained in a linebuffer. */
6744 static void
6745 linebuffer_setlen (lbp, toksize)
6746      linebuffer *lbp;
6747      int toksize;
6748 {
6749   while (lbp->size <= toksize)
6750     {
6751       lbp->size *= 2;
6752       xrnew (lbp->buffer, lbp->size, char);
6753     }
6754   lbp->len = toksize;
6755 }
6756
6757 /* Like malloc but get fatal error if memory is exhausted. */
6758 static PTR
6759 xmalloc (size)
6760      unsigned int size;
6761 {
6762   PTR result = (PTR) malloc (size);
6763   if (result == NULL)
6764     fatal ("virtual memory exhausted", (char *)NULL);
6765   return result;
6766 }
6767
6768 static PTR
6769 xrealloc (ptr, size)
6770      char *ptr;
6771      unsigned int size;
6772 {
6773   PTR result = (PTR) realloc (ptr, size);
6774   if (result == NULL)
6775     fatal ("virtual memory exhausted", (char *)NULL);
6776   return result;
6777 }
6778
6779 /*
6780  * Local Variables:
6781  * c-indentation-style: gnu
6782  * indent-tabs-mode: t
6783  * tab-width: 8
6784  * fill-column: 79
6785  * c-font-lock-extra-types: ("FILE" "bool" "language" "linebuffer" "fdesc" "node" "regexp")
6786  * End:
6787  */