(make-autoload): Add define-global-minor-mode.
[emacs.git] / lib-src / etags.c
blobfecf79041176261fa9bc9c1e5f7f12f784eee431
1 /* Tags file maker to go with GNU Emacs -*- coding: latin-1 -*-
2 Copyright (C) 1984, 1987, 1988, 1989, 1993, 1994, 1995,
3 1998, 1999, 2000, 2001, 2002, 2003, 2004,
4 2005 Free Software Foundation, Inc. and Ken Arnold
6 This file is not considered part of GNU Emacs.
8 This program is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 2 of the License, or
11 (at your option) any later version.
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software Foundation,
20 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
23 * Authors:
24 * Ctags originally by Ken Arnold.
25 * Fortran added by Jim Kleckner.
26 * Ed Pelegri-Llopart added C typedefs.
27 * Gnu Emacs TAGS format and modifications by RMS?
28 * 1989 Sam Kendall added C++.
29 * 1992 Joseph B. Wells improved C and C++ parsing.
30 * 1993 Francesco Potortì reorganised C and C++.
31 * 1994 Line-by-line regexp tags by Tom Tromey.
32 * 2001 Nested classes by Francesco Potortì (concept by Mykola Dzyuba).
33 * 2002 #line directives by Francesco Potortì.
35 * Francesco Potortì <pot@gnu.org> has maintained and improved it since 1993.
39 * If you want to add support for a new language, start by looking at the LUA
40 * language, which is the simplest. Alternatively, consider shipping a
41 * configuration file containing regexp definitions for etags.
44 char pot_etags_version[] = "@(#) pot revision number is 17.15";
46 #define TRUE 1
47 #define FALSE 0
49 #ifdef DEBUG
50 # undef DEBUG
51 # define DEBUG TRUE
52 #else
53 # define DEBUG FALSE
54 # define NDEBUG /* disable assert */
55 #endif
57 #ifdef HAVE_CONFIG_H
58 # include <config.h>
59 /* On some systems, Emacs defines static as nothing for the sake
60 of unexec. We don't want that here since we don't use unexec. */
61 # undef static
62 # define ETAGS_REGEXPS /* use the regexp features */
63 # define LONG_OPTIONS /* accept long options */
64 # ifndef PTR /* for Xemacs */
65 # define PTR void *
66 # endif
67 # ifndef __P /* for Xemacs */
68 # define __P(args) args
69 # endif
70 #else /* no config.h */
71 # if defined(__STDC__) && (__STDC__ || defined(__SUNPRO_C))
72 # define __P(args) args /* use prototypes */
73 # define PTR void * /* for generic pointers */
74 # else /* not standard C */
75 # define __P(args) () /* no prototypes */
76 # define const /* remove const for old compilers' sake */
77 # define PTR long * /* don't use void* */
78 # endif
79 #endif /* !HAVE_CONFIG_H */
81 #ifndef _GNU_SOURCE
82 # define _GNU_SOURCE 1 /* enables some compiler checks on GNU */
83 #endif
85 #ifdef LONG_OPTIONS
86 # undef LONG_OPTIONS
87 # define LONG_OPTIONS TRUE
88 #else
89 # define LONG_OPTIONS FALSE
90 #endif
92 /* WIN32_NATIVE is for Xemacs.
93 MSDOS, WINDOWSNT, DOS_NT are for Emacs. */
94 #ifdef WIN32_NATIVE
95 # undef MSDOS
96 # undef WINDOWSNT
97 # define WINDOWSNT
98 #endif /* WIN32_NATIVE */
100 #ifdef MSDOS
101 # undef MSDOS
102 # define MSDOS TRUE
103 # include <fcntl.h>
104 # include <sys/param.h>
105 # include <io.h>
106 # ifndef HAVE_CONFIG_H
107 # define DOS_NT
108 # include <sys/config.h>
109 # endif
110 #else
111 # define MSDOS FALSE
112 #endif /* MSDOS */
114 #ifdef WINDOWSNT
115 # include <stdlib.h>
116 # include <fcntl.h>
117 # include <string.h>
118 # include <direct.h>
119 # include <io.h>
120 # define MAXPATHLEN _MAX_PATH
121 # undef HAVE_NTGUI
122 # undef DOS_NT
123 # define DOS_NT
124 # ifndef HAVE_GETCWD
125 # define HAVE_GETCWD
126 # endif /* undef HAVE_GETCWD */
127 #else /* not WINDOWSNT */
128 # ifdef STDC_HEADERS
129 # include <stdlib.h>
130 # include <string.h>
131 # else /* no standard C headers */
132 extern char *getenv ();
133 # ifdef VMS
134 # define EXIT_SUCCESS 1
135 # define EXIT_FAILURE 0
136 # else /* no VMS */
137 # define EXIT_SUCCESS 0
138 # define EXIT_FAILURE 1
139 # endif
140 # endif
141 #endif /* !WINDOWSNT */
143 #ifdef HAVE_UNISTD_H
144 # include <unistd.h>
145 #else
146 # if defined (HAVE_GETCWD) && !defined (WINDOWSNT)
147 extern char *getcwd (char *buf, size_t size);
148 # endif
149 #endif /* HAVE_UNISTD_H */
151 #include <stdio.h>
152 #include <ctype.h>
153 #include <errno.h>
154 #ifndef errno
155 extern int errno;
156 #endif
157 #include <sys/types.h>
158 #include <sys/stat.h>
160 #include <assert.h>
161 #ifdef NDEBUG
162 # undef assert /* some systems have a buggy assert.h */
163 # define assert(x) ((void) 0)
164 #endif
166 #if !defined (S_ISREG) && defined (S_IFREG)
167 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
168 #endif
170 #if LONG_OPTIONS
171 # include <getopt.h>
172 #else
173 # define getopt_long(argc,argv,optstr,lopts,lind) getopt (argc, argv, optstr)
174 extern char *optarg;
175 extern int optind, opterr;
176 #endif /* LONG_OPTIONS */
178 #ifdef ETAGS_REGEXPS
179 # ifndef HAVE_CONFIG_H /* this is a standalone compilation */
180 # ifdef __CYGWIN__ /* compiling on Cygwin */
181 !!! NOTICE !!!
182 the regex.h distributed with Cygwin is not compatible with etags, alas!
183 If you want regular expression support, you should delete this notice and
184 arrange to use the GNU regex.h and regex.c.
185 # endif
186 # endif
187 # include <regex.h>
188 #endif /* ETAGS_REGEXPS */
190 /* Define CTAGS to make the program "ctags" compatible with the usual one.
191 Leave it undefined to make the program "etags", which makes emacs-style
192 tag tables and tags typedefs, #defines and struct/union/enum by default. */
193 #ifdef CTAGS
194 # undef CTAGS
195 # define CTAGS TRUE
196 #else
197 # define CTAGS FALSE
198 #endif
200 #define streq(s,t) (assert((s)!=NULL || (t)!=NULL), !strcmp (s, t))
201 #define strcaseeq(s,t) (assert((s)!=NULL && (t)!=NULL), !etags_strcasecmp (s, t))
202 #define strneq(s,t,n) (assert((s)!=NULL || (t)!=NULL), !strncmp (s, t, n))
203 #define strncaseeq(s,t,n) (assert((s)!=NULL && (t)!=NULL), !etags_strncasecmp (s, t, n))
205 #define CHARS 256 /* 2^sizeof(char) */
206 #define CHAR(x) ((unsigned int)(x) & (CHARS - 1))
207 #define iswhite(c) (_wht[CHAR(c)]) /* c is white (see white) */
208 #define notinname(c) (_nin[CHAR(c)]) /* c is not in a name (see nonam) */
209 #define begtoken(c) (_btk[CHAR(c)]) /* c can start token (see begtk) */
210 #define intoken(c) (_itk[CHAR(c)]) /* c can be in token (see midtk) */
211 #define endtoken(c) (_etk[CHAR(c)]) /* c ends tokens (see endtk) */
213 #define ISALNUM(c) isalnum (CHAR(c))
214 #define ISALPHA(c) isalpha (CHAR(c))
215 #define ISDIGIT(c) isdigit (CHAR(c))
216 #define ISLOWER(c) islower (CHAR(c))
218 #define lowcase(c) tolower (CHAR(c))
219 #define upcase(c) toupper (CHAR(c))
223 * xnew, xrnew -- allocate, reallocate storage
225 * SYNOPSIS: Type *xnew (int n, Type);
226 * void xrnew (OldPointer, int n, Type);
228 #if DEBUG
229 # include "chkmalloc.h"
230 # define xnew(n,Type) ((Type *) trace_malloc (__FILE__, __LINE__, \
231 (n) * sizeof (Type)))
232 # define xrnew(op,n,Type) ((op) = (Type *) trace_realloc (__FILE__, __LINE__, \
233 (char *) (op), (n) * sizeof (Type)))
234 #else
235 # define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type)))
236 # define xrnew(op,n,Type) ((op) = (Type *) xrealloc ( \
237 (char *) (op), (n) * sizeof (Type)))
238 #endif
240 #define bool int
242 typedef void Lang_function __P((FILE *));
244 typedef struct
246 char *suffix; /* file name suffix for this compressor */
247 char *command; /* takes one arg and decompresses to stdout */
248 } compressor;
250 typedef struct
252 char *name; /* language name */
253 char *help; /* detailed help for the language */
254 Lang_function *function; /* parse function */
255 char **suffixes; /* name suffixes of this language's files */
256 char **filenames; /* names of this language's files */
257 char **interpreters; /* interpreters for this language */
258 bool metasource; /* source used to generate other sources */
259 } language;
261 typedef struct fdesc
263 struct fdesc *next; /* for the linked list */
264 char *infname; /* uncompressed input file name */
265 char *infabsname; /* absolute uncompressed input file name */
266 char *infabsdir; /* absolute dir of input file */
267 char *taggedfname; /* file name to write in tagfile */
268 language *lang; /* language of file */
269 char *prop; /* file properties to write in tagfile */
270 bool usecharno; /* etags tags shall contain char number */
271 bool written; /* entry written in the tags file */
272 } fdesc;
274 typedef struct node_st
275 { /* sorting structure */
276 struct node_st *left, *right; /* left and right sons */
277 fdesc *fdp; /* description of file to whom tag belongs */
278 char *name; /* tag name */
279 char *regex; /* search regexp */
280 bool valid; /* write this tag on the tag file */
281 bool is_func; /* function tag: use regexp in CTAGS mode */
282 bool been_warned; /* warning already given for duplicated tag */
283 int lno; /* line number tag is on */
284 long cno; /* character number line starts on */
285 } node;
288 * A `linebuffer' is a structure which holds a line of text.
289 * `readline_internal' reads a line from a stream into a linebuffer
290 * and works regardless of the length of the line.
291 * SIZE is the size of BUFFER, LEN is the length of the string in
292 * BUFFER after readline reads it.
294 typedef struct
296 long size;
297 int len;
298 char *buffer;
299 } linebuffer;
301 /* Used to support mixing of --lang and file names. */
302 typedef struct
304 enum {
305 at_language, /* a language specification */
306 at_regexp, /* a regular expression */
307 at_filename, /* a file name */
308 at_stdin, /* read from stdin here */
309 at_end /* stop parsing the list */
310 } arg_type; /* argument type */
311 language *lang; /* language associated with the argument */
312 char *what; /* the argument itself */
313 } argument;
315 #ifdef ETAGS_REGEXPS
316 /* Structure defining a regular expression. */
317 typedef struct regexp
319 struct regexp *p_next; /* pointer to next in list */
320 language *lang; /* if set, use only for this language */
321 char *pattern; /* the regexp pattern */
322 char *name; /* tag name */
323 struct re_pattern_buffer *pat; /* the compiled pattern */
324 struct re_registers regs; /* re registers */
325 bool error_signaled; /* already signaled for this regexp */
326 bool force_explicit_name; /* do not allow implict tag name */
327 bool ignore_case; /* ignore case when matching */
328 bool multi_line; /* do a multi-line match on the whole file */
329 } regexp;
330 #endif /* ETAGS_REGEXPS */
333 /* Many compilers barf on this:
334 Lang_function Ada_funcs;
335 so let's write it this way */
336 static void Ada_funcs __P((FILE *));
337 static void Asm_labels __P((FILE *));
338 static void C_entries __P((int c_ext, FILE *));
339 static void default_C_entries __P((FILE *));
340 static void plain_C_entries __P((FILE *));
341 static void Cjava_entries __P((FILE *));
342 static void Cobol_paragraphs __P((FILE *));
343 static void Cplusplus_entries __P((FILE *));
344 static void Cstar_entries __P((FILE *));
345 static void Erlang_functions __P((FILE *));
346 static void Forth_words __P((FILE *));
347 static void Fortran_functions __P((FILE *));
348 static void HTML_labels __P((FILE *));
349 static void Lisp_functions __P((FILE *));
350 static void Lua_functions __P((FILE *));
351 static void Makefile_targets __P((FILE *));
352 static void Pascal_functions __P((FILE *));
353 static void Perl_functions __P((FILE *));
354 static void PHP_functions __P((FILE *));
355 static void PS_functions __P((FILE *));
356 static void Prolog_functions __P((FILE *));
357 static void Python_functions __P((FILE *));
358 static void Scheme_functions __P((FILE *));
359 static void TeX_commands __P((FILE *));
360 static void Texinfo_nodes __P((FILE *));
361 static void Yacc_entries __P((FILE *));
362 static void just_read_file __P((FILE *));
364 static void print_language_names __P((void));
365 static void print_version __P((void));
366 static void print_help __P((argument *));
367 int main __P((int, char **));
369 static compressor *get_compressor_from_suffix __P((char *, char **));
370 static language *get_language_from_langname __P((const char *));
371 static language *get_language_from_interpreter __P((char *));
372 static language *get_language_from_filename __P((char *, bool));
373 static void readline __P((linebuffer *, FILE *));
374 static long readline_internal __P((linebuffer *, FILE *));
375 static bool nocase_tail __P((char *));
376 static void get_tag __P((char *, char **));
378 #ifdef ETAGS_REGEXPS
379 static void analyse_regex __P((char *));
380 static void free_regexps __P((void));
381 static void regex_tag_multiline __P((void));
382 #endif /* ETAGS_REGEXPS */
383 static void error __P((const char *, const char *));
384 static void suggest_asking_for_help __P((void));
385 void fatal __P((char *, char *));
386 static void pfatal __P((char *));
387 static void add_node __P((node *, node **));
389 static void init __P((void));
390 static void process_file_name __P((char *, language *));
391 static void process_file __P((FILE *, char *, language *));
392 static void find_entries __P((FILE *));
393 static void free_tree __P((node *));
394 static void free_fdesc __P((fdesc *));
395 static void pfnote __P((char *, bool, char *, int, int, long));
396 static void make_tag __P((char *, int, bool, char *, int, int, long));
397 static void invalidate_nodes __P((fdesc *, node **));
398 static void put_entries __P((node *));
400 static char *concat __P((char *, char *, char *));
401 static char *skip_spaces __P((char *));
402 static char *skip_non_spaces __P((char *));
403 static char *savenstr __P((char *, int));
404 static char *savestr __P((char *));
405 static char *etags_strchr __P((const char *, int));
406 static char *etags_strrchr __P((const char *, int));
407 static int etags_strcasecmp __P((const char *, const char *));
408 static int etags_strncasecmp __P((const char *, const char *, int));
409 static char *etags_getcwd __P((void));
410 static char *relative_filename __P((char *, char *));
411 static char *absolute_filename __P((char *, char *));
412 static char *absolute_dirname __P((char *, char *));
413 static bool filename_is_absolute __P((char *f));
414 static void canonicalize_filename __P((char *));
415 static void linebuffer_init __P((linebuffer *));
416 static void linebuffer_setlen __P((linebuffer *, int));
417 static PTR xmalloc __P((unsigned int));
418 static PTR xrealloc __P((char *, unsigned int));
421 static char searchar = '/'; /* use /.../ searches */
423 static char *tagfile; /* output file */
424 static char *progname; /* name this program was invoked with */
425 static char *cwd; /* current working directory */
426 static char *tagfiledir; /* directory of tagfile */
427 static FILE *tagf; /* ioptr for tags file */
429 static fdesc *fdhead; /* head of file description list */
430 static fdesc *curfdp; /* current file description */
431 static int lineno; /* line number of current line */
432 static long charno; /* current character number */
433 static long linecharno; /* charno of start of current line */
434 static char *dbp; /* pointer to start of current tag */
436 static const int invalidcharno = -1;
438 static node *nodehead; /* the head of the binary tree of tags */
439 static node *last_node; /* the last node created */
441 static linebuffer lb; /* the current line */
442 static linebuffer filebuf; /* a buffer containing the whole file */
443 static linebuffer token_name; /* a buffer containing a tag name */
445 /* boolean "functions" (see init) */
446 static bool _wht[CHARS], _nin[CHARS], _itk[CHARS], _btk[CHARS], _etk[CHARS];
447 static char
448 /* white chars */
449 *white = " \f\t\n\r\v",
450 /* not in a name */
451 *nonam = " \f\t\n\r()=,;", /* look at make_tag before modifying! */
452 /* token ending chars */
453 *endtk = " \t\n\r\"'#()[]{}=-+%*/&|^~!<>;,.:?",
454 /* token starting chars */
455 *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~@",
456 /* valid in-token chars */
457 *midtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789";
459 static bool append_to_tagfile; /* -a: append to tags */
460 /* The next four default to TRUE for etags, but to FALSE for ctags. */
461 static bool typedefs; /* -t: create tags for C and Ada typedefs */
462 static bool typedefs_or_cplusplus; /* -T: create tags for C typedefs, level */
463 /* 0 struct/enum/union decls, and C++ */
464 /* member functions. */
465 static bool constantypedefs; /* -d: create tags for C #define, enum */
466 /* constants and variables. */
467 /* -D: opposite of -d. Default under ctags. */
468 static bool globals; /* create tags for global variables */
469 static bool declarations; /* --declarations: tag them and extern in C&Co*/
470 static bool members; /* create tags for C member variables */
471 static bool no_line_directive; /* ignore #line directives (undocumented) */
472 static bool update; /* -u: update tags */
473 static bool vgrind_style; /* -v: create vgrind style index output */
474 static bool no_warnings; /* -w: suppress warnings */
475 static bool cxref_style; /* -x: create cxref style output */
476 static bool cplusplus; /* .[hc] means C++, not C */
477 static bool ignoreindent; /* -I: ignore indentation in C */
478 static bool packages_only; /* --packages-only: in Ada, only tag packages*/
480 /* STDIN is defined in LynxOS system headers */
481 #ifdef STDIN
482 # undef STDIN
483 #endif
485 #define STDIN 0x1001 /* returned by getopt_long on --parse-stdin */
486 static bool parsing_stdin; /* --parse-stdin used */
488 #ifdef ETAGS_REGEXPS
489 static regexp *p_head; /* list of all regexps */
490 static bool need_filebuf; /* some regexes are multi-line */
491 #else
492 # define need_filebuf FALSE
493 #endif /* ETAGS_REGEXPS */
495 #if LONG_OPTIONS
496 static struct option longopts[] =
498 { "append", no_argument, NULL, 'a' },
499 { "packages-only", no_argument, &packages_only, TRUE },
500 { "c++", no_argument, NULL, 'C' },
501 { "declarations", no_argument, &declarations, TRUE },
502 { "no-line-directive", no_argument, &no_line_directive, TRUE },
503 { "help", no_argument, NULL, 'h' },
504 { "help", no_argument, NULL, 'H' },
505 { "ignore-indentation", no_argument, NULL, 'I' },
506 { "language", required_argument, NULL, 'l' },
507 { "members", no_argument, &members, TRUE },
508 { "no-members", no_argument, &members, FALSE },
509 { "output", required_argument, NULL, 'o' },
510 #ifdef ETAGS_REGEXPS
511 { "regex", required_argument, NULL, 'r' },
512 { "no-regex", no_argument, NULL, 'R' },
513 { "ignore-case-regex", required_argument, NULL, 'c' },
514 #endif /* ETAGS_REGEXPS */
515 { "parse-stdin", required_argument, NULL, STDIN },
516 { "version", no_argument, NULL, 'V' },
518 #if CTAGS /* Ctags options */
519 { "backward-search", no_argument, NULL, 'B' },
520 { "cxref", no_argument, NULL, 'x' },
521 { "defines", no_argument, NULL, 'd' },
522 { "globals", no_argument, &globals, TRUE },
523 { "typedefs", no_argument, NULL, 't' },
524 { "typedefs-and-c++", no_argument, NULL, 'T' },
525 { "update", no_argument, NULL, 'u' },
526 { "vgrind", no_argument, NULL, 'v' },
527 { "no-warn", no_argument, NULL, 'w' },
529 #else /* Etags options */
530 { "no-defines", no_argument, NULL, 'D' },
531 { "no-globals", no_argument, &globals, FALSE },
532 { "include", required_argument, NULL, 'i' },
533 #endif
534 { NULL }
536 #endif /* LONG_OPTIONS */
538 static compressor compressors[] =
540 { "z", "gzip -d -c"},
541 { "Z", "gzip -d -c"},
542 { "gz", "gzip -d -c"},
543 { "GZ", "gzip -d -c"},
544 { "bz2", "bzip2 -d -c" },
545 { NULL }
549 * Language stuff.
552 /* Ada code */
553 static char *Ada_suffixes [] =
554 { "ads", "adb", "ada", NULL };
555 static char Ada_help [] =
556 "In Ada code, functions, procedures, packages, tasks and types are\n\
557 tags. Use the `--packages-only' option to create tags for\n\
558 packages only.\n\
559 Ada tag names have suffixes indicating the type of entity:\n\
560 Entity type: Qualifier:\n\
561 ------------ ----------\n\
562 function /f\n\
563 procedure /p\n\
564 package spec /s\n\
565 package body /b\n\
566 type /t\n\
567 task /k\n\
568 Thus, `M-x find-tag <RET> bidule/b <RET>' will go directly to the\n\
569 body of the package `bidule', while `M-x find-tag <RET> bidule <RET>'\n\
570 will just search for any tag `bidule'.";
572 /* Assembly code */
573 static char *Asm_suffixes [] =
574 { "a", /* Unix assembler */
575 "asm", /* Microcontroller assembly */
576 "def", /* BSO/Tasking definition includes */
577 "inc", /* Microcontroller include files */
578 "ins", /* Microcontroller include files */
579 "s", "sa", /* Unix assembler */
580 "S", /* cpp-processed Unix assembler */
581 "src", /* BSO/Tasking C compiler output */
582 NULL
584 static char Asm_help [] =
585 "In assembler code, labels appearing at the beginning of a line,\n\
586 followed by a colon, are tags.";
589 /* Note that .c and .h can be considered C++, if the --c++ flag was
590 given, or if the `class' or `template' keyowrds are met inside the file.
591 That is why default_C_entries is called for these. */
592 static char *default_C_suffixes [] =
593 { "c", "h", NULL };
594 static char default_C_help [] =
595 "In C code, any C function or typedef is a tag, and so are\n\
596 definitions of `struct', `union' and `enum'. `#define' macro\n\
597 definitions and `enum' constants are tags unless you specify\n\
598 `--no-defines'. Global variables are tags unless you specify\n\
599 `--no-globals'. Use of `--no-globals' and `--no-defines'\n\
600 can make the tags table file much smaller.\n\
601 You can tag function declarations and external variables by\n\
602 using `--declarations', and struct members by using `--members'.";
604 static char *Cplusplus_suffixes [] =
605 { "C", "c++", "cc", "cpp", "cxx", "H", "h++", "hh", "hpp", "hxx",
606 "M", /* Objective C++ */
607 "pdb", /* Postscript with C syntax */
608 NULL };
609 static char Cplusplus_help [] =
610 "In C++ code, all the tag constructs of C code are tagged. (Use\n\
611 --help --lang=c --lang=c++ for full help.)\n\
612 In addition to C tags, member functions are also recognized, and\n\
613 optionally member variables if you use the `--members' option.\n\
614 Tags for variables and functions in classes are named `CLASS::VARIABLE'\n\
615 and `CLASS::FUNCTION'. `operator' definitions have tag names like\n\
616 `operator+'.";
618 static char *Cjava_suffixes [] =
619 { "java", NULL };
620 static char Cjava_help [] =
621 "In Java code, all the tags constructs of C and C++ code are\n\
622 tagged. (Use --help --lang=c --lang=c++ --lang=java for full help.)";
625 static char *Cobol_suffixes [] =
626 { "COB", "cob", NULL };
627 static char Cobol_help [] =
628 "In Cobol code, tags are paragraph names; that is, any word\n\
629 starting in column 8 and followed by a period.";
631 static char *Cstar_suffixes [] =
632 { "cs", "hs", NULL };
634 static char *Erlang_suffixes [] =
635 { "erl", "hrl", NULL };
636 static char Erlang_help [] =
637 "In Erlang code, the tags are the functions, records and macros\n\
638 defined in the file.";
640 char *Forth_suffixes [] =
641 { "fth", "tok", NULL };
642 static char Forth_help [] =
643 "In Forth code, tags are words defined by `:',\n\
644 constant, code, create, defer, value, variable, buffer:, field.";
646 static char *Fortran_suffixes [] =
647 { "F", "f", "f90", "for", NULL };
648 static char Fortran_help [] =
649 "In Fortran code, functions, subroutines and block data are tags.";
651 static char *HTML_suffixes [] =
652 { "htm", "html", "shtml", NULL };
653 static char HTML_help [] =
654 "In HTML input files, the tags are the `title' and the `h1', `h2',\n\
655 `h3' headers. Also, tags are `name=' in anchors and all\n\
656 occurrences of `id='.";
658 static char *Lisp_suffixes [] =
659 { "cl", "clisp", "el", "l", "lisp", "LSP", "lsp", "ml", NULL };
660 static char Lisp_help [] =
661 "In Lisp code, any function defined with `defun', any variable\n\
662 defined with `defvar' or `defconst', and in general the first\n\
663 argument of any expression that starts with `(def' in column zero\n\
664 is a tag.";
666 static char *Lua_suffixes [] =
667 { "lua", "LUA", NULL };
668 static char Lua_help [] =
669 "In Lua scripts, all functions are tags.";
671 static char *Makefile_filenames [] =
672 { "Makefile", "makefile", "GNUMakefile", "Makefile.in", "Makefile.am", NULL};
673 static char Makefile_help [] =
674 "In makefiles, targets are tags; additionally, variables are tags\n\
675 unless you specify `--no-globals'.";
677 static char *Objc_suffixes [] =
678 { "lm", /* Objective lex file */
679 "m", /* Objective C file */
680 NULL };
681 static char Objc_help [] =
682 "In Objective C code, tags include Objective C definitions for classes,\n\
683 class categories, methods and protocols. Tags for variables and\n\
684 functions in classes are named `CLASS::VARIABLE' and `CLASS::FUNCTION'.";
686 static char *Pascal_suffixes [] =
687 { "p", "pas", NULL };
688 static char Pascal_help [] =
689 "In Pascal code, the tags are the functions and procedures defined\n\
690 in the file.";
692 static char *Perl_suffixes [] =
693 { "pl", "pm", NULL };
694 static char *Perl_interpreters [] =
695 { "perl", "@PERL@", NULL };
696 static char Perl_help [] =
697 "In Perl code, the tags are the packages, subroutines and variables\n\
698 defined by the `package', `sub', `my' and `local' keywords. Use\n\
699 `--globals' if you want to tag global variables. Tags for\n\
700 subroutines are named `PACKAGE::SUB'. The name for subroutines\n\
701 defined in the default package is `main::SUB'.";
703 static char *PHP_suffixes [] =
704 { "php", "php3", "php4", NULL };
705 static char PHP_help [] =
706 "In PHP code, tags are functions, classes and defines. When using\n\
707 the `--members' option, vars are tags too.";
709 static char *plain_C_suffixes [] =
710 { "pc", /* Pro*C file */
711 NULL };
713 static char *PS_suffixes [] =
714 { "ps", "psw", NULL }; /* .psw is for PSWrap */
715 static char PS_help [] =
716 "In PostScript code, the tags are the functions.";
718 static char *Prolog_suffixes [] =
719 { "prolog", NULL };
720 static char Prolog_help [] =
721 "In Prolog code, tags are predicates and rules at the beginning of\n\
722 line.";
724 static char *Python_suffixes [] =
725 { "py", NULL };
726 static char Python_help [] =
727 "In Python code, `def' or `class' at the beginning of a line\n\
728 generate a tag.";
730 /* Can't do the `SCM' or `scm' prefix with a version number. */
731 static char *Scheme_suffixes [] =
732 { "oak", "sch", "scheme", "SCM", "scm", "SM", "sm", "ss", "t", NULL };
733 static char Scheme_help [] =
734 "In Scheme code, tags include anything defined with `def' or with a\n\
735 construct whose name starts with `def'. They also include\n\
736 variables set with `set!' at top level in the file.";
738 static char *TeX_suffixes [] =
739 { "bib", "clo", "cls", "ltx", "sty", "TeX", "tex", NULL };
740 static char TeX_help [] =
741 "In LaTeX text, the argument of any of the commands `\\chapter',\n\
742 `\\section', `\\subsection', `\\subsubsection', `\\eqno', `\\label',\n\
743 `\\ref', `\\cite', `\\bibitem', `\\part', `\\appendix', `\\entry',\n\
744 `\\index', `\\def', `\\newcommand', `\\renewcommand',\n\
745 `\\newenvironment' or `\\renewenvironment' is a tag.\n\
747 Other commands can be specified by setting the environment variable\n\
748 `TEXTAGS' to a colon-separated list like, for example,\n\
749 TEXTAGS=\"mycommand:myothercommand\".";
752 static char *Texinfo_suffixes [] =
753 { "texi", "texinfo", "txi", NULL };
754 static char Texinfo_help [] =
755 "for texinfo files, lines starting with @node are tagged.";
757 static char *Yacc_suffixes [] =
758 { "y", "y++", "ym", "yxx", "yy", NULL }; /* .ym is Objective yacc file */
759 static char Yacc_help [] =
760 "In Bison or Yacc input files, each rule defines as a tag the\n\
761 nonterminal it constructs. The portions of the file that contain\n\
762 C code are parsed as C code (use --help --lang=c --lang=yacc\n\
763 for full help).";
765 static char auto_help [] =
766 "`auto' is not a real language, it indicates to use\n\
767 a default language for files base on file name suffix and file contents.";
769 static char none_help [] =
770 "`none' is not a real language, it indicates to only do\n\
771 regexp processing on files.";
773 static char no_lang_help [] =
774 "No detailed help available for this language.";
778 * Table of languages.
780 * It is ok for a given function to be listed under more than one
781 * name. I just didn't.
784 static language lang_names [] =
786 { "ada", Ada_help, Ada_funcs, Ada_suffixes },
787 { "asm", Asm_help, Asm_labels, Asm_suffixes },
788 { "c", default_C_help, default_C_entries, default_C_suffixes },
789 { "c++", Cplusplus_help, Cplusplus_entries, Cplusplus_suffixes },
790 { "c*", no_lang_help, Cstar_entries, Cstar_suffixes },
791 { "cobol", Cobol_help, Cobol_paragraphs, Cobol_suffixes },
792 { "erlang", Erlang_help, Erlang_functions, Erlang_suffixes },
793 { "forth", Forth_help, Forth_words, Forth_suffixes },
794 { "fortran", Fortran_help, Fortran_functions, Fortran_suffixes },
795 { "html", HTML_help, HTML_labels, HTML_suffixes },
796 { "java", Cjava_help, Cjava_entries, Cjava_suffixes },
797 { "lisp", Lisp_help, Lisp_functions, Lisp_suffixes },
798 { "lua", Lua_help, Lua_functions, Lua_suffixes },
799 { "makefile", Makefile_help,Makefile_targets,NULL,Makefile_filenames},
800 { "objc", Objc_help, plain_C_entries, Objc_suffixes },
801 { "pascal", Pascal_help, Pascal_functions, Pascal_suffixes },
802 { "perl",Perl_help,Perl_functions,Perl_suffixes,NULL,Perl_interpreters},
803 { "php", PHP_help, PHP_functions, PHP_suffixes },
804 { "postscript",PS_help, PS_functions, PS_suffixes },
805 { "proc", no_lang_help, plain_C_entries, plain_C_suffixes },
806 { "prolog", Prolog_help, Prolog_functions, Prolog_suffixes },
807 { "python", Python_help, Python_functions, Python_suffixes },
808 { "scheme", Scheme_help, Scheme_functions, Scheme_suffixes },
809 { "tex", TeX_help, TeX_commands, TeX_suffixes },
810 { "texinfo", Texinfo_help, Texinfo_nodes, Texinfo_suffixes },
811 { "yacc", Yacc_help,Yacc_entries,Yacc_suffixes,NULL,NULL,TRUE},
812 { "auto", auto_help }, /* default guessing scheme */
813 { "none", none_help, just_read_file }, /* regexp matching only */
814 { NULL } /* end of list */
818 static void
819 print_language_names ()
821 language *lang;
822 char **name, **ext;
824 puts ("\nThese are the currently supported languages, along with the\n\
825 default file names and dot suffixes:");
826 for (lang = lang_names; lang->name != NULL; lang++)
828 printf (" %-*s", 10, lang->name);
829 if (lang->filenames != NULL)
830 for (name = lang->filenames; *name != NULL; name++)
831 printf (" %s", *name);
832 if (lang->suffixes != NULL)
833 for (ext = lang->suffixes; *ext != NULL; ext++)
834 printf (" .%s", *ext);
835 puts ("");
837 puts ("where `auto' means use default language for files based on file\n\
838 name suffix, and `none' means only do regexp processing on files.\n\
839 If no language is specified and no matching suffix is found,\n\
840 the first line of the file is read for a sharp-bang (#!) sequence\n\
841 followed by the name of an interpreter. If no such sequence is found,\n\
842 Fortran is tried first; if no tags are found, C is tried next.\n\
843 When parsing any C file, a \"class\" or \"template\" keyword\n\
844 switches to C++.");
845 puts ("Compressed files are supported using gzip and bzip2.\n\
847 For detailed help on a given language use, for example,\n\
848 etags --help --lang=ada.");
851 #ifndef EMACS_NAME
852 # define EMACS_NAME "standalone"
853 #endif
854 #ifndef VERSION
855 # define VERSION "version"
856 #endif
857 static void
858 print_version ()
860 printf ("%s (%s %s)\n", (CTAGS) ? "ctags" : "etags", EMACS_NAME, VERSION);
861 puts ("Copyright (C) 2002 Free Software Foundation, Inc. and Ken Arnold");
862 puts ("This program is distributed under the same terms as Emacs");
864 exit (EXIT_SUCCESS);
867 static void
868 print_help (argbuffer)
869 argument *argbuffer;
871 bool help_for_lang = FALSE;
873 for (; argbuffer->arg_type != at_end; argbuffer++)
874 if (argbuffer->arg_type == at_language)
876 if (help_for_lang)
877 puts ("");
878 puts (argbuffer->lang->help);
879 help_for_lang = TRUE;
882 if (help_for_lang)
883 exit (EXIT_SUCCESS);
885 printf ("Usage: %s [options] [[regex-option ...] file-name] ...\n\
887 These are the options accepted by %s.\n", progname, progname);
888 if (LONG_OPTIONS)
889 puts ("You may use unambiguous abbreviations for the long option names.");
890 else
891 puts ("Long option names do not work with this executable, as it is not\n\
892 linked with GNU getopt.");
893 puts (" A - as file name means read names from stdin (one per line).\n\
894 Absolute names are stored in the output file as they are.\n\
895 Relative ones are stored relative to the output file's directory.\n");
897 puts ("-a, --append\n\
898 Append tag entries to existing tags file.");
900 puts ("--packages-only\n\
901 For Ada files, only generate tags for packages.");
903 if (CTAGS)
904 puts ("-B, --backward-search\n\
905 Write the search commands for the tag entries using '?', the\n\
906 backward-search command instead of '/', the forward-search command.");
908 /* This option is mostly obsolete, because etags can now automatically
909 detect C++. Retained for backward compatibility and for debugging and
910 experimentation. In principle, we could want to tag as C++ even
911 before any "class" or "template" keyword.
912 puts ("-C, --c++\n\
913 Treat files whose name suffix defaults to C language as C++ files.");
916 puts ("--declarations\n\
917 In C and derived languages, create tags for function declarations,");
918 if (CTAGS)
919 puts ("\tand create tags for extern variables if --globals is used.");
920 else
921 puts
922 ("\tand create tags for extern variables unless --no-globals is used.");
924 if (CTAGS)
925 puts ("-d, --defines\n\
926 Create tag entries for C #define constants and enum constants, too.");
927 else
928 puts ("-D, --no-defines\n\
929 Don't create tag entries for C #define constants and enum constants.\n\
930 This makes the tags file smaller.");
932 if (!CTAGS)
933 puts ("-i FILE, --include=FILE\n\
934 Include a note in tag file indicating that, when searching for\n\
935 a tag, one should also consult the tags file FILE after\n\
936 checking the current file.");
938 puts ("-l LANG, --language=LANG\n\
939 Force the following files to be considered as written in the\n\
940 named language up to the next --language=LANG option.");
942 if (CTAGS)
943 puts ("--globals\n\
944 Create tag entries for global variables in some languages.");
945 else
946 puts ("--no-globals\n\
947 Do not create tag entries for global variables in some\n\
948 languages. This makes the tags file smaller.");
949 puts ("--members\n\
950 Create tag entries for members of structures in some languages.");
952 #ifdef ETAGS_REGEXPS
953 puts ("-r REGEXP, --regex=REGEXP or --regex=@regexfile\n\
954 Make a tag for each line matching a regular expression pattern\n\
955 in the following files. {LANGUAGE}REGEXP uses REGEXP for LANGUAGE\n\
956 files only. REGEXFILE is a file containing one REGEXP per line.\n\
957 REGEXP takes the form /TAGREGEXP/TAGNAME/MODS, where TAGNAME/ is\n\
958 optional. The TAGREGEXP pattern is anchored (as if preceded by ^).");
959 puts (" If TAGNAME/ is present, the tags created are named.\n\
960 For example Tcl named tags can be created with:\n\
961 --regex=\"/proc[ \\t]+\\([^ \\t]+\\)/\\1/.\".\n\
962 MODS are optional one-letter modifiers: `i' means to ignore case,\n\
963 `m' means to allow multi-line matches, `s' implies `m' and\n\
964 causes dot to match any character, including newline.");
965 puts ("-R, --no-regex\n\
966 Don't create tags from regexps for the following files.");
967 #endif /* ETAGS_REGEXPS */
968 puts ("-I, --ignore-indentation\n\
969 In C and C++ do not assume that a closing brace in the first\n\
970 column is the final brace of a function or structure definition.");
971 puts ("-o FILE, --output=FILE\n\
972 Write the tags to FILE.");
973 puts ("--parse-stdin=NAME\n\
974 Read from standard input and record tags as belonging to file NAME.");
976 if (CTAGS)
978 puts ("-t, --typedefs\n\
979 Generate tag entries for C and Ada typedefs.");
980 puts ("-T, --typedefs-and-c++\n\
981 Generate tag entries for C typedefs, C struct/enum/union tags,\n\
982 and C++ member functions.");
985 if (CTAGS)
986 puts ("-u, --update\n\
987 Update the tag entries for the given files, leaving tag\n\
988 entries for other files in place. Currently, this is\n\
989 implemented by deleting the existing entries for the given\n\
990 files and then rewriting the new entries at the end of the\n\
991 tags file. It is often faster to simply rebuild the entire\n\
992 tag file than to use this.");
994 if (CTAGS)
996 puts ("-v, --vgrind\n\
997 Print on the standard output an index of items intended for\n\
998 human consumption, similar to the output of vgrind. The index\n\
999 is sorted, and gives the page number of each item.");
1000 puts ("-w, --no-warn\n\
1001 Suppress warning messages about entries defined in multiple\n\
1002 files.");
1003 puts ("-x, --cxref\n\
1004 Like --vgrind, but in the style of cxref, rather than vgrind.\n\
1005 The output uses line numbers instead of page numbers, but\n\
1006 beyond that the differences are cosmetic; try both to see\n\
1007 which you like.");
1010 puts ("-V, --version\n\
1011 Print the version of the program.\n\
1012 -h, --help\n\
1013 Print this help message.\n\
1014 Followed by one or more `--language' options prints detailed\n\
1015 help about tag generation for the specified languages.");
1017 print_language_names ();
1019 puts ("");
1020 puts ("Report bugs to bug-gnu-emacs@gnu.org");
1022 exit (EXIT_SUCCESS);
1026 #ifdef VMS /* VMS specific functions */
1028 #define EOS '\0'
1030 /* This is a BUG! ANY arbitrary limit is a BUG!
1031 Won't someone please fix this? */
1032 #define MAX_FILE_SPEC_LEN 255
1033 typedef struct {
1034 short curlen;
1035 char body[MAX_FILE_SPEC_LEN + 1];
1036 } vspec;
1039 v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names
1040 returning in each successive call the next file name matching the input
1041 spec. The function expects that each in_spec passed
1042 to it will be processed to completion; in particular, up to and
1043 including the call following that in which the last matching name
1044 is returned, the function ignores the value of in_spec, and will
1045 only start processing a new spec with the following call.
1046 If an error occurs, on return out_spec contains the value
1047 of in_spec when the error occurred.
1049 With each successive file name returned in out_spec, the
1050 function's return value is one. When there are no more matching
1051 names the function returns zero. If on the first call no file
1052 matches in_spec, or there is any other error, -1 is returned.
1055 #include <rmsdef.h>
1056 #include <descrip.h>
1057 #define OUTSIZE MAX_FILE_SPEC_LEN
1058 static short
1059 fn_exp (out, in)
1060 vspec *out;
1061 char *in;
1063 static long context = 0;
1064 static struct dsc$descriptor_s o;
1065 static struct dsc$descriptor_s i;
1066 static bool pass1 = TRUE;
1067 long status;
1068 short retval;
1070 if (pass1)
1072 pass1 = FALSE;
1073 o.dsc$a_pointer = (char *) out;
1074 o.dsc$w_length = (short)OUTSIZE;
1075 i.dsc$a_pointer = in;
1076 i.dsc$w_length = (short)strlen(in);
1077 i.dsc$b_dtype = DSC$K_DTYPE_T;
1078 i.dsc$b_class = DSC$K_CLASS_S;
1079 o.dsc$b_dtype = DSC$K_DTYPE_VT;
1080 o.dsc$b_class = DSC$K_CLASS_VS;
1082 if ((status = lib$find_file(&i, &o, &context, 0, 0)) == RMS$_NORMAL)
1084 out->body[out->curlen] = EOS;
1085 return 1;
1087 else if (status == RMS$_NMF)
1088 retval = 0;
1089 else
1091 strcpy(out->body, in);
1092 retval = -1;
1094 lib$find_file_end(&context);
1095 pass1 = TRUE;
1096 return retval;
1100 v1.01 nmm 19-Aug-85 gfnames - return in successive calls the
1101 name of each file specified by the provided arg expanding wildcards.
1103 static char *
1104 gfnames (arg, p_error)
1105 char *arg;
1106 bool *p_error;
1108 static vspec filename = {MAX_FILE_SPEC_LEN, "\0"};
1110 switch (fn_exp (&filename, arg))
1112 case 1:
1113 *p_error = FALSE;
1114 return filename.body;
1115 case 0:
1116 *p_error = FALSE;
1117 return NULL;
1118 default:
1119 *p_error = TRUE;
1120 return filename.body;
1124 #ifndef OLD /* Newer versions of VMS do provide `system'. */
1125 system (cmd)
1126 char *cmd;
1128 error ("%s", "system() function not implemented under VMS");
1130 #endif
1132 #define VERSION_DELIM ';'
1133 char *massage_name (s)
1134 char *s;
1136 char *start = s;
1138 for ( ; *s; s++)
1139 if (*s == VERSION_DELIM)
1141 *s = EOS;
1142 break;
1144 else
1145 *s = lowcase (*s);
1146 return start;
1148 #endif /* VMS */
1152 main (argc, argv)
1153 int argc;
1154 char *argv[];
1156 int i;
1157 unsigned int nincluded_files;
1158 char **included_files;
1159 argument *argbuffer;
1160 int current_arg, file_count;
1161 linebuffer filename_lb;
1162 bool help_asked = FALSE;
1163 #ifdef VMS
1164 bool got_err;
1165 #endif
1166 char *optstring;
1167 int opt;
1170 #ifdef DOS_NT
1171 _fmode = O_BINARY; /* all of files are treated as binary files */
1172 #endif /* DOS_NT */
1174 progname = argv[0];
1175 nincluded_files = 0;
1176 included_files = xnew (argc, char *);
1177 current_arg = 0;
1178 file_count = 0;
1180 /* Allocate enough no matter what happens. Overkill, but each one
1181 is small. */
1182 argbuffer = xnew (argc, argument);
1185 * If etags, always find typedefs and structure tags. Why not?
1186 * Also default to find macro constants, enum constants and
1187 * global variables.
1189 if (!CTAGS)
1191 typedefs = typedefs_or_cplusplus = constantypedefs = TRUE;
1192 globals = TRUE;
1195 /* When the optstring begins with a '-' getopt_long does not rearrange the
1196 non-options arguments to be at the end, but leaves them alone. */
1197 optstring = "-";
1198 #ifdef ETAGS_REGEXPS
1199 optstring = "-r:Rc:";
1200 #endif /* ETAGS_REGEXPS */
1201 if (!LONG_OPTIONS)
1202 optstring += 1; /* remove the initial '-' */
1203 optstring = concat (optstring,
1204 "aCf:Il:o:SVhH",
1205 (CTAGS) ? "BxdtTuvw" : "Di:");
1207 while ((opt = getopt_long (argc, argv, optstring, longopts, NULL)) != EOF)
1208 switch (opt)
1210 case 0:
1211 /* If getopt returns 0, then it has already processed a
1212 long-named option. We should do nothing. */
1213 break;
1215 case 1:
1216 /* This means that a file name has been seen. Record it. */
1217 argbuffer[current_arg].arg_type = at_filename;
1218 argbuffer[current_arg].what = optarg;
1219 ++current_arg;
1220 ++file_count;
1221 break;
1223 case STDIN:
1224 /* Parse standard input. Idea by Vivek <vivek@etla.org>. */
1225 argbuffer[current_arg].arg_type = at_stdin;
1226 argbuffer[current_arg].what = optarg;
1227 ++current_arg;
1228 ++file_count;
1229 if (parsing_stdin)
1230 fatal ("cannot parse standard input more than once", (char *)NULL);
1231 parsing_stdin = TRUE;
1232 break;
1234 /* Common options. */
1235 case 'a': append_to_tagfile = TRUE; break;
1236 case 'C': cplusplus = TRUE; break;
1237 case 'f': /* for compatibility with old makefiles */
1238 case 'o':
1239 if (tagfile)
1241 error ("-o option may only be given once.", (char *)NULL);
1242 suggest_asking_for_help ();
1243 /* NOTREACHED */
1245 tagfile = optarg;
1246 break;
1247 case 'I':
1248 case 'S': /* for backward compatibility */
1249 ignoreindent = TRUE;
1250 break;
1251 case 'l':
1253 language *lang = get_language_from_langname (optarg);
1254 if (lang != NULL)
1256 argbuffer[current_arg].lang = lang;
1257 argbuffer[current_arg].arg_type = at_language;
1258 ++current_arg;
1261 break;
1262 case 'c':
1263 /* Backward compatibility: support obsolete --ignore-case-regexp. */
1264 optarg = concat (optarg, "i", ""); /* memory leak here */
1265 /* FALLTHRU */
1266 case 'r':
1267 argbuffer[current_arg].arg_type = at_regexp;
1268 argbuffer[current_arg].what = optarg;
1269 ++current_arg;
1270 break;
1271 case 'R':
1272 argbuffer[current_arg].arg_type = at_regexp;
1273 argbuffer[current_arg].what = NULL;
1274 ++current_arg;
1275 break;
1276 case 'V':
1277 print_version ();
1278 break;
1279 case 'h':
1280 case 'H':
1281 help_asked = TRUE;
1282 break;
1284 /* Etags options */
1285 case 'D': constantypedefs = FALSE; break;
1286 case 'i': included_files[nincluded_files++] = optarg; break;
1288 /* Ctags options. */
1289 case 'B': searchar = '?'; break;
1290 case 'd': constantypedefs = TRUE; break;
1291 case 't': typedefs = TRUE; break;
1292 case 'T': typedefs = typedefs_or_cplusplus = TRUE; break;
1293 case 'u': update = TRUE; break;
1294 case 'v': vgrind_style = TRUE; /*FALLTHRU*/
1295 case 'x': cxref_style = TRUE; break;
1296 case 'w': no_warnings = TRUE; break;
1297 default:
1298 suggest_asking_for_help ();
1299 /* NOTREACHED */
1302 /* No more options. Store the rest of arguments. */
1303 for (; optind < argc; optind++)
1305 argbuffer[current_arg].arg_type = at_filename;
1306 argbuffer[current_arg].what = argv[optind];
1307 ++current_arg;
1308 ++file_count;
1311 argbuffer[current_arg].arg_type = at_end;
1313 if (help_asked)
1314 print_help (argbuffer);
1315 /* NOTREACHED */
1317 if (nincluded_files == 0 && file_count == 0)
1319 error ("no input files specified.", (char *)NULL);
1320 suggest_asking_for_help ();
1321 /* NOTREACHED */
1324 if (tagfile == NULL)
1325 tagfile = CTAGS ? "tags" : "TAGS";
1326 cwd = etags_getcwd (); /* the current working directory */
1327 if (cwd[strlen (cwd) - 1] != '/')
1329 char *oldcwd = cwd;
1330 cwd = concat (oldcwd, "/", "");
1331 free (oldcwd);
1333 /* Relative file names are made relative to the current directory. */
1334 if (streq (tagfile, "-")
1335 || strneq (tagfile, "/dev/", 5))
1336 tagfiledir = cwd;
1337 else
1338 tagfiledir = absolute_dirname (tagfile, cwd);
1340 init (); /* set up boolean "functions" */
1342 linebuffer_init (&lb);
1343 linebuffer_init (&filename_lb);
1344 linebuffer_init (&filebuf);
1345 linebuffer_init (&token_name);
1347 if (!CTAGS)
1349 if (streq (tagfile, "-"))
1351 tagf = stdout;
1352 #ifdef DOS_NT
1353 /* Switch redirected `stdout' to binary mode (setting `_fmode'
1354 doesn't take effect until after `stdout' is already open). */
1355 if (!isatty (fileno (stdout)))
1356 setmode (fileno (stdout), O_BINARY);
1357 #endif /* DOS_NT */
1359 else
1360 tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1361 if (tagf == NULL)
1362 pfatal (tagfile);
1366 * Loop through files finding functions.
1368 for (i = 0; i < current_arg; i++)
1370 static language *lang; /* non-NULL if language is forced */
1371 char *this_file;
1373 switch (argbuffer[i].arg_type)
1375 case at_language:
1376 lang = argbuffer[i].lang;
1377 break;
1378 #ifdef ETAGS_REGEXPS
1379 case at_regexp:
1380 analyse_regex (argbuffer[i].what);
1381 break;
1382 #endif
1383 case at_filename:
1384 #ifdef VMS
1385 while ((this_file = gfnames (argbuffer[i].what, &got_err)) != NULL)
1387 if (got_err)
1389 error ("can't find file %s\n", this_file);
1390 argc--, argv++;
1392 else
1394 this_file = massage_name (this_file);
1396 #else
1397 this_file = argbuffer[i].what;
1398 #endif
1399 /* Input file named "-" means read file names from stdin
1400 (one per line) and use them. */
1401 if (streq (this_file, "-"))
1403 if (parsing_stdin)
1404 fatal ("cannot parse standard input AND read file names from it",
1405 (char *)NULL);
1406 while (readline_internal (&filename_lb, stdin) > 0)
1407 process_file_name (filename_lb.buffer, lang);
1409 else
1410 process_file_name (this_file, lang);
1411 #ifdef VMS
1413 #endif
1414 break;
1415 case at_stdin:
1416 this_file = argbuffer[i].what;
1417 process_file (stdin, this_file, lang);
1418 break;
1422 #ifdef ETAGS_REGEXPS
1423 free_regexps ();
1424 #endif /* ETAGS_REGEXPS */
1425 free (lb.buffer);
1426 free (filebuf.buffer);
1427 free (token_name.buffer);
1429 if (!CTAGS || cxref_style)
1431 /* Write the remaining tags to tagf (ETAGS) or stdout (CXREF). */
1432 put_entries (nodehead);
1433 free_tree (nodehead);
1434 nodehead = NULL;
1435 if (!CTAGS)
1437 fdesc *fdp;
1439 /* Output file entries that have no tags. */
1440 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
1441 if (!fdp->written)
1442 fprintf (tagf, "\f\n%s,0\n", fdp->taggedfname);
1444 while (nincluded_files-- > 0)
1445 fprintf (tagf, "\f\n%s,include\n", *included_files++);
1447 if (fclose (tagf) == EOF)
1448 pfatal (tagfile);
1451 exit (EXIT_SUCCESS);
1454 if (update)
1456 char cmd[BUFSIZ];
1457 for (i = 0; i < current_arg; ++i)
1459 switch (argbuffer[i].arg_type)
1461 case at_filename:
1462 case at_stdin:
1463 break;
1464 default:
1465 continue; /* the for loop */
1467 sprintf (cmd,
1468 "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS",
1469 tagfile, argbuffer[i].what, tagfile);
1470 if (system (cmd) != EXIT_SUCCESS)
1471 fatal ("failed to execute shell command", (char *)NULL);
1473 append_to_tagfile = TRUE;
1476 tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1477 if (tagf == NULL)
1478 pfatal (tagfile);
1479 put_entries (nodehead); /* write all the tags (CTAGS) */
1480 free_tree (nodehead);
1481 nodehead = NULL;
1482 if (fclose (tagf) == EOF)
1483 pfatal (tagfile);
1485 if (CTAGS)
1486 if (append_to_tagfile || update)
1488 char cmd[2*BUFSIZ+10];
1489 sprintf (cmd, "sort -o %.*s %.*s", BUFSIZ, tagfile, BUFSIZ, tagfile);
1490 exit (system (cmd));
1492 return EXIT_SUCCESS;
1497 * Return a compressor given the file name. If EXTPTR is non-zero,
1498 * return a pointer into FILE where the compressor-specific
1499 * extension begins. If no compressor is found, NULL is returned
1500 * and EXTPTR is not significant.
1501 * Idea by Vladimir Alexiev <vladimir@cs.ualberta.ca> (1998)
1503 static compressor *
1504 get_compressor_from_suffix (file, extptr)
1505 char *file;
1506 char **extptr;
1508 compressor *compr;
1509 char *slash, *suffix;
1511 /* This relies on FN to be after canonicalize_filename,
1512 so we don't need to consider backslashes on DOS_NT. */
1513 slash = etags_strrchr (file, '/');
1514 suffix = etags_strrchr (file, '.');
1515 if (suffix == NULL || suffix < slash)
1516 return NULL;
1517 if (extptr != NULL)
1518 *extptr = suffix;
1519 suffix += 1;
1520 /* Let those poor souls who live with DOS 8+3 file name limits get
1521 some solace by treating foo.cgz as if it were foo.c.gz, etc.
1522 Only the first do loop is run if not MSDOS */
1525 for (compr = compressors; compr->suffix != NULL; compr++)
1526 if (streq (compr->suffix, suffix))
1527 return compr;
1528 if (!MSDOS)
1529 break; /* do it only once: not really a loop */
1530 if (extptr != NULL)
1531 *extptr = ++suffix;
1532 } while (*suffix != '\0');
1533 return NULL;
1539 * Return a language given the name.
1541 static language *
1542 get_language_from_langname (name)
1543 const char *name;
1545 language *lang;
1547 if (name == NULL)
1548 error ("empty language name", (char *)NULL);
1549 else
1551 for (lang = lang_names; lang->name != NULL; lang++)
1552 if (streq (name, lang->name))
1553 return lang;
1554 error ("unknown language \"%s\"", name);
1557 return NULL;
1562 * Return a language given the interpreter name.
1564 static language *
1565 get_language_from_interpreter (interpreter)
1566 char *interpreter;
1568 language *lang;
1569 char **iname;
1571 if (interpreter == NULL)
1572 return NULL;
1573 for (lang = lang_names; lang->name != NULL; lang++)
1574 if (lang->interpreters != NULL)
1575 for (iname = lang->interpreters; *iname != NULL; iname++)
1576 if (streq (*iname, interpreter))
1577 return lang;
1579 return NULL;
1585 * Return a language given the file name.
1587 static language *
1588 get_language_from_filename (file, case_sensitive)
1589 char *file;
1590 bool case_sensitive;
1592 language *lang;
1593 char **name, **ext, *suffix;
1595 /* Try whole file name first. */
1596 for (lang = lang_names; lang->name != NULL; lang++)
1597 if (lang->filenames != NULL)
1598 for (name = lang->filenames; *name != NULL; name++)
1599 if ((case_sensitive)
1600 ? streq (*name, file)
1601 : strcaseeq (*name, file))
1602 return lang;
1604 /* If not found, try suffix after last dot. */
1605 suffix = etags_strrchr (file, '.');
1606 if (suffix == NULL)
1607 return NULL;
1608 suffix += 1;
1609 for (lang = lang_names; lang->name != NULL; lang++)
1610 if (lang->suffixes != NULL)
1611 for (ext = lang->suffixes; *ext != NULL; ext++)
1612 if ((case_sensitive)
1613 ? streq (*ext, suffix)
1614 : strcaseeq (*ext, suffix))
1615 return lang;
1616 return NULL;
1621 * This routine is called on each file argument.
1623 static void
1624 process_file_name (file, lang)
1625 char *file;
1626 language *lang;
1628 struct stat stat_buf;
1629 FILE *inf;
1630 fdesc *fdp;
1631 compressor *compr;
1632 char *compressed_name, *uncompressed_name;
1633 char *ext, *real_name;
1634 int retval;
1636 canonicalize_filename (file);
1637 if (streq (file, tagfile) && !streq (tagfile, "-"))
1639 error ("skipping inclusion of %s in self.", file);
1640 return;
1642 if ((compr = get_compressor_from_suffix (file, &ext)) == NULL)
1644 compressed_name = NULL;
1645 real_name = uncompressed_name = savestr (file);
1647 else
1649 real_name = compressed_name = savestr (file);
1650 uncompressed_name = savenstr (file, ext - file);
1653 /* If the canonicalized uncompressed name
1654 has already been dealt with, skip it silently. */
1655 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
1657 assert (fdp->infname != NULL);
1658 if (streq (uncompressed_name, fdp->infname))
1659 goto cleanup;
1662 if (stat (real_name, &stat_buf) != 0)
1664 /* Reset real_name and try with a different name. */
1665 real_name = NULL;
1666 if (compressed_name != NULL) /* try with the given suffix */
1668 if (stat (uncompressed_name, &stat_buf) == 0)
1669 real_name = uncompressed_name;
1671 else /* try all possible suffixes */
1673 for (compr = compressors; compr->suffix != NULL; compr++)
1675 compressed_name = concat (file, ".", compr->suffix);
1676 if (stat (compressed_name, &stat_buf) != 0)
1678 if (MSDOS)
1680 char *suf = compressed_name + strlen (file);
1681 size_t suflen = strlen (compr->suffix) + 1;
1682 for ( ; suf[1]; suf++, suflen--)
1684 memmove (suf, suf + 1, suflen);
1685 if (stat (compressed_name, &stat_buf) == 0)
1687 real_name = compressed_name;
1688 break;
1691 if (real_name != NULL)
1692 break;
1693 } /* MSDOS */
1694 free (compressed_name);
1695 compressed_name = NULL;
1697 else
1699 real_name = compressed_name;
1700 break;
1704 if (real_name == NULL)
1706 perror (file);
1707 goto cleanup;
1709 } /* try with a different name */
1711 if (!S_ISREG (stat_buf.st_mode))
1713 error ("skipping %s: it is not a regular file.", real_name);
1714 goto cleanup;
1716 if (real_name == compressed_name)
1718 char *cmd = concat (compr->command, " ", real_name);
1719 inf = (FILE *) popen (cmd, "r");
1720 free (cmd);
1722 else
1723 inf = fopen (real_name, "r");
1724 if (inf == NULL)
1726 perror (real_name);
1727 goto cleanup;
1730 process_file (inf, uncompressed_name, lang);
1732 if (real_name == compressed_name)
1733 retval = pclose (inf);
1734 else
1735 retval = fclose (inf);
1736 if (retval < 0)
1737 pfatal (file);
1739 cleanup:
1740 if (compressed_name) free (compressed_name);
1741 if (uncompressed_name) free (uncompressed_name);
1742 last_node = NULL;
1743 curfdp = NULL;
1744 return;
1747 static void
1748 process_file (fh, fn, lang)
1749 FILE *fh;
1750 char *fn;
1751 language *lang;
1753 static const fdesc emptyfdesc;
1754 fdesc *fdp;
1756 /* Create a new input file description entry. */
1757 fdp = xnew (1, fdesc);
1758 *fdp = emptyfdesc;
1759 fdp->next = fdhead;
1760 fdp->infname = savestr (fn);
1761 fdp->lang = lang;
1762 fdp->infabsname = absolute_filename (fn, cwd);
1763 fdp->infabsdir = absolute_dirname (fn, cwd);
1764 if (filename_is_absolute (fn))
1766 /* An absolute file name. Canonicalize it. */
1767 fdp->taggedfname = absolute_filename (fn, NULL);
1769 else
1771 /* A file name relative to cwd. Make it relative
1772 to the directory of the tags file. */
1773 fdp->taggedfname = relative_filename (fn, tagfiledir);
1775 fdp->usecharno = TRUE; /* use char position when making tags */
1776 fdp->prop = NULL;
1777 fdp->written = FALSE; /* not written on tags file yet */
1779 fdhead = fdp;
1780 curfdp = fdhead; /* the current file description */
1782 find_entries (fh);
1784 /* If not Ctags, and if this is not metasource and if it contained no #line
1785 directives, we can write the tags and free all nodes pointing to
1786 curfdp. */
1787 if (!CTAGS
1788 && curfdp->usecharno /* no #line directives in this file */
1789 && !curfdp->lang->metasource)
1791 node *np, *prev;
1793 /* Look for the head of the sublist relative to this file. See add_node
1794 for the structure of the node tree. */
1795 prev = NULL;
1796 for (np = nodehead; np != NULL; prev = np, np = np->left)
1797 if (np->fdp == curfdp)
1798 break;
1800 /* If we generated tags for this file, write and delete them. */
1801 if (np != NULL)
1803 /* This is the head of the last sublist, if any. The following
1804 instructions depend on this being true. */
1805 assert (np->left == NULL);
1807 assert (fdhead == curfdp);
1808 assert (last_node->fdp == curfdp);
1809 put_entries (np); /* write tags for file curfdp->taggedfname */
1810 free_tree (np); /* remove the written nodes */
1811 if (prev == NULL)
1812 nodehead = NULL; /* no nodes left */
1813 else
1814 prev->left = NULL; /* delete the pointer to the sublist */
1820 * This routine sets up the boolean pseudo-functions which work
1821 * by setting boolean flags dependent upon the corresponding character.
1822 * Every char which is NOT in that string is not a white char. Therefore,
1823 * all of the array "_wht" is set to FALSE, and then the elements
1824 * subscripted by the chars in "white" are set to TRUE. Thus "_wht"
1825 * of a char is TRUE if it is the string "white", else FALSE.
1827 static void
1828 init ()
1830 register char *sp;
1831 register int i;
1833 for (i = 0; i < CHARS; i++)
1834 iswhite(i) = notinname(i) = begtoken(i) = intoken(i) = endtoken(i) = FALSE;
1835 for (sp = white; *sp != '\0'; sp++) iswhite (*sp) = TRUE;
1836 for (sp = nonam; *sp != '\0'; sp++) notinname (*sp) = TRUE;
1837 notinname('\0') = notinname('\n');
1838 for (sp = begtk; *sp != '\0'; sp++) begtoken (*sp) = TRUE;
1839 begtoken('\0') = begtoken('\n');
1840 for (sp = midtk; *sp != '\0'; sp++) intoken (*sp) = TRUE;
1841 intoken('\0') = intoken('\n');
1842 for (sp = endtk; *sp != '\0'; sp++) endtoken (*sp) = TRUE;
1843 endtoken('\0') = endtoken('\n');
1847 * This routine opens the specified file and calls the function
1848 * which finds the function and type definitions.
1850 static void
1851 find_entries (inf)
1852 FILE *inf;
1854 char *cp;
1855 language *lang = curfdp->lang;
1856 Lang_function *parser = NULL;
1858 /* If user specified a language, use it. */
1859 if (lang != NULL && lang->function != NULL)
1861 parser = lang->function;
1864 /* Else try to guess the language given the file name. */
1865 if (parser == NULL)
1867 lang = get_language_from_filename (curfdp->infname, TRUE);
1868 if (lang != NULL && lang->function != NULL)
1870 curfdp->lang = lang;
1871 parser = lang->function;
1875 /* Else look for sharp-bang as the first two characters. */
1876 if (parser == NULL
1877 && readline_internal (&lb, inf) > 0
1878 && lb.len >= 2
1879 && lb.buffer[0] == '#'
1880 && lb.buffer[1] == '!')
1882 char *lp;
1884 /* Set lp to point at the first char after the last slash in the
1885 line or, if no slashes, at the first nonblank. Then set cp to
1886 the first successive blank and terminate the string. */
1887 lp = etags_strrchr (lb.buffer+2, '/');
1888 if (lp != NULL)
1889 lp += 1;
1890 else
1891 lp = skip_spaces (lb.buffer + 2);
1892 cp = skip_non_spaces (lp);
1893 *cp = '\0';
1895 if (strlen (lp) > 0)
1897 lang = get_language_from_interpreter (lp);
1898 if (lang != NULL && lang->function != NULL)
1900 curfdp->lang = lang;
1901 parser = lang->function;
1906 /* We rewind here, even if inf may be a pipe. We fail if the
1907 length of the first line is longer than the pipe block size,
1908 which is unlikely. */
1909 rewind (inf);
1911 /* Else try to guess the language given the case insensitive file name. */
1912 if (parser == NULL)
1914 lang = get_language_from_filename (curfdp->infname, FALSE);
1915 if (lang != NULL && lang->function != NULL)
1917 curfdp->lang = lang;
1918 parser = lang->function;
1922 /* Else try Fortran or C. */
1923 if (parser == NULL)
1925 node *old_last_node = last_node;
1927 curfdp->lang = get_language_from_langname ("fortran");
1928 find_entries (inf);
1930 if (old_last_node == last_node)
1931 /* No Fortran entries found. Try C. */
1933 /* We do not tag if rewind fails.
1934 Only the file name will be recorded in the tags file. */
1935 rewind (inf);
1936 curfdp->lang = get_language_from_langname (cplusplus ? "c++" : "c");
1937 find_entries (inf);
1939 return;
1942 if (!no_line_directive
1943 && curfdp->lang != NULL && curfdp->lang->metasource)
1944 /* It may be that this is a bingo.y file, and we already parsed a bingo.c
1945 file, or anyway we parsed a file that is automatically generated from
1946 this one. If this is the case, the bingo.c file contained #line
1947 directives that generated tags pointing to this file. Let's delete
1948 them all before parsing this file, which is the real source. */
1950 fdesc **fdpp = &fdhead;
1951 while (*fdpp != NULL)
1952 if (*fdpp != curfdp
1953 && streq ((*fdpp)->taggedfname, curfdp->taggedfname))
1954 /* We found one of those! We must delete both the file description
1955 and all tags referring to it. */
1957 fdesc *badfdp = *fdpp;
1959 /* Delete the tags referring to badfdp->taggedfname
1960 that were obtained from badfdp->infname. */
1961 invalidate_nodes (badfdp, &nodehead);
1963 *fdpp = badfdp->next; /* remove the bad description from the list */
1964 free_fdesc (badfdp);
1966 else
1967 fdpp = &(*fdpp)->next; /* advance the list pointer */
1970 assert (parser != NULL);
1972 /* Generic initialisations before reading from file. */
1973 linebuffer_setlen (&filebuf, 0); /* reset the file buffer */
1975 /* Generic initialisations before parsing file with readline. */
1976 lineno = 0; /* reset global line number */
1977 charno = 0; /* reset global char number */
1978 linecharno = 0; /* reset global char number of line start */
1980 parser (inf);
1982 #ifdef ETAGS_REGEXPS
1983 regex_tag_multiline ();
1984 #endif /* ETAGS_REGEXPS */
1989 * Check whether an implicitly named tag should be created,
1990 * then call `pfnote'.
1991 * NAME is a string that is internally copied by this function.
1993 * TAGS format specification
1994 * Idea by Sam Kendall <kendall@mv.mv.com> (1997)
1995 * The following is explained in some more detail in etc/ETAGS.EBNF.
1997 * make_tag creates tags with "implicit tag names" (unnamed tags)
1998 * if the following are all true, assuming NONAM=" \f\t\n\r()=,;":
1999 * 1. NAME does not contain any of the characters in NONAM;
2000 * 2. LINESTART contains name as either a rightmost, or rightmost but
2001 * one character, substring;
2002 * 3. the character, if any, immediately before NAME in LINESTART must
2003 * be a character in NONAM;
2004 * 4. the character, if any, immediately after NAME in LINESTART must
2005 * also be a character in NONAM.
2007 * The implementation uses the notinname() macro, which recognises the
2008 * characters stored in the string `nonam'.
2009 * etags.el needs to use the same characters that are in NONAM.
2011 static void
2012 make_tag (name, namelen, is_func, linestart, linelen, lno, cno)
2013 char *name; /* tag name, or NULL if unnamed */
2014 int namelen; /* tag length */
2015 bool is_func; /* tag is a function */
2016 char *linestart; /* start of the line where tag is */
2017 int linelen; /* length of the line where tag is */
2018 int lno; /* line number */
2019 long cno; /* character number */
2021 bool named = (name != NULL && namelen > 0);
2023 if (!CTAGS && named) /* maybe set named to false */
2024 /* Let's try to make an implicit tag name, that is, create an unnamed tag
2025 such that etags.el can guess a name from it. */
2027 int i;
2028 register char *cp = name;
2030 for (i = 0; i < namelen; i++)
2031 if (notinname (*cp++))
2032 break;
2033 if (i == namelen) /* rule #1 */
2035 cp = linestart + linelen - namelen;
2036 if (notinname (linestart[linelen-1]))
2037 cp -= 1; /* rule #4 */
2038 if (cp >= linestart /* rule #2 */
2039 && (cp == linestart
2040 || notinname (cp[-1])) /* rule #3 */
2041 && strneq (name, cp, namelen)) /* rule #2 */
2042 named = FALSE; /* use implicit tag name */
2046 if (named)
2047 name = savenstr (name, namelen);
2048 else
2049 name = NULL;
2050 pfnote (name, is_func, linestart, linelen, lno, cno);
2053 /* Record a tag. */
2054 static void
2055 pfnote (name, is_func, linestart, linelen, lno, cno)
2056 char *name; /* tag name, or NULL if unnamed */
2057 bool is_func; /* tag is a function */
2058 char *linestart; /* start of the line where tag is */
2059 int linelen; /* length of the line where tag is */
2060 int lno; /* line number */
2061 long cno; /* character number */
2063 register node *np;
2065 assert (name == NULL || name[0] != '\0');
2066 if (CTAGS && name == NULL)
2067 return;
2069 np = xnew (1, node);
2071 /* If ctags mode, change name "main" to M<thisfilename>. */
2072 if (CTAGS && !cxref_style && streq (name, "main"))
2074 register char *fp = etags_strrchr (curfdp->taggedfname, '/');
2075 np->name = concat ("M", fp == NULL ? curfdp->taggedfname : fp + 1, "");
2076 fp = etags_strrchr (np->name, '.');
2077 if (fp != NULL && fp[1] != '\0' && fp[2] == '\0')
2078 fp[0] = '\0';
2080 else
2081 np->name = name;
2082 np->valid = TRUE;
2083 np->been_warned = FALSE;
2084 np->fdp = curfdp;
2085 np->is_func = is_func;
2086 np->lno = lno;
2087 if (np->fdp->usecharno)
2088 /* Our char numbers are 0-base, because of C language tradition?
2089 ctags compatibility? old versions compatibility? I don't know.
2090 Anyway, since emacs's are 1-base we expect etags.el to take care
2091 of the difference. If we wanted to have 1-based numbers, we would
2092 uncomment the +1 below. */
2093 np->cno = cno /* + 1 */ ;
2094 else
2095 np->cno = invalidcharno;
2096 np->left = np->right = NULL;
2097 if (CTAGS && !cxref_style)
2099 if (strlen (linestart) < 50)
2100 np->regex = concat (linestart, "$", "");
2101 else
2102 np->regex = savenstr (linestart, 50);
2104 else
2105 np->regex = savenstr (linestart, linelen);
2107 add_node (np, &nodehead);
2111 * free_tree ()
2112 * recurse on left children, iterate on right children.
2114 static void
2115 free_tree (np)
2116 register node *np;
2118 while (np)
2120 register node *node_right = np->right;
2121 free_tree (np->left);
2122 if (np->name != NULL)
2123 free (np->name);
2124 free (np->regex);
2125 free (np);
2126 np = node_right;
2131 * free_fdesc ()
2132 * delete a file description
2134 static void
2135 free_fdesc (fdp)
2136 register fdesc *fdp;
2138 if (fdp->infname != NULL) free (fdp->infname);
2139 if (fdp->infabsname != NULL) free (fdp->infabsname);
2140 if (fdp->infabsdir != NULL) free (fdp->infabsdir);
2141 if (fdp->taggedfname != NULL) free (fdp->taggedfname);
2142 if (fdp->prop != NULL) free (fdp->prop);
2143 free (fdp);
2147 * add_node ()
2148 * Adds a node to the tree of nodes. In etags mode, sort by file
2149 * name. In ctags mode, sort by tag name. Make no attempt at
2150 * balancing.
2152 * add_node is the only function allowed to add nodes, so it can
2153 * maintain state.
2155 static void
2156 add_node (np, cur_node_p)
2157 node *np, **cur_node_p;
2159 register int dif;
2160 register node *cur_node = *cur_node_p;
2162 if (cur_node == NULL)
2164 *cur_node_p = np;
2165 last_node = np;
2166 return;
2169 if (!CTAGS)
2170 /* Etags Mode */
2172 /* For each file name, tags are in a linked sublist on the right
2173 pointer. The first tags of different files are a linked list
2174 on the left pointer. last_node points to the end of the last
2175 used sublist. */
2176 if (last_node != NULL && last_node->fdp == np->fdp)
2178 /* Let's use the same sublist as the last added node. */
2179 assert (last_node->right == NULL);
2180 last_node->right = np;
2181 last_node = np;
2183 else if (cur_node->fdp == np->fdp)
2185 /* Scanning the list we found the head of a sublist which is
2186 good for us. Let's scan this sublist. */
2187 add_node (np, &cur_node->right);
2189 else
2190 /* The head of this sublist is not good for us. Let's try the
2191 next one. */
2192 add_node (np, &cur_node->left);
2193 } /* if ETAGS mode */
2195 else
2197 /* Ctags Mode */
2198 dif = strcmp (np->name, cur_node->name);
2201 * If this tag name matches an existing one, then
2202 * do not add the node, but maybe print a warning.
2204 if (!dif)
2206 if (np->fdp == cur_node->fdp)
2208 if (!no_warnings)
2210 fprintf (stderr, "Duplicate entry in file %s, line %d: %s\n",
2211 np->fdp->infname, lineno, np->name);
2212 fprintf (stderr, "Second entry ignored\n");
2215 else if (!cur_node->been_warned && !no_warnings)
2217 fprintf
2218 (stderr,
2219 "Duplicate entry in files %s and %s: %s (Warning only)\n",
2220 np->fdp->infname, cur_node->fdp->infname, np->name);
2221 cur_node->been_warned = TRUE;
2223 return;
2226 /* Actually add the node */
2227 add_node (np, dif < 0 ? &cur_node->left : &cur_node->right);
2228 } /* if CTAGS mode */
2232 * invalidate_nodes ()
2233 * Scan the node tree and invalidate all nodes pointing to the
2234 * given file description (CTAGS case) or free them (ETAGS case).
2236 static void
2237 invalidate_nodes (badfdp, npp)
2238 fdesc *badfdp;
2239 node **npp;
2241 node *np = *npp;
2243 if (np == NULL)
2244 return;
2246 if (CTAGS)
2248 if (np->left != NULL)
2249 invalidate_nodes (badfdp, &np->left);
2250 if (np->fdp == badfdp)
2251 np->valid = FALSE;
2252 if (np->right != NULL)
2253 invalidate_nodes (badfdp, &np->right);
2255 else
2257 assert (np->fdp != NULL);
2258 if (np->fdp == badfdp)
2260 *npp = np->left; /* detach the sublist from the list */
2261 np->left = NULL; /* isolate it */
2262 free_tree (np); /* free it */
2263 invalidate_nodes (badfdp, npp);
2265 else
2266 invalidate_nodes (badfdp, &np->left);
2271 static int total_size_of_entries __P((node *));
2272 static int number_len __P((long));
2274 /* Length of a non-negative number's decimal representation. */
2275 static int
2276 number_len (num)
2277 long num;
2279 int len = 1;
2280 while ((num /= 10) > 0)
2281 len += 1;
2282 return len;
2286 * Return total number of characters that put_entries will output for
2287 * the nodes in the linked list at the right of the specified node.
2288 * This count is irrelevant with etags.el since emacs 19.34 at least,
2289 * but is still supplied for backward compatibility.
2291 static int
2292 total_size_of_entries (np)
2293 register node *np;
2295 register int total = 0;
2297 for (; np != NULL; np = np->right)
2298 if (np->valid)
2300 total += strlen (np->regex) + 1; /* pat\177 */
2301 if (np->name != NULL)
2302 total += strlen (np->name) + 1; /* name\001 */
2303 total += number_len ((long) np->lno) + 1; /* lno, */
2304 if (np->cno != invalidcharno) /* cno */
2305 total += number_len (np->cno);
2306 total += 1; /* newline */
2309 return total;
2312 static void
2313 put_entries (np)
2314 register node *np;
2316 register char *sp;
2317 static fdesc *fdp = NULL;
2319 if (np == NULL)
2320 return;
2322 /* Output subentries that precede this one */
2323 if (CTAGS)
2324 put_entries (np->left);
2326 /* Output this entry */
2327 if (np->valid)
2329 if (!CTAGS)
2331 /* Etags mode */
2332 if (fdp != np->fdp)
2334 fdp = np->fdp;
2335 fprintf (tagf, "\f\n%s,%d\n",
2336 fdp->taggedfname, total_size_of_entries (np));
2337 fdp->written = TRUE;
2339 fputs (np->regex, tagf);
2340 fputc ('\177', tagf);
2341 if (np->name != NULL)
2343 fputs (np->name, tagf);
2344 fputc ('\001', tagf);
2346 fprintf (tagf, "%d,", np->lno);
2347 if (np->cno != invalidcharno)
2348 fprintf (tagf, "%ld", np->cno);
2349 fputs ("\n", tagf);
2351 else
2353 /* Ctags mode */
2354 if (np->name == NULL)
2355 error ("internal error: NULL name in ctags mode.", (char *)NULL);
2357 if (cxref_style)
2359 if (vgrind_style)
2360 fprintf (stdout, "%s %s %d\n",
2361 np->name, np->fdp->taggedfname, (np->lno + 63) / 64);
2362 else
2363 fprintf (stdout, "%-16s %3d %-16s %s\n",
2364 np->name, np->lno, np->fdp->taggedfname, np->regex);
2366 else
2368 fprintf (tagf, "%s\t%s\t", np->name, np->fdp->taggedfname);
2370 if (np->is_func)
2371 { /* function or #define macro with args */
2372 putc (searchar, tagf);
2373 putc ('^', tagf);
2375 for (sp = np->regex; *sp; sp++)
2377 if (*sp == '\\' || *sp == searchar)
2378 putc ('\\', tagf);
2379 putc (*sp, tagf);
2381 putc (searchar, tagf);
2383 else
2384 { /* anything else; text pattern inadequate */
2385 fprintf (tagf, "%d", np->lno);
2387 putc ('\n', tagf);
2390 } /* if this node contains a valid tag */
2392 /* Output subentries that follow this one */
2393 put_entries (np->right);
2394 if (!CTAGS)
2395 put_entries (np->left);
2399 /* C extensions. */
2400 #define C_EXT 0x00fff /* C extensions */
2401 #define C_PLAIN 0x00000 /* C */
2402 #define C_PLPL 0x00001 /* C++ */
2403 #define C_STAR 0x00003 /* C* */
2404 #define C_JAVA 0x00005 /* JAVA */
2405 #define C_AUTO 0x01000 /* C, but switch to C++ if `class' is met */
2406 #define YACC 0x10000 /* yacc file */
2409 * The C symbol tables.
2411 enum sym_type
2413 st_none,
2414 st_C_objprot, st_C_objimpl, st_C_objend,
2415 st_C_gnumacro,
2416 st_C_ignore, st_C_attribute,
2417 st_C_javastruct,
2418 st_C_operator,
2419 st_C_class, st_C_template,
2420 st_C_struct, st_C_extern, st_C_enum, st_C_define, st_C_typedef
2423 static unsigned int hash __P((const char *, unsigned int));
2424 static struct C_stab_entry * in_word_set __P((const char *, unsigned int));
2425 static enum sym_type C_symtype __P((char *, int, int));
2427 /* Feed stuff between (but not including) %[ and %] lines to:
2428 gperf -m 5
2430 %compare-strncmp
2431 %enum
2432 %struct-type
2433 struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
2435 if, 0, st_C_ignore
2436 for, 0, st_C_ignore
2437 while, 0, st_C_ignore
2438 switch, 0, st_C_ignore
2439 return, 0, st_C_ignore
2440 __attribute__, 0, st_C_attribute
2441 @interface, 0, st_C_objprot
2442 @protocol, 0, st_C_objprot
2443 @implementation,0, st_C_objimpl
2444 @end, 0, st_C_objend
2445 import, (C_JAVA & !C_PLPL), st_C_ignore
2446 package, (C_JAVA & !C_PLPL), st_C_ignore
2447 friend, C_PLPL, st_C_ignore
2448 extends, (C_JAVA & !C_PLPL), st_C_javastruct
2449 implements, (C_JAVA & !C_PLPL), st_C_javastruct
2450 interface, (C_JAVA & !C_PLPL), st_C_struct
2451 class, 0, st_C_class
2452 namespace, C_PLPL, st_C_struct
2453 domain, C_STAR, st_C_struct
2454 union, 0, st_C_struct
2455 struct, 0, st_C_struct
2456 extern, 0, st_C_extern
2457 enum, 0, st_C_enum
2458 typedef, 0, st_C_typedef
2459 define, 0, st_C_define
2460 operator, C_PLPL, st_C_operator
2461 template, 0, st_C_template
2462 # DEFUN used in emacs, the next three used in glibc (SYSCALL only for mach).
2463 DEFUN, 0, st_C_gnumacro
2464 SYSCALL, 0, st_C_gnumacro
2465 ENTRY, 0, st_C_gnumacro
2466 PSEUDO, 0, st_C_gnumacro
2467 # These are defined inside C functions, so currently they are not met.
2468 # EXFUN used in glibc, DEFVAR_* in emacs.
2469 #EXFUN, 0, st_C_gnumacro
2470 #DEFVAR_, 0, st_C_gnumacro
2472 and replace lines between %< and %> with its output, then:
2473 - remove the #if characterset check
2474 - make in_word_set static and not inline. */
2475 /*%<*/
2476 /* C code produced by gperf version 3.0.1 */
2477 /* Command-line: gperf -m 5 */
2478 /* Computed positions: -k'1-2' */
2480 struct C_stab_entry { char *name; int c_ext; enum sym_type type; };
2481 /* maximum key range = 31, duplicates = 0 */
2483 #ifdef __GNUC__
2484 __inline
2485 #else
2486 #ifdef __cplusplus
2487 inline
2488 #endif
2489 #endif
2490 static unsigned int
2491 hash (str, len)
2492 register const char *str;
2493 register unsigned int len;
2495 static unsigned char asso_values[] =
2497 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2498 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2499 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2500 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2501 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2502 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2503 34, 34, 34, 34, 1, 34, 34, 34, 14, 14,
2504 34, 34, 34, 34, 34, 34, 34, 34, 13, 34,
2505 13, 34, 34, 12, 34, 34, 34, 34, 34, 11,
2506 34, 34, 34, 34, 34, 8, 34, 11, 34, 12,
2507 11, 0, 1, 34, 7, 0, 34, 34, 11, 9,
2508 0, 4, 0, 34, 7, 4, 14, 21, 34, 15,
2509 0, 2, 34, 34, 34, 34, 34, 34, 34, 34,
2510 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2511 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2512 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2513 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2514 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2515 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2516 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2517 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2518 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2519 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2520 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2521 34, 34, 34, 34, 34, 34, 34, 34, 34, 34,
2522 34, 34, 34, 34, 34, 34
2524 return len + asso_values[(unsigned char)str[1]] + asso_values[(unsigned char)str[0]];
2527 static struct C_stab_entry *
2528 in_word_set (str, len)
2529 register const char *str;
2530 register unsigned int len;
2532 enum
2534 TOTAL_KEYWORDS = 31,
2535 MIN_WORD_LENGTH = 2,
2536 MAX_WORD_LENGTH = 15,
2537 MIN_HASH_VALUE = 3,
2538 MAX_HASH_VALUE = 33
2541 static struct C_stab_entry wordlist[] =
2543 {""}, {""}, {""},
2544 {"if", 0, st_C_ignore},
2545 {"enum", 0, st_C_enum},
2546 {"@end", 0, st_C_objend},
2547 {"extern", 0, st_C_extern},
2548 {"extends", (C_JAVA & !C_PLPL), st_C_javastruct},
2549 {"for", 0, st_C_ignore},
2550 {"interface", (C_JAVA & !C_PLPL), st_C_struct},
2551 {"@protocol", 0, st_C_objprot},
2552 {"@interface", 0, st_C_objprot},
2553 {"operator", C_PLPL, st_C_operator},
2554 {"return", 0, st_C_ignore},
2555 {"friend", C_PLPL, st_C_ignore},
2556 {"import", (C_JAVA & !C_PLPL), st_C_ignore},
2557 {"@implementation",0, st_C_objimpl},
2558 {"define", 0, st_C_define},
2559 {"package", (C_JAVA & !C_PLPL), st_C_ignore},
2560 {"implements", (C_JAVA & !C_PLPL), st_C_javastruct},
2561 {"namespace", C_PLPL, st_C_struct},
2562 {"domain", C_STAR, st_C_struct},
2563 {"template", 0, st_C_template},
2564 {"typedef", 0, st_C_typedef},
2565 {"struct", 0, st_C_struct},
2566 {"switch", 0, st_C_ignore},
2567 {"union", 0, st_C_struct},
2568 {"while", 0, st_C_ignore},
2569 {"class", 0, st_C_class},
2570 {"__attribute__", 0, st_C_attribute},
2571 {"SYSCALL", 0, st_C_gnumacro},
2572 {"PSEUDO", 0, st_C_gnumacro},
2573 {"ENTRY", 0, st_C_gnumacro},
2574 {"DEFUN", 0, st_C_gnumacro}
2577 if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH)
2579 register int key = hash (str, len);
2581 if (key <= MAX_HASH_VALUE && key >= 0)
2583 register const char *s = wordlist[key].name;
2585 if (*str == *s && !strncmp (str + 1, s + 1, len - 1) && s[len] == '\0')
2586 return &wordlist[key];
2589 return 0;
2591 /*%>*/
2593 static enum sym_type
2594 C_symtype (str, len, c_ext)
2595 char *str;
2596 int len;
2597 int c_ext;
2599 register struct C_stab_entry *se = in_word_set (str, len);
2601 if (se == NULL || (se->c_ext && !(c_ext & se->c_ext)))
2602 return st_none;
2603 return se->type;
2608 * Ignoring __attribute__ ((list))
2610 static bool inattribute; /* looking at an __attribute__ construct */
2613 * C functions and variables are recognized using a simple
2614 * finite automaton. fvdef is its state variable.
2616 static enum
2618 fvnone, /* nothing seen */
2619 fdefunkey, /* Emacs DEFUN keyword seen */
2620 fdefunname, /* Emacs DEFUN name seen */
2621 foperator, /* func: operator keyword seen (cplpl) */
2622 fvnameseen, /* function or variable name seen */
2623 fstartlist, /* func: just after open parenthesis */
2624 finlist, /* func: in parameter list */
2625 flistseen, /* func: after parameter list */
2626 fignore, /* func: before open brace */
2627 vignore /* var-like: ignore until ';' */
2628 } fvdef;
2630 static bool fvextern; /* func or var: extern keyword seen; */
2633 * typedefs are recognized using a simple finite automaton.
2634 * typdef is its state variable.
2636 static enum
2638 tnone, /* nothing seen */
2639 tkeyseen, /* typedef keyword seen */
2640 ttypeseen, /* defined type seen */
2641 tinbody, /* inside typedef body */
2642 tend, /* just before typedef tag */
2643 tignore /* junk after typedef tag */
2644 } typdef;
2647 * struct-like structures (enum, struct and union) are recognized
2648 * using another simple finite automaton. `structdef' is its state
2649 * variable.
2651 static enum
2653 snone, /* nothing seen yet,
2654 or in struct body if bracelev > 0 */
2655 skeyseen, /* struct-like keyword seen */
2656 stagseen, /* struct-like tag seen */
2657 scolonseen /* colon seen after struct-like tag */
2658 } structdef;
2661 * When objdef is different from onone, objtag is the name of the class.
2663 static char *objtag = "<uninited>";
2666 * Yet another little state machine to deal with preprocessor lines.
2668 static enum
2670 dnone, /* nothing seen */
2671 dsharpseen, /* '#' seen as first char on line */
2672 ddefineseen, /* '#' and 'define' seen */
2673 dignorerest /* ignore rest of line */
2674 } definedef;
2677 * State machine for Objective C protocols and implementations.
2678 * Idea by Tom R.Hageman <tom@basil.icce.rug.nl> (1995)
2680 static enum
2682 onone, /* nothing seen */
2683 oprotocol, /* @interface or @protocol seen */
2684 oimplementation, /* @implementations seen */
2685 otagseen, /* class name seen */
2686 oparenseen, /* parenthesis before category seen */
2687 ocatseen, /* category name seen */
2688 oinbody, /* in @implementation body */
2689 omethodsign, /* in @implementation body, after +/- */
2690 omethodtag, /* after method name */
2691 omethodcolon, /* after method colon */
2692 omethodparm, /* after method parameter */
2693 oignore /* wait for @end */
2694 } objdef;
2698 * Use this structure to keep info about the token read, and how it
2699 * should be tagged. Used by the make_C_tag function to build a tag.
2701 static struct tok
2703 char *line; /* string containing the token */
2704 int offset; /* where the token starts in LINE */
2705 int length; /* token length */
2707 The previous members can be used to pass strings around for generic
2708 purposes. The following ones specifically refer to creating tags. In this
2709 case the token contained here is the pattern that will be used to create a
2710 tag.
2712 bool valid; /* do not create a tag; the token should be
2713 invalidated whenever a state machine is
2714 reset prematurely */
2715 bool named; /* create a named tag */
2716 int lineno; /* source line number of tag */
2717 long linepos; /* source char number of tag */
2718 } token; /* latest token read */
2721 * Variables and functions for dealing with nested structures.
2722 * Idea by Mykola Dzyuba <mdzyuba@yahoo.com> (2001)
2724 static void pushclass_above __P((int, char *, int));
2725 static void popclass_above __P((int));
2726 static void write_classname __P((linebuffer *, char *qualifier));
2728 static struct {
2729 char **cname; /* nested class names */
2730 int *bracelev; /* nested class brace level */
2731 int nl; /* class nesting level (elements used) */
2732 int size; /* length of the array */
2733 } cstack; /* stack for nested declaration tags */
2734 /* Current struct nesting depth (namespace, class, struct, union, enum). */
2735 #define nestlev (cstack.nl)
2736 /* After struct keyword or in struct body, not inside a nested function. */
2737 #define instruct (structdef == snone && nestlev > 0 \
2738 && bracelev == cstack.bracelev[nestlev-1] + 1)
2740 static void
2741 pushclass_above (bracelev, str, len)
2742 int bracelev;
2743 char *str;
2744 int len;
2746 int nl;
2748 popclass_above (bracelev);
2749 nl = cstack.nl;
2750 if (nl >= cstack.size)
2752 int size = cstack.size *= 2;
2753 xrnew (cstack.cname, size, char *);
2754 xrnew (cstack.bracelev, size, int);
2756 assert (nl == 0 || cstack.bracelev[nl-1] < bracelev);
2757 cstack.cname[nl] = (str == NULL) ? NULL : savenstr (str, len);
2758 cstack.bracelev[nl] = bracelev;
2759 cstack.nl = nl + 1;
2762 static void
2763 popclass_above (bracelev)
2764 int bracelev;
2766 int nl;
2768 for (nl = cstack.nl - 1;
2769 nl >= 0 && cstack.bracelev[nl] >= bracelev;
2770 nl--)
2772 if (cstack.cname[nl] != NULL)
2773 free (cstack.cname[nl]);
2774 cstack.nl = nl;
2778 static void
2779 write_classname (cn, qualifier)
2780 linebuffer *cn;
2781 char *qualifier;
2783 int i, len;
2784 int qlen = strlen (qualifier);
2786 if (cstack.nl == 0 || cstack.cname[0] == NULL)
2788 len = 0;
2789 cn->len = 0;
2790 cn->buffer[0] = '\0';
2792 else
2794 len = strlen (cstack.cname[0]);
2795 linebuffer_setlen (cn, len);
2796 strcpy (cn->buffer, cstack.cname[0]);
2798 for (i = 1; i < cstack.nl; i++)
2800 char *s;
2801 int slen;
2803 s = cstack.cname[i];
2804 if (s == NULL)
2805 continue;
2806 slen = strlen (s);
2807 len += slen + qlen;
2808 linebuffer_setlen (cn, len);
2809 strncat (cn->buffer, qualifier, qlen);
2810 strncat (cn->buffer, s, slen);
2815 static bool consider_token __P((char *, int, int, int *, int, int, bool *));
2816 static void make_C_tag __P((bool));
2819 * consider_token ()
2820 * checks to see if the current token is at the start of a
2821 * function or variable, or corresponds to a typedef, or
2822 * is a struct/union/enum tag, or #define, or an enum constant.
2824 * *IS_FUNC gets TRUE iff the token is a function or #define macro
2825 * with args. C_EXTP points to which language we are looking at.
2827 * Globals
2828 * fvdef IN OUT
2829 * structdef IN OUT
2830 * definedef IN OUT
2831 * typdef IN OUT
2832 * objdef IN OUT
2835 static bool
2836 consider_token (str, len, c, c_extp, bracelev, parlev, is_func_or_var)
2837 register char *str; /* IN: token pointer */
2838 register int len; /* IN: token length */
2839 register int c; /* IN: first char after the token */
2840 int *c_extp; /* IN, OUT: C extensions mask */
2841 int bracelev; /* IN: brace level */
2842 int parlev; /* IN: parenthesis level */
2843 bool *is_func_or_var; /* OUT: function or variable found */
2845 /* When structdef is stagseen, scolonseen, or snone with bracelev > 0,
2846 structtype is the type of the preceding struct-like keyword, and
2847 structbracelev is the brace level where it has been seen. */
2848 static enum sym_type structtype;
2849 static int structbracelev;
2850 static enum sym_type toktype;
2853 toktype = C_symtype (str, len, *c_extp);
2856 * Skip __attribute__
2858 if (toktype == st_C_attribute)
2860 inattribute = TRUE;
2861 return FALSE;
2865 * Advance the definedef state machine.
2867 switch (definedef)
2869 case dnone:
2870 /* We're not on a preprocessor line. */
2871 if (toktype == st_C_gnumacro)
2873 fvdef = fdefunkey;
2874 return FALSE;
2876 break;
2877 case dsharpseen:
2878 if (toktype == st_C_define)
2880 definedef = ddefineseen;
2882 else
2884 definedef = dignorerest;
2886 return FALSE;
2887 case ddefineseen:
2889 * Make a tag for any macro, unless it is a constant
2890 * and constantypedefs is FALSE.
2892 definedef = dignorerest;
2893 *is_func_or_var = (c == '(');
2894 if (!*is_func_or_var && !constantypedefs)
2895 return FALSE;
2896 else
2897 return TRUE;
2898 case dignorerest:
2899 return FALSE;
2900 default:
2901 error ("internal error: definedef value.", (char *)NULL);
2905 * Now typedefs
2907 switch (typdef)
2909 case tnone:
2910 if (toktype == st_C_typedef)
2912 if (typedefs)
2913 typdef = tkeyseen;
2914 fvextern = FALSE;
2915 fvdef = fvnone;
2916 return FALSE;
2918 break;
2919 case tkeyseen:
2920 switch (toktype)
2922 case st_none:
2923 case st_C_class:
2924 case st_C_struct:
2925 case st_C_enum:
2926 typdef = ttypeseen;
2928 break;
2929 case ttypeseen:
2930 if (structdef == snone && fvdef == fvnone)
2932 fvdef = fvnameseen;
2933 return TRUE;
2935 break;
2936 case tend:
2937 switch (toktype)
2939 case st_C_class:
2940 case st_C_struct:
2941 case st_C_enum:
2942 return FALSE;
2944 return TRUE;
2948 * This structdef business is NOT invoked when we are ctags and the
2949 * file is plain C. This is because a struct tag may have the same
2950 * name as another tag, and this loses with ctags.
2952 switch (toktype)
2954 case st_C_javastruct:
2955 if (structdef == stagseen)
2956 structdef = scolonseen;
2957 return FALSE;
2958 case st_C_template:
2959 case st_C_class:
2960 if ((*c_extp & C_AUTO) /* automatic detection of C++ language */
2961 && bracelev == 0
2962 && definedef == dnone && structdef == snone
2963 && typdef == tnone && fvdef == fvnone)
2964 *c_extp = (*c_extp | C_PLPL) & ~C_AUTO;
2965 if (toktype == st_C_template)
2966 break;
2967 /* FALLTHRU */
2968 case st_C_struct:
2969 case st_C_enum:
2970 if (parlev == 0
2971 && fvdef != vignore
2972 && (typdef == tkeyseen
2973 || (typedefs_or_cplusplus && structdef == snone)))
2975 structdef = skeyseen;
2976 structtype = toktype;
2977 structbracelev = bracelev;
2978 if (fvdef == fvnameseen)
2979 fvdef = fvnone;
2981 return FALSE;
2984 if (structdef == skeyseen)
2986 structdef = stagseen;
2987 return TRUE;
2990 if (typdef != tnone)
2991 definedef = dnone;
2993 /* Detect Objective C constructs. */
2994 switch (objdef)
2996 case onone:
2997 switch (toktype)
2999 case st_C_objprot:
3000 objdef = oprotocol;
3001 return FALSE;
3002 case st_C_objimpl:
3003 objdef = oimplementation;
3004 return FALSE;
3006 break;
3007 case oimplementation:
3008 /* Save the class tag for functions or variables defined inside. */
3009 objtag = savenstr (str, len);
3010 objdef = oinbody;
3011 return FALSE;
3012 case oprotocol:
3013 /* Save the class tag for categories. */
3014 objtag = savenstr (str, len);
3015 objdef = otagseen;
3016 *is_func_or_var = TRUE;
3017 return TRUE;
3018 case oparenseen:
3019 objdef = ocatseen;
3020 *is_func_or_var = TRUE;
3021 return TRUE;
3022 case oinbody:
3023 break;
3024 case omethodsign:
3025 if (parlev == 0)
3027 fvdef = fvnone;
3028 objdef = omethodtag;
3029 linebuffer_setlen (&token_name, len);
3030 strncpy (token_name.buffer, str, len);
3031 token_name.buffer[len] = '\0';
3032 return TRUE;
3034 return FALSE;
3035 case omethodcolon:
3036 if (parlev == 0)
3037 objdef = omethodparm;
3038 return FALSE;
3039 case omethodparm:
3040 if (parlev == 0)
3042 fvdef = fvnone;
3043 objdef = omethodtag;
3044 linebuffer_setlen (&token_name, token_name.len + len);
3045 strncat (token_name.buffer, str, len);
3046 return TRUE;
3048 return FALSE;
3049 case oignore:
3050 if (toktype == st_C_objend)
3052 /* Memory leakage here: the string pointed by objtag is
3053 never released, because many tests would be needed to
3054 avoid breaking on incorrect input code. The amount of
3055 memory leaked here is the sum of the lengths of the
3056 class tags.
3057 free (objtag); */
3058 objdef = onone;
3060 return FALSE;
3063 /* A function, variable or enum constant? */
3064 switch (toktype)
3066 case st_C_extern:
3067 fvextern = TRUE;
3068 switch (fvdef)
3070 case finlist:
3071 case flistseen:
3072 case fignore:
3073 case vignore:
3074 break;
3075 default:
3076 fvdef = fvnone;
3078 return FALSE;
3079 case st_C_ignore:
3080 fvextern = FALSE;
3081 fvdef = vignore;
3082 return FALSE;
3083 case st_C_operator:
3084 fvdef = foperator;
3085 *is_func_or_var = TRUE;
3086 return TRUE;
3087 case st_none:
3088 if (constantypedefs
3089 && structdef == snone
3090 && structtype == st_C_enum && bracelev > structbracelev)
3091 return TRUE; /* enum constant */
3092 switch (fvdef)
3094 case fdefunkey:
3095 if (bracelev > 0)
3096 break;
3097 fvdef = fdefunname; /* GNU macro */
3098 *is_func_or_var = TRUE;
3099 return TRUE;
3100 case fvnone:
3101 switch (typdef)
3103 case ttypeseen:
3104 return FALSE;
3105 case tnone:
3106 if ((strneq (str, "asm", 3) && endtoken (str[3]))
3107 || (strneq (str, "__asm__", 7) && endtoken (str[7])))
3109 fvdef = vignore;
3110 return FALSE;
3112 break;
3114 /* FALLTHRU */
3115 case fvnameseen:
3116 if (len >= 10 && strneq (str+len-10, "::operator", 10))
3118 if (*c_extp & C_AUTO) /* automatic detection of C++ */
3119 *c_extp = (*c_extp | C_PLPL) & ~C_AUTO;
3120 fvdef = foperator;
3121 *is_func_or_var = TRUE;
3122 return TRUE;
3124 if (bracelev > 0 && !instruct)
3125 break;
3126 fvdef = fvnameseen; /* function or variable */
3127 *is_func_or_var = TRUE;
3128 return TRUE;
3130 break;
3133 return FALSE;
3138 * C_entries often keeps pointers to tokens or lines which are older than
3139 * the line currently read. By keeping two line buffers, and switching
3140 * them at end of line, it is possible to use those pointers.
3142 static struct
3144 long linepos;
3145 linebuffer lb;
3146 } lbs[2];
3148 #define current_lb_is_new (newndx == curndx)
3149 #define switch_line_buffers() (curndx = 1 - curndx)
3151 #define curlb (lbs[curndx].lb)
3152 #define newlb (lbs[newndx].lb)
3153 #define curlinepos (lbs[curndx].linepos)
3154 #define newlinepos (lbs[newndx].linepos)
3156 #define plainc ((c_ext & C_EXT) == C_PLAIN)
3157 #define cplpl (c_ext & C_PLPL)
3158 #define cjava ((c_ext & C_JAVA) == C_JAVA)
3160 #define CNL_SAVE_DEFINEDEF() \
3161 do { \
3162 curlinepos = charno; \
3163 readline (&curlb, inf); \
3164 lp = curlb.buffer; \
3165 quotednl = FALSE; \
3166 newndx = curndx; \
3167 } while (0)
3169 #define CNL() \
3170 do { \
3171 CNL_SAVE_DEFINEDEF(); \
3172 if (savetoken.valid) \
3174 token = savetoken; \
3175 savetoken.valid = FALSE; \
3177 definedef = dnone; \
3178 } while (0)
3181 static void
3182 make_C_tag (isfun)
3183 bool isfun;
3185 /* This function should never be called when token.valid is FALSE, but
3186 we must protect against invalid input or internal errors. */
3187 if (!DEBUG && !token.valid)
3188 return;
3190 if (token.valid)
3191 make_tag (token_name.buffer, token_name.len, isfun, token.line,
3192 token.offset+token.length+1, token.lineno, token.linepos);
3193 else /* this case is optimised away if !DEBUG */
3194 make_tag (concat ("INVALID TOKEN:-->", token_name.buffer, ""),
3195 token_name.len + 17, isfun, token.line,
3196 token.offset+token.length+1, token.lineno, token.linepos);
3198 token.valid = FALSE;
3203 * C_entries ()
3204 * This routine finds functions, variables, typedefs,
3205 * #define's, enum constants and struct/union/enum definitions in
3206 * C syntax and adds them to the list.
3208 static void
3209 C_entries (c_ext, inf)
3210 int c_ext; /* extension of C */
3211 FILE *inf; /* input file */
3213 register char c; /* latest char read; '\0' for end of line */
3214 register char *lp; /* pointer one beyond the character `c' */
3215 int curndx, newndx; /* indices for current and new lb */
3216 register int tokoff; /* offset in line of start of current token */
3217 register int toklen; /* length of current token */
3218 char *qualifier; /* string used to qualify names */
3219 int qlen; /* length of qualifier */
3220 int bracelev; /* current brace level */
3221 int bracketlev; /* current bracket level */
3222 int parlev; /* current parenthesis level */
3223 int attrparlev; /* __attribute__ parenthesis level */
3224 int templatelev; /* current template level */
3225 int typdefbracelev; /* bracelev 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 */
3231 linebuffer_init (&lbs[0].lb);
3232 linebuffer_init (&lbs[1].lb);
3233 if (cstack.size == 0)
3235 cstack.size = (DEBUG) ? 1 : 4;
3236 cstack.nl = 0;
3237 cstack.cname = xnew (cstack.size, char *);
3238 cstack.bracelev = xnew (cstack.size, int);
3241 tokoff = toklen = typdefbracelev = 0; /* keep compiler quiet */
3242 curndx = newndx = 0;
3243 lp = curlb.buffer;
3244 *lp = 0;
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 bracelev = bracketlev = parlev = attrparlev = templatelev = 0;
3252 if (cjava)
3253 { qualifier = "."; qlen = 1; }
3254 else
3255 { qualifier = "::"; qlen = 2; }
3258 while (!feof (inf))
3260 c = *lp++;
3261 if (c == '\\')
3263 /* If we are at the end of the line, the next character is a
3264 '\0'; do not skip it, because it is what tells us
3265 to read the next line. */
3266 if (*lp == '\0')
3268 quotednl = TRUE;
3269 continue;
3271 lp++;
3272 c = ' ';
3274 else if (incomm)
3276 switch (c)
3278 case '*':
3279 if (*lp == '/')
3281 c = *lp++;
3282 incomm = FALSE;
3284 break;
3285 case '\0':
3286 /* Newlines inside comments do not end macro definitions in
3287 traditional cpp. */
3288 CNL_SAVE_DEFINEDEF ();
3289 break;
3291 continue;
3293 else if (inquote)
3295 switch (c)
3297 case '"':
3298 inquote = FALSE;
3299 break;
3300 case '\0':
3301 /* Newlines inside strings do not end macro definitions
3302 in traditional cpp, even though compilers don't
3303 usually accept them. */
3304 CNL_SAVE_DEFINEDEF ();
3305 break;
3307 continue;
3309 else if (inchar)
3311 switch (c)
3313 case '\0':
3314 /* Hmmm, something went wrong. */
3315 CNL ();
3316 /* FALLTHRU */
3317 case '\'':
3318 inchar = FALSE;
3319 break;
3321 continue;
3323 else if (bracketlev > 0)
3325 switch (c)
3327 case ']':
3328 if (--bracketlev > 0)
3329 continue;
3330 break;
3331 case '\0':
3332 CNL_SAVE_DEFINEDEF ();
3333 break;
3335 continue;
3337 else switch (c)
3339 case '"':
3340 inquote = TRUE;
3341 if (inattribute)
3342 break;
3343 switch (fvdef)
3345 case fdefunkey:
3346 case fstartlist:
3347 case finlist:
3348 case fignore:
3349 case vignore:
3350 break;
3351 default:
3352 fvextern = FALSE;
3353 fvdef = fvnone;
3355 continue;
3356 case '\'':
3357 inchar = TRUE;
3358 if (inattribute)
3359 break;
3360 if (fvdef != finlist && fvdef != fignore && fvdef !=vignore)
3362 fvextern = FALSE;
3363 fvdef = fvnone;
3365 continue;
3366 case '/':
3367 if (*lp == '*')
3369 lp++;
3370 incomm = TRUE;
3371 continue;
3373 else if (/* cplpl && */ *lp == '/')
3375 c = '\0';
3376 break;
3378 else
3379 break;
3380 case '%':
3381 if ((c_ext & YACC) && *lp == '%')
3383 /* Entering or exiting rules section in yacc file. */
3384 lp++;
3385 definedef = dnone; fvdef = fvnone; fvextern = FALSE;
3386 typdef = tnone; structdef = snone;
3387 midtoken = inquote = inchar = incomm = quotednl = FALSE;
3388 bracelev = 0;
3389 yacc_rules = !yacc_rules;
3390 continue;
3392 else
3393 break;
3394 case '#':
3395 if (definedef == dnone)
3397 char *cp;
3398 bool cpptoken = TRUE;
3400 /* Look back on this line. If all blanks, or nonblanks
3401 followed by an end of comment, this is a preprocessor
3402 token. */
3403 for (cp = newlb.buffer; cp < lp-1; cp++)
3404 if (!iswhite (*cp))
3406 if (*cp == '*' && *(cp+1) == '/')
3408 cp++;
3409 cpptoken = TRUE;
3411 else
3412 cpptoken = FALSE;
3414 if (cpptoken)
3415 definedef = dsharpseen;
3416 } /* if (definedef == dnone) */
3417 continue;
3418 case '[':
3419 bracketlev++;
3420 continue;
3421 } /* switch (c) */
3424 /* Consider token only if some involved conditions are satisfied. */
3425 if (typdef != tignore
3426 && definedef != dignorerest
3427 && fvdef != finlist
3428 && templatelev == 0
3429 && (definedef != dnone
3430 || structdef != scolonseen)
3431 && !inattribute)
3433 if (midtoken)
3435 if (endtoken (c))
3437 if (c == ':' && *lp == ':' && begtoken (lp[1]))
3438 /* This handles :: in the middle,
3439 but not at the beginning of an identifier.
3440 Also, space-separated :: is not recognised. */
3442 if (c_ext & C_AUTO) /* automatic detection of C++ */
3443 c_ext = (c_ext | C_PLPL) & ~C_AUTO;
3444 lp += 2;
3445 toklen += 2;
3446 c = lp[-1];
3447 goto still_in_token;
3449 else
3451 bool funorvar = FALSE;
3453 if (yacc_rules
3454 || consider_token (newlb.buffer + tokoff, toklen, c,
3455 &c_ext, bracelev, parlev,
3456 &funorvar))
3458 if (fvdef == foperator)
3460 char *oldlp = lp;
3461 lp = skip_spaces (lp-1);
3462 if (*lp != '\0')
3463 lp += 1;
3464 while (*lp != '\0'
3465 && !iswhite (*lp) && *lp != '(')
3466 lp += 1;
3467 c = *lp++;
3468 toklen += lp - oldlp;
3470 token.named = FALSE;
3471 if (!plainc
3472 && nestlev > 0 && definedef == dnone)
3473 /* in struct body */
3475 write_classname (&token_name, qualifier);
3476 linebuffer_setlen (&token_name,
3477 token_name.len+qlen+toklen);
3478 strcat (token_name.buffer, qualifier);
3479 strncat (token_name.buffer,
3480 newlb.buffer + tokoff, toklen);
3481 token.named = TRUE;
3483 else if (objdef == ocatseen)
3484 /* Objective C category */
3486 int len = strlen (objtag) + 2 + toklen;
3487 linebuffer_setlen (&token_name, len);
3488 strcpy (token_name.buffer, objtag);
3489 strcat (token_name.buffer, "(");
3490 strncat (token_name.buffer,
3491 newlb.buffer + tokoff, toklen);
3492 strcat (token_name.buffer, ")");
3493 token.named = TRUE;
3495 else if (objdef == omethodtag
3496 || objdef == omethodparm)
3497 /* Objective C method */
3499 token.named = TRUE;
3501 else if (fvdef == fdefunname)
3502 /* GNU DEFUN and similar macros */
3504 bool defun = (newlb.buffer[tokoff] == 'F');
3505 int off = tokoff;
3506 int len = toklen;
3508 /* Rewrite the tag so that emacs lisp DEFUNs
3509 can be found by their elisp name */
3510 if (defun)
3512 off += 1;
3513 len -= 1;
3515 len = toklen;
3516 linebuffer_setlen (&token_name, len);
3517 strncpy (token_name.buffer,
3518 newlb.buffer + off, len);
3519 token_name.buffer[len] = '\0';
3520 if (defun)
3521 while (--len >= 0)
3522 if (token_name.buffer[len] == '_')
3523 token_name.buffer[len] = '-';
3524 token.named = defun;
3526 else
3528 linebuffer_setlen (&token_name, toklen);
3529 strncpy (token_name.buffer,
3530 newlb.buffer + tokoff, toklen);
3531 token_name.buffer[toklen] = '\0';
3532 /* Name macros and members. */
3533 token.named = (structdef == stagseen
3534 || typdef == ttypeseen
3535 || typdef == tend
3536 || (funorvar
3537 && definedef == dignorerest)
3538 || (funorvar
3539 && definedef == dnone
3540 && structdef == snone
3541 && bracelev > 0));
3543 token.lineno = lineno;
3544 token.offset = tokoff;
3545 token.length = toklen;
3546 token.line = newlb.buffer;
3547 token.linepos = newlinepos;
3548 token.valid = TRUE;
3550 if (definedef == dnone
3551 && (fvdef == fvnameseen
3552 || fvdef == foperator
3553 || structdef == stagseen
3554 || typdef == tend
3555 || typdef == ttypeseen
3556 || objdef != onone))
3558 if (current_lb_is_new)
3559 switch_line_buffers ();
3561 else if (definedef != dnone
3562 || fvdef == fdefunname
3563 || instruct)
3564 make_C_tag (funorvar);
3566 else /* not yacc and consider_token failed */
3568 if (inattribute && fvdef == fignore)
3570 /* We have just met __attribute__ after a
3571 function parameter list: do not tag the
3572 function again. */
3573 fvdef = fvnone;
3576 midtoken = FALSE;
3578 } /* if (endtoken (c)) */
3579 else if (intoken (c))
3580 still_in_token:
3582 toklen++;
3583 continue;
3585 } /* if (midtoken) */
3586 else if (begtoken (c))
3588 switch (definedef)
3590 case dnone:
3591 switch (fvdef)
3593 case fstartlist:
3594 /* This prevents tagging fb in
3595 void (__attribute__((noreturn)) *fb) (void);
3596 Fixing this is not easy and not very important. */
3597 fvdef = finlist;
3598 continue;
3599 case flistseen:
3600 if (plainc || declarations)
3602 make_C_tag (TRUE); /* a function */
3603 fvdef = fignore;
3605 break;
3607 if (structdef == stagseen && !cjava)
3609 popclass_above (bracelev);
3610 structdef = snone;
3612 break;
3613 case dsharpseen:
3614 savetoken = token;
3615 break;
3617 if (!yacc_rules || lp == newlb.buffer + 1)
3619 tokoff = lp - 1 - newlb.buffer;
3620 toklen = 1;
3621 midtoken = TRUE;
3623 continue;
3624 } /* if (begtoken) */
3625 } /* if must look at token */
3628 /* Detect end of line, colon, comma, semicolon and various braces
3629 after having handled a token.*/
3630 switch (c)
3632 case ':':
3633 if (inattribute)
3634 break;
3635 if (yacc_rules && token.offset == 0 && token.valid)
3637 make_C_tag (FALSE); /* a yacc function */
3638 break;
3640 if (definedef != dnone)
3641 break;
3642 switch (objdef)
3644 case otagseen:
3645 objdef = oignore;
3646 make_C_tag (TRUE); /* an Objective C class */
3647 break;
3648 case omethodtag:
3649 case omethodparm:
3650 objdef = omethodcolon;
3651 linebuffer_setlen (&token_name, token_name.len + 1);
3652 strcat (token_name.buffer, ":");
3653 break;
3655 if (structdef == stagseen)
3657 structdef = scolonseen;
3658 break;
3660 /* Should be useless, but may be work as a safety net. */
3661 if (cplpl && fvdef == flistseen)
3663 make_C_tag (TRUE); /* a function */
3664 fvdef = fignore;
3665 break;
3667 break;
3668 case ';':
3669 if (definedef != dnone || inattribute)
3670 break;
3671 switch (typdef)
3673 case tend:
3674 case ttypeseen:
3675 make_C_tag (FALSE); /* a typedef */
3676 typdef = tnone;
3677 fvdef = fvnone;
3678 break;
3679 case tnone:
3680 case tinbody:
3681 case tignore:
3682 switch (fvdef)
3684 case fignore:
3685 if (typdef == tignore || cplpl)
3686 fvdef = fvnone;
3687 break;
3688 case fvnameseen:
3689 if ((globals && bracelev == 0 && (!fvextern || declarations))
3690 || (members && instruct))
3691 make_C_tag (FALSE); /* a variable */
3692 fvextern = FALSE;
3693 fvdef = fvnone;
3694 token.valid = FALSE;
3695 break;
3696 case flistseen:
3697 if ((declarations
3698 && (cplpl || !instruct)
3699 && (typdef == tnone || (typdef != tignore && instruct)))
3700 || (members
3701 && plainc && instruct))
3702 make_C_tag (TRUE); /* a function */
3703 /* FALLTHRU */
3704 default:
3705 fvextern = FALSE;
3706 fvdef = fvnone;
3707 if (declarations
3708 && cplpl && structdef == stagseen)
3709 make_C_tag (FALSE); /* forward declaration */
3710 else
3711 token.valid = FALSE;
3712 } /* switch (fvdef) */
3713 /* FALLTHRU */
3714 default:
3715 if (!instruct)
3716 typdef = tnone;
3718 if (structdef == stagseen)
3719 structdef = snone;
3720 break;
3721 case ',':
3722 if (definedef != dnone || inattribute)
3723 break;
3724 switch (objdef)
3726 case omethodtag:
3727 case omethodparm:
3728 make_C_tag (TRUE); /* an Objective C method */
3729 objdef = oinbody;
3730 break;
3732 switch (fvdef)
3734 case fdefunkey:
3735 case foperator:
3736 case fstartlist:
3737 case finlist:
3738 case fignore:
3739 case vignore:
3740 break;
3741 case fdefunname:
3742 fvdef = fignore;
3743 break;
3744 case fvnameseen:
3745 if (parlev == 0
3746 && ((globals
3747 && bracelev == 0
3748 && templatelev == 0
3749 && (!fvextern || declarations))
3750 || (members && instruct)))
3751 make_C_tag (FALSE); /* a variable */
3752 break;
3753 case flistseen:
3754 if ((declarations && typdef == tnone && !instruct)
3755 || (members && typdef != tignore && instruct))
3757 make_C_tag (TRUE); /* a function */
3758 fvdef = fvnameseen;
3760 else if (!declarations)
3761 fvdef = fvnone;
3762 token.valid = FALSE;
3763 break;
3764 default:
3765 fvdef = fvnone;
3767 if (structdef == stagseen)
3768 structdef = snone;
3769 break;
3770 case ']':
3771 if (definedef != dnone || inattribute)
3772 break;
3773 if (structdef == stagseen)
3774 structdef = snone;
3775 switch (typdef)
3777 case ttypeseen:
3778 case tend:
3779 typdef = tignore;
3780 make_C_tag (FALSE); /* a typedef */
3781 break;
3782 case tnone:
3783 case tinbody:
3784 switch (fvdef)
3786 case foperator:
3787 case finlist:
3788 case fignore:
3789 case vignore:
3790 break;
3791 case fvnameseen:
3792 if ((members && bracelev == 1)
3793 || (globals && bracelev == 0
3794 && (!fvextern || declarations)))
3795 make_C_tag (FALSE); /* a variable */
3796 /* FALLTHRU */
3797 default:
3798 fvdef = fvnone;
3800 break;
3802 break;
3803 case '(':
3804 if (inattribute)
3806 attrparlev++;
3807 break;
3809 if (definedef != dnone)
3810 break;
3811 if (objdef == otagseen && parlev == 0)
3812 objdef = oparenseen;
3813 switch (fvdef)
3815 case fvnameseen:
3816 if (typdef == ttypeseen
3817 && *lp != '*'
3818 && !instruct)
3820 /* This handles constructs like:
3821 typedef void OperatorFun (int fun); */
3822 make_C_tag (FALSE);
3823 typdef = tignore;
3824 fvdef = fignore;
3825 break;
3827 /* FALLTHRU */
3828 case foperator:
3829 fvdef = fstartlist;
3830 break;
3831 case flistseen:
3832 fvdef = finlist;
3833 break;
3835 parlev++;
3836 break;
3837 case ')':
3838 if (inattribute)
3840 if (--attrparlev == 0)
3841 inattribute = FALSE;
3842 break;
3844 if (definedef != dnone)
3845 break;
3846 if (objdef == ocatseen && parlev == 1)
3848 make_C_tag (TRUE); /* an Objective C category */
3849 objdef = oignore;
3851 if (--parlev == 0)
3853 switch (fvdef)
3855 case fstartlist:
3856 case finlist:
3857 fvdef = flistseen;
3858 break;
3860 if (!instruct
3861 && (typdef == tend
3862 || typdef == ttypeseen))
3864 typdef = tignore;
3865 make_C_tag (FALSE); /* a typedef */
3868 else if (parlev < 0) /* can happen due to ill-conceived #if's. */
3869 parlev = 0;
3870 break;
3871 case '{':
3872 if (definedef != dnone)
3873 break;
3874 if (typdef == ttypeseen)
3876 /* Whenever typdef is set to tinbody (currently only
3877 here), typdefbracelev should be set to bracelev. */
3878 typdef = tinbody;
3879 typdefbracelev = bracelev;
3881 switch (fvdef)
3883 case flistseen:
3884 make_C_tag (TRUE); /* a function */
3885 /* FALLTHRU */
3886 case fignore:
3887 fvdef = fvnone;
3888 break;
3889 case fvnone:
3890 switch (objdef)
3892 case otagseen:
3893 make_C_tag (TRUE); /* an Objective C class */
3894 objdef = oignore;
3895 break;
3896 case omethodtag:
3897 case omethodparm:
3898 make_C_tag (TRUE); /* an Objective C method */
3899 objdef = oinbody;
3900 break;
3901 default:
3902 /* Neutralize `extern "C" {' grot. */
3903 if (bracelev == 0 && structdef == snone && nestlev == 0
3904 && typdef == tnone)
3905 bracelev = -1;
3907 break;
3909 switch (structdef)
3911 case skeyseen: /* unnamed struct */
3912 pushclass_above (bracelev, NULL, 0);
3913 structdef = snone;
3914 break;
3915 case stagseen: /* named struct or enum */
3916 case scolonseen: /* a class */
3917 pushclass_above (bracelev,token.line+token.offset, token.length);
3918 structdef = snone;
3919 make_C_tag (FALSE); /* a struct or enum */
3920 break;
3922 bracelev++;
3923 break;
3924 case '*':
3925 if (definedef != dnone)
3926 break;
3927 if (fvdef == fstartlist)
3929 fvdef = fvnone; /* avoid tagging `foo' in `foo (*bar()) ()' */
3930 token.valid = FALSE;
3932 break;
3933 case '}':
3934 if (definedef != dnone)
3935 break;
3936 if (!ignoreindent && lp == newlb.buffer + 1)
3938 if (bracelev != 0)
3939 token.valid = FALSE;
3940 bracelev = 0; /* reset brace level if first column */
3941 parlev = 0; /* also reset paren level, just in case... */
3943 else if (bracelev > 0)
3944 bracelev--;
3945 else
3946 token.valid = FALSE; /* something gone amiss, token unreliable */
3947 popclass_above (bracelev);
3948 structdef = snone;
3949 /* Only if typdef == tinbody is typdefbracelev significant. */
3950 if (typdef == tinbody && bracelev <= typdefbracelev)
3952 assert (bracelev == typdefbracelev);
3953 typdef = tend;
3955 break;
3956 case '=':
3957 if (definedef != dnone)
3958 break;
3959 switch (fvdef)
3961 case foperator:
3962 case finlist:
3963 case fignore:
3964 case vignore:
3965 break;
3966 case fvnameseen:
3967 if ((members && bracelev == 1)
3968 || (globals && bracelev == 0 && (!fvextern || declarations)))
3969 make_C_tag (FALSE); /* a variable */
3970 /* FALLTHRU */
3971 default:
3972 fvdef = vignore;
3974 break;
3975 case '<':
3976 if (cplpl
3977 && (structdef == stagseen || fvdef == fvnameseen))
3979 templatelev++;
3980 break;
3982 goto resetfvdef;
3983 case '>':
3984 if (templatelev > 0)
3986 templatelev--;
3987 break;
3989 goto resetfvdef;
3990 case '+':
3991 case '-':
3992 if (objdef == oinbody && bracelev == 0)
3994 objdef = omethodsign;
3995 break;
3997 /* FALLTHRU */
3998 resetfvdef:
3999 case '#': case '~': case '&': case '%': case '/':
4000 case '|': case '^': case '!': case '.': case '?':
4001 if (definedef != dnone)
4002 break;
4003 /* These surely cannot follow a function tag in C. */
4004 switch (fvdef)
4006 case foperator:
4007 case finlist:
4008 case fignore:
4009 case vignore:
4010 break;
4011 default:
4012 fvdef = fvnone;
4014 break;
4015 case '\0':
4016 if (objdef == otagseen)
4018 make_C_tag (TRUE); /* an Objective C class */
4019 objdef = oignore;
4021 /* If a macro spans multiple lines don't reset its state. */
4022 if (quotednl)
4023 CNL_SAVE_DEFINEDEF ();
4024 else
4025 CNL ();
4026 break;
4027 } /* switch (c) */
4029 } /* while not eof */
4031 free (lbs[0].lb.buffer);
4032 free (lbs[1].lb.buffer);
4036 * Process either a C++ file or a C file depending on the setting
4037 * of a global flag.
4039 static void
4040 default_C_entries (inf)
4041 FILE *inf;
4043 C_entries (cplusplus ? C_PLPL : C_AUTO, inf);
4046 /* Always do plain C. */
4047 static void
4048 plain_C_entries (inf)
4049 FILE *inf;
4051 C_entries (0, inf);
4054 /* Always do C++. */
4055 static void
4056 Cplusplus_entries (inf)
4057 FILE *inf;
4059 C_entries (C_PLPL, inf);
4062 /* Always do Java. */
4063 static void
4064 Cjava_entries (inf)
4065 FILE *inf;
4067 C_entries (C_JAVA, inf);
4070 /* Always do C*. */
4071 static void
4072 Cstar_entries (inf)
4073 FILE *inf;
4075 C_entries (C_STAR, inf);
4078 /* Always do Yacc. */
4079 static void
4080 Yacc_entries (inf)
4081 FILE *inf;
4083 C_entries (YACC, inf);
4087 /* Useful macros. */
4088 #define LOOP_ON_INPUT_LINES(file_pointer, line_buffer, char_pointer) \
4089 for (; /* loop initialization */ \
4090 !feof (file_pointer) /* loop test */ \
4091 && /* instructions at start of loop */ \
4092 (readline (&line_buffer, file_pointer), \
4093 char_pointer = line_buffer.buffer, \
4094 TRUE); \
4097 #define LOOKING_AT(cp, kw) /* kw is the keyword, a literal string */ \
4098 ((assert("" kw), TRUE) /* syntax error if not a literal string */ \
4099 && strneq ((cp), kw, sizeof(kw)-1) /* cp points at kw */ \
4100 && notinname ((cp)[sizeof(kw)-1]) /* end of kw */ \
4101 && ((cp) = skip_spaces((cp)+sizeof(kw)-1))) /* skip spaces */
4103 /* Similar to LOOKING_AT but does not use notinname, does not skip */
4104 #define LOOKING_AT_NOCASE(cp, kw) /* the keyword is a literal string */ \
4105 ((assert("" kw), TRUE) /* syntax error if not a literal string */ \
4106 && strncaseeq ((cp), kw, sizeof(kw)-1) /* cp points at kw */ \
4107 && ((cp) += sizeof(kw)-1)) /* skip spaces */
4110 * Read a file, but do no processing. This is used to do regexp
4111 * matching on files that have no language defined.
4113 static void
4114 just_read_file (inf)
4115 FILE *inf;
4117 register char *dummy;
4119 LOOP_ON_INPUT_LINES (inf, lb, dummy)
4120 continue;
4124 /* Fortran parsing */
4126 static void F_takeprec __P((void));
4127 static void F_getit __P((FILE *));
4129 static void
4130 F_takeprec ()
4132 dbp = skip_spaces (dbp);
4133 if (*dbp != '*')
4134 return;
4135 dbp++;
4136 dbp = skip_spaces (dbp);
4137 if (strneq (dbp, "(*)", 3))
4139 dbp += 3;
4140 return;
4142 if (!ISDIGIT (*dbp))
4144 --dbp; /* force failure */
4145 return;
4148 dbp++;
4149 while (ISDIGIT (*dbp));
4152 static void
4153 F_getit (inf)
4154 FILE *inf;
4156 register char *cp;
4158 dbp = skip_spaces (dbp);
4159 if (*dbp == '\0')
4161 readline (&lb, inf);
4162 dbp = lb.buffer;
4163 if (dbp[5] != '&')
4164 return;
4165 dbp += 6;
4166 dbp = skip_spaces (dbp);
4168 if (!ISALPHA (*dbp) && *dbp != '_' && *dbp != '$')
4169 return;
4170 for (cp = dbp + 1; *cp != '\0' && intoken (*cp); cp++)
4171 continue;
4172 make_tag (dbp, cp-dbp, TRUE,
4173 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4177 static void
4178 Fortran_functions (inf)
4179 FILE *inf;
4181 LOOP_ON_INPUT_LINES (inf, lb, dbp)
4183 if (*dbp == '%')
4184 dbp++; /* Ratfor escape to fortran */
4185 dbp = skip_spaces (dbp);
4186 if (*dbp == '\0')
4187 continue;
4188 switch (lowcase (*dbp))
4190 case 'i':
4191 if (nocase_tail ("integer"))
4192 F_takeprec ();
4193 break;
4194 case 'r':
4195 if (nocase_tail ("real"))
4196 F_takeprec ();
4197 break;
4198 case 'l':
4199 if (nocase_tail ("logical"))
4200 F_takeprec ();
4201 break;
4202 case 'c':
4203 if (nocase_tail ("complex") || nocase_tail ("character"))
4204 F_takeprec ();
4205 break;
4206 case 'd':
4207 if (nocase_tail ("double"))
4209 dbp = skip_spaces (dbp);
4210 if (*dbp == '\0')
4211 continue;
4212 if (nocase_tail ("precision"))
4213 break;
4214 continue;
4216 break;
4218 dbp = skip_spaces (dbp);
4219 if (*dbp == '\0')
4220 continue;
4221 switch (lowcase (*dbp))
4223 case 'f':
4224 if (nocase_tail ("function"))
4225 F_getit (inf);
4226 continue;
4227 case 's':
4228 if (nocase_tail ("subroutine"))
4229 F_getit (inf);
4230 continue;
4231 case 'e':
4232 if (nocase_tail ("entry"))
4233 F_getit (inf);
4234 continue;
4235 case 'b':
4236 if (nocase_tail ("blockdata") || nocase_tail ("block data"))
4238 dbp = skip_spaces (dbp);
4239 if (*dbp == '\0') /* assume un-named */
4240 make_tag ("blockdata", 9, TRUE,
4241 lb.buffer, dbp - lb.buffer, lineno, linecharno);
4242 else
4243 F_getit (inf); /* look for name */
4245 continue;
4252 * Ada parsing
4253 * Original code by
4254 * Philippe Waroquiers (1998)
4257 static void Ada_getit __P((FILE *, char *));
4259 /* Once we are positioned after an "interesting" keyword, let's get
4260 the real tag value necessary. */
4261 static void
4262 Ada_getit (inf, name_qualifier)
4263 FILE *inf;
4264 char *name_qualifier;
4266 register char *cp;
4267 char *name;
4268 char c;
4270 while (!feof (inf))
4272 dbp = skip_spaces (dbp);
4273 if (*dbp == '\0'
4274 || (dbp[0] == '-' && dbp[1] == '-'))
4276 readline (&lb, inf);
4277 dbp = lb.buffer;
4279 switch (lowcase(*dbp))
4281 case 'b':
4282 if (nocase_tail ("body"))
4284 /* Skipping body of procedure body or package body or ....
4285 resetting qualifier to body instead of spec. */
4286 name_qualifier = "/b";
4287 continue;
4289 break;
4290 case 't':
4291 /* Skipping type of task type or protected type ... */
4292 if (nocase_tail ("type"))
4293 continue;
4294 break;
4296 if (*dbp == '"')
4298 dbp += 1;
4299 for (cp = dbp; *cp != '\0' && *cp != '"'; cp++)
4300 continue;
4302 else
4304 dbp = skip_spaces (dbp);
4305 for (cp = dbp;
4306 (*cp != '\0'
4307 && (ISALPHA (*cp) || ISDIGIT (*cp) || *cp == '_' || *cp == '.'));
4308 cp++)
4309 continue;
4310 if (cp == dbp)
4311 return;
4313 c = *cp;
4314 *cp = '\0';
4315 name = concat (dbp, name_qualifier, "");
4316 *cp = c;
4317 make_tag (name, strlen (name), TRUE,
4318 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4319 free (name);
4320 if (c == '"')
4321 dbp = cp + 1;
4322 return;
4326 static void
4327 Ada_funcs (inf)
4328 FILE *inf;
4330 bool inquote = FALSE;
4331 bool skip_till_semicolumn = FALSE;
4333 LOOP_ON_INPUT_LINES (inf, lb, dbp)
4335 while (*dbp != '\0')
4337 /* Skip a string i.e. "abcd". */
4338 if (inquote || (*dbp == '"'))
4340 dbp = etags_strchr ((inquote) ? dbp : dbp+1, '"');
4341 if (dbp != NULL)
4343 inquote = FALSE;
4344 dbp += 1;
4345 continue; /* advance char */
4347 else
4349 inquote = TRUE;
4350 break; /* advance line */
4354 /* Skip comments. */
4355 if (dbp[0] == '-' && dbp[1] == '-')
4356 break; /* advance line */
4358 /* Skip character enclosed in single quote i.e. 'a'
4359 and skip single quote starting an attribute i.e. 'Image. */
4360 if (*dbp == '\'')
4362 dbp++ ;
4363 if (*dbp != '\0')
4364 dbp++;
4365 continue;
4368 if (skip_till_semicolumn)
4370 if (*dbp == ';')
4371 skip_till_semicolumn = FALSE;
4372 dbp++;
4373 continue; /* advance char */
4376 /* Search for beginning of a token. */
4377 if (!begtoken (*dbp))
4379 dbp++;
4380 continue; /* advance char */
4383 /* We are at the beginning of a token. */
4384 switch (lowcase(*dbp))
4386 case 'f':
4387 if (!packages_only && nocase_tail ("function"))
4388 Ada_getit (inf, "/f");
4389 else
4390 break; /* from switch */
4391 continue; /* advance char */
4392 case 'p':
4393 if (!packages_only && nocase_tail ("procedure"))
4394 Ada_getit (inf, "/p");
4395 else if (nocase_tail ("package"))
4396 Ada_getit (inf, "/s");
4397 else if (nocase_tail ("protected")) /* protected type */
4398 Ada_getit (inf, "/t");
4399 else
4400 break; /* from switch */
4401 continue; /* advance char */
4403 case 'u':
4404 if (typedefs && !packages_only && nocase_tail ("use"))
4406 /* when tagging types, avoid tagging use type Pack.Typename;
4407 for this, we will skip everything till a ; */
4408 skip_till_semicolumn = TRUE;
4409 continue; /* advance char */
4412 case 't':
4413 if (!packages_only && nocase_tail ("task"))
4414 Ada_getit (inf, "/k");
4415 else if (typedefs && !packages_only && nocase_tail ("type"))
4417 Ada_getit (inf, "/t");
4418 while (*dbp != '\0')
4419 dbp += 1;
4421 else
4422 break; /* from switch */
4423 continue; /* advance char */
4426 /* Look for the end of the token. */
4427 while (!endtoken (*dbp))
4428 dbp++;
4430 } /* advance char */
4431 } /* advance line */
4436 * Unix and microcontroller assembly tag handling
4437 * Labels: /^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]/
4438 * Idea by Bob Weiner, Motorola Inc. (1994)
4440 static void
4441 Asm_labels (inf)
4442 FILE *inf;
4444 register char *cp;
4446 LOOP_ON_INPUT_LINES (inf, lb, cp)
4448 /* If first char is alphabetic or one of [_.$], test for colon
4449 following identifier. */
4450 if (ISALPHA (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
4452 /* Read past label. */
4453 cp++;
4454 while (ISALNUM (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
4455 cp++;
4456 if (*cp == ':' || iswhite (*cp))
4457 /* Found end of label, so copy it and add it to the table. */
4458 make_tag (lb.buffer, cp - lb.buffer, TRUE,
4459 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4466 * Perl support
4467 * Perl sub names: /^sub[ \t\n]+[^ \t\n{]+/
4468 * Perl variable names: /^(my|local).../
4469 * Original code by Bart Robinson <lomew@cs.utah.edu> (1995)
4470 * Additions by Michael Ernst <mernst@alum.mit.edu> (1997)
4471 * Ideas by Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> (2001)
4473 static void
4474 Perl_functions (inf)
4475 FILE *inf;
4477 char *package = savestr ("main"); /* current package name */
4478 register char *cp;
4480 LOOP_ON_INPUT_LINES (inf, lb, cp)
4482 skip_spaces(cp);
4484 if (LOOKING_AT (cp, "package"))
4486 free (package);
4487 get_tag (cp, &package);
4489 else if (LOOKING_AT (cp, "sub"))
4491 char *pos;
4492 char *sp = cp;
4494 while (!notinname (*cp))
4495 cp++;
4496 if (cp == sp)
4497 continue; /* nothing found */
4498 if ((pos = etags_strchr (sp, ':')) != NULL
4499 && pos < cp && pos[1] == ':')
4500 /* The name is already qualified. */
4501 make_tag (sp, cp - sp, TRUE,
4502 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4503 else
4504 /* Qualify it. */
4506 char savechar, *name;
4508 savechar = *cp;
4509 *cp = '\0';
4510 name = concat (package, "::", sp);
4511 *cp = savechar;
4512 make_tag (name, strlen(name), TRUE,
4513 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4514 free (name);
4517 else if (globals) /* only if we are tagging global vars */
4519 /* Skip a qualifier, if any. */
4520 bool qual = LOOKING_AT (cp, "my") || LOOKING_AT (cp, "local");
4521 /* After "my" or "local", but before any following paren or space. */
4522 char *varstart = cp;
4524 if (qual /* should this be removed? If yes, how? */
4525 && (*cp == '$' || *cp == '@' || *cp == '%'))
4527 varstart += 1;
4529 cp++;
4530 while (ISALNUM (*cp) || *cp == '_');
4532 else if (qual)
4534 /* Should be examining a variable list at this point;
4535 could insist on seeing an open parenthesis. */
4536 while (*cp != '\0' && *cp != ';' && *cp != '=' && *cp != ')')
4537 cp++;
4539 else
4540 continue;
4542 make_tag (varstart, cp - varstart, FALSE,
4543 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4550 * Python support
4551 * Look for /^[\t]*def[ \t\n]+[^ \t\n(:]+/ or /^class[ \t\n]+[^ \t\n(:]+/
4552 * Idea by Eric S. Raymond <esr@thyrsus.com> (1997)
4553 * More ideas by seb bacon <seb@jamkit.com> (2002)
4555 static void
4556 Python_functions (inf)
4557 FILE *inf;
4559 register char *cp;
4561 LOOP_ON_INPUT_LINES (inf, lb, cp)
4563 cp = skip_spaces (cp);
4564 if (LOOKING_AT (cp, "def") || LOOKING_AT (cp, "class"))
4566 char *name = cp;
4567 while (!notinname (*cp) && *cp != ':')
4568 cp++;
4569 make_tag (name, cp - name, TRUE,
4570 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4577 * PHP support
4578 * Look for:
4579 * - /^[ \t]*function[ \t\n]+[^ \t\n(]+/
4580 * - /^[ \t]*class[ \t\n]+[^ \t\n]+/
4581 * - /^[ \t]*define\(\"[^\"]+/
4582 * Only with --members:
4583 * - /^[ \t]*var[ \t\n]+\$[^ \t\n=;]/
4584 * Idea by Diez B. Roggisch (2001)
4586 static void
4587 PHP_functions (inf)
4588 FILE *inf;
4590 register char *cp, *name;
4591 bool search_identifier = FALSE;
4593 LOOP_ON_INPUT_LINES (inf, lb, cp)
4595 cp = skip_spaces (cp);
4596 name = cp;
4597 if (search_identifier
4598 && *cp != '\0')
4600 while (!notinname (*cp))
4601 cp++;
4602 make_tag (name, cp - name, TRUE,
4603 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4604 search_identifier = FALSE;
4606 else if (LOOKING_AT (cp, "function"))
4608 if(*cp == '&')
4609 cp = skip_spaces (cp+1);
4610 if(*cp != '\0')
4612 name = cp;
4613 while (!notinname (*cp))
4614 cp++;
4615 make_tag (name, cp - name, TRUE,
4616 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4618 else
4619 search_identifier = TRUE;
4621 else if (LOOKING_AT (cp, "class"))
4623 if (*cp != '\0')
4625 name = cp;
4626 while (*cp != '\0' && !iswhite (*cp))
4627 cp++;
4628 make_tag (name, cp - name, FALSE,
4629 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4631 else
4632 search_identifier = TRUE;
4634 else if (strneq (cp, "define", 6)
4635 && (cp = skip_spaces (cp+6))
4636 && *cp++ == '('
4637 && (*cp == '"' || *cp == '\''))
4639 char quote = *cp++;
4640 name = cp;
4641 while (*cp != quote && *cp != '\0')
4642 cp++;
4643 make_tag (name, cp - name, FALSE,
4644 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4646 else if (members
4647 && LOOKING_AT (cp, "var")
4648 && *cp == '$')
4650 name = cp;
4651 while (!notinname(*cp))
4652 cp++;
4653 make_tag (name, cp - name, FALSE,
4654 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4661 * Cobol tag functions
4662 * We could look for anything that could be a paragraph name.
4663 * i.e. anything that starts in column 8 is one word and ends in a full stop.
4664 * Idea by Corny de Souza (1993)
4666 static void
4667 Cobol_paragraphs (inf)
4668 FILE *inf;
4670 register char *bp, *ep;
4672 LOOP_ON_INPUT_LINES (inf, lb, bp)
4674 if (lb.len < 9)
4675 continue;
4676 bp += 8;
4678 /* If eoln, compiler option or comment ignore whole line. */
4679 if (bp[-1] != ' ' || !ISALNUM (bp[0]))
4680 continue;
4682 for (ep = bp; ISALNUM (*ep) || *ep == '-'; ep++)
4683 continue;
4684 if (*ep++ == '.')
4685 make_tag (bp, ep - bp, TRUE,
4686 lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
4692 * Makefile support
4693 * Ideas by Assar Westerlund <assar@sics.se> (2001)
4695 static void
4696 Makefile_targets (inf)
4697 FILE *inf;
4699 register char *bp;
4701 LOOP_ON_INPUT_LINES (inf, lb, bp)
4703 if (*bp == '\t' || *bp == '#')
4704 continue;
4705 while (*bp != '\0' && *bp != '=' && *bp != ':')
4706 bp++;
4707 if (*bp == ':' || (globals && *bp == '='))
4708 make_tag (lb.buffer, bp - lb.buffer, TRUE,
4709 lb.buffer, bp - lb.buffer + 1, lineno, linecharno);
4715 * Pascal parsing
4716 * Original code by Mosur K. Mohan (1989)
4718 * Locates tags for procedures & functions. Doesn't do any type- or
4719 * var-definitions. It does look for the keyword "extern" or
4720 * "forward" immediately following the procedure statement; if found,
4721 * the tag is skipped.
4723 static void
4724 Pascal_functions (inf)
4725 FILE *inf;
4727 linebuffer tline; /* mostly copied from C_entries */
4728 long save_lcno;
4729 int save_lineno, namelen, taglen;
4730 char c, *name;
4732 bool /* each of these flags is TRUE iff: */
4733 incomment, /* point is inside a comment */
4734 inquote, /* point is inside '..' string */
4735 get_tagname, /* point is after PROCEDURE/FUNCTION
4736 keyword, so next item = potential tag */
4737 found_tag, /* point is after a potential tag */
4738 inparms, /* point is within parameter-list */
4739 verify_tag; /* point has passed the parm-list, so the
4740 next token will determine whether this
4741 is a FORWARD/EXTERN to be ignored, or
4742 whether it is a real tag */
4744 save_lcno = save_lineno = namelen = taglen = 0; /* keep compiler quiet */
4745 name = NULL; /* keep compiler quiet */
4746 dbp = lb.buffer;
4747 *dbp = '\0';
4748 linebuffer_init (&tline);
4750 incomment = inquote = FALSE;
4751 found_tag = FALSE; /* have a proc name; check if extern */
4752 get_tagname = FALSE; /* found "procedure" keyword */
4753 inparms = FALSE; /* found '(' after "proc" */
4754 verify_tag = FALSE; /* check if "extern" is ahead */
4757 while (!feof (inf)) /* long main loop to get next char */
4759 c = *dbp++;
4760 if (c == '\0') /* if end of line */
4762 readline (&lb, inf);
4763 dbp = lb.buffer;
4764 if (*dbp == '\0')
4765 continue;
4766 if (!((found_tag && verify_tag)
4767 || get_tagname))
4768 c = *dbp++; /* only if don't need *dbp pointing
4769 to the beginning of the name of
4770 the procedure or function */
4772 if (incomment)
4774 if (c == '}') /* within { } comments */
4775 incomment = FALSE;
4776 else if (c == '*' && *dbp == ')') /* within (* *) comments */
4778 dbp++;
4779 incomment = FALSE;
4781 continue;
4783 else if (inquote)
4785 if (c == '\'')
4786 inquote = FALSE;
4787 continue;
4789 else
4790 switch (c)
4792 case '\'':
4793 inquote = TRUE; /* found first quote */
4794 continue;
4795 case '{': /* found open { comment */
4796 incomment = TRUE;
4797 continue;
4798 case '(':
4799 if (*dbp == '*') /* found open (* comment */
4801 incomment = TRUE;
4802 dbp++;
4804 else if (found_tag) /* found '(' after tag, i.e., parm-list */
4805 inparms = TRUE;
4806 continue;
4807 case ')': /* end of parms list */
4808 if (inparms)
4809 inparms = FALSE;
4810 continue;
4811 case ';':
4812 if (found_tag && !inparms) /* end of proc or fn stmt */
4814 verify_tag = TRUE;
4815 break;
4817 continue;
4819 if (found_tag && verify_tag && (*dbp != ' '))
4821 /* Check if this is an "extern" declaration. */
4822 if (*dbp == '\0')
4823 continue;
4824 if (lowcase (*dbp == 'e'))
4826 if (nocase_tail ("extern")) /* superfluous, really! */
4828 found_tag = FALSE;
4829 verify_tag = FALSE;
4832 else if (lowcase (*dbp) == 'f')
4834 if (nocase_tail ("forward")) /* check for forward reference */
4836 found_tag = FALSE;
4837 verify_tag = FALSE;
4840 if (found_tag && verify_tag) /* not external proc, so make tag */
4842 found_tag = FALSE;
4843 verify_tag = FALSE;
4844 make_tag (name, namelen, TRUE,
4845 tline.buffer, taglen, save_lineno, save_lcno);
4846 continue;
4849 if (get_tagname) /* grab name of proc or fn */
4851 char *cp;
4853 if (*dbp == '\0')
4854 continue;
4856 /* Find block name. */
4857 for (cp = dbp + 1; *cp != '\0' && !endtoken (*cp); cp++)
4858 continue;
4860 /* Save all values for later tagging. */
4861 linebuffer_setlen (&tline, lb.len);
4862 strcpy (tline.buffer, lb.buffer);
4863 save_lineno = lineno;
4864 save_lcno = linecharno;
4865 name = tline.buffer + (dbp - lb.buffer);
4866 namelen = cp - dbp;
4867 taglen = cp - lb.buffer + 1;
4869 dbp = cp; /* set dbp to e-o-token */
4870 get_tagname = FALSE;
4871 found_tag = TRUE;
4872 continue;
4874 /* And proceed to check for "extern". */
4876 else if (!incomment && !inquote && !found_tag)
4878 /* Check for proc/fn keywords. */
4879 switch (lowcase (c))
4881 case 'p':
4882 if (nocase_tail ("rocedure")) /* c = 'p', dbp has advanced */
4883 get_tagname = TRUE;
4884 continue;
4885 case 'f':
4886 if (nocase_tail ("unction"))
4887 get_tagname = TRUE;
4888 continue;
4891 } /* while not eof */
4893 free (tline.buffer);
4898 * Lisp tag functions
4899 * look for (def or (DEF, quote or QUOTE
4902 static void L_getit __P((void));
4904 static void
4905 L_getit ()
4907 if (*dbp == '\'') /* Skip prefix quote */
4908 dbp++;
4909 else if (*dbp == '(')
4911 dbp++;
4912 /* Try to skip "(quote " */
4913 if (!LOOKING_AT (dbp, "quote") && !LOOKING_AT (dbp, "QUOTE"))
4914 /* Ok, then skip "(" before name in (defstruct (foo)) */
4915 dbp = skip_spaces (dbp);
4917 get_tag (dbp, NULL);
4920 static void
4921 Lisp_functions (inf)
4922 FILE *inf;
4924 LOOP_ON_INPUT_LINES (inf, lb, dbp)
4926 if (dbp[0] != '(')
4927 continue;
4929 if (strneq (dbp+1, "def", 3) || strneq (dbp+1, "DEF", 3))
4931 dbp = skip_non_spaces (dbp);
4932 dbp = skip_spaces (dbp);
4933 L_getit ();
4935 else
4937 /* Check for (foo::defmumble name-defined ... */
4939 dbp++;
4940 while (!notinname (*dbp) && *dbp != ':');
4941 if (*dbp == ':')
4944 dbp++;
4945 while (*dbp == ':');
4947 if (strneq (dbp, "def", 3) || strneq (dbp, "DEF", 3))
4949 dbp = skip_non_spaces (dbp);
4950 dbp = skip_spaces (dbp);
4951 L_getit ();
4960 * Lua script language parsing
4961 * Original code by David A. Capello <dacap@users.sourceforge.net> (2004)
4963 * "function" and "local function" are tags if they start at column 1.
4965 static void
4966 Lua_functions (inf)
4967 FILE *inf;
4969 register char *bp;
4971 LOOP_ON_INPUT_LINES (inf, lb, bp)
4973 if (bp[0] != 'f' && bp[0] != 'l')
4974 continue;
4976 LOOKING_AT (bp, "local"); /* skip possible "local" */
4978 if (LOOKING_AT (bp, "function"))
4979 get_tag (bp, NULL);
4985 * Postscript tags
4986 * Just look for lines where the first character is '/'
4987 * Also look at "defineps" for PSWrap
4988 * Ideas by:
4989 * Richard Mlynarik <mly@adoc.xerox.com> (1997)
4990 * Masatake Yamato <masata-y@is.aist-nara.ac.jp> (1999)
4992 static void
4993 PS_functions (inf)
4994 FILE *inf;
4996 register char *bp, *ep;
4998 LOOP_ON_INPUT_LINES (inf, lb, bp)
5000 if (bp[0] == '/')
5002 for (ep = bp+1;
5003 *ep != '\0' && *ep != ' ' && *ep != '{';
5004 ep++)
5005 continue;
5006 make_tag (bp, ep - bp, TRUE,
5007 lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
5009 else if (LOOKING_AT (bp, "defineps"))
5010 get_tag (bp, NULL);
5016 * Forth tags
5017 * Ignore anything after \ followed by space or in ( )
5018 * Look for words defined by :
5019 * Look for constant, code, create, defer, value, and variable
5020 * OBP extensions: Look for buffer:, field,
5021 * Ideas by Eduardo Horvath <eeh@netbsd.org> (2004)
5023 static void
5024 Forth_words (inf)
5025 FILE *inf;
5027 register char *bp;
5029 LOOP_ON_INPUT_LINES (inf, lb, bp)
5030 while ((bp = skip_spaces (bp))[0] != '\0')
5031 if (bp[0] == '\\' && iswhite(bp[1]))
5032 break; /* read next line */
5033 else if (bp[0] == '(' && iswhite(bp[1]))
5034 do /* skip to ) or eol */
5035 bp++;
5036 while (*bp != ')' && *bp != '\0');
5037 else if ((bp[0] == ':' && iswhite(bp[1]) && bp++)
5038 || LOOKING_AT_NOCASE (bp, "constant")
5039 || LOOKING_AT_NOCASE (bp, "code")
5040 || LOOKING_AT_NOCASE (bp, "create")
5041 || LOOKING_AT_NOCASE (bp, "defer")
5042 || LOOKING_AT_NOCASE (bp, "value")
5043 || LOOKING_AT_NOCASE (bp, "variable")
5044 || LOOKING_AT_NOCASE (bp, "buffer:")
5045 || LOOKING_AT_NOCASE (bp, "field"))
5046 get_tag (skip_spaces (bp), NULL); /* Yay! A definition! */
5047 else
5048 bp = skip_non_spaces (bp);
5053 * Scheme tag functions
5054 * look for (def... xyzzy
5055 * (def... (xyzzy
5056 * (def ... ((...(xyzzy ....
5057 * (set! xyzzy
5058 * Original code by Ken Haase (1985?)
5060 static void
5061 Scheme_functions (inf)
5062 FILE *inf;
5064 register char *bp;
5066 LOOP_ON_INPUT_LINES (inf, lb, bp)
5068 if (strneq (bp, "(def", 4) || strneq (bp, "(DEF", 4))
5070 bp = skip_non_spaces (bp+4);
5071 /* Skip over open parens and white space */
5072 while (notinname (*bp))
5073 bp++;
5074 get_tag (bp, NULL);
5076 if (LOOKING_AT (bp, "(SET!") || LOOKING_AT (bp, "(set!"))
5077 get_tag (bp, NULL);
5082 /* Find tags in TeX and LaTeX input files. */
5084 /* TEX_toktab is a table of TeX control sequences that define tags.
5085 * Each entry records one such control sequence.
5087 * Original code from who knows whom.
5088 * Ideas by:
5089 * Stefan Monnier (2002)
5092 static linebuffer *TEX_toktab = NULL; /* Table with tag tokens */
5094 /* Default set of control sequences to put into TEX_toktab.
5095 The value of environment var TEXTAGS is prepended to this. */
5096 static char *TEX_defenv = "\
5097 :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem\
5098 :part:appendix:entry:index:def\
5099 :newcommand:renewcommand:newenvironment:renewenvironment";
5101 static void TEX_mode __P((FILE *));
5102 static void TEX_decode_env __P((char *, char *));
5104 static char TEX_esc = '\\';
5105 static char TEX_opgrp = '{';
5106 static char TEX_clgrp = '}';
5109 * TeX/LaTeX scanning loop.
5111 static void
5112 TeX_commands (inf)
5113 FILE *inf;
5115 char *cp;
5116 linebuffer *key;
5118 /* Select either \ or ! as escape character. */
5119 TEX_mode (inf);
5121 /* Initialize token table once from environment. */
5122 if (TEX_toktab == NULL)
5123 TEX_decode_env ("TEXTAGS", TEX_defenv);
5125 LOOP_ON_INPUT_LINES (inf, lb, cp)
5127 /* Look at each TEX keyword in line. */
5128 for (;;)
5130 /* Look for a TEX escape. */
5131 while (*cp++ != TEX_esc)
5132 if (cp[-1] == '\0' || cp[-1] == '%')
5133 goto tex_next_line;
5135 for (key = TEX_toktab; key->buffer != NULL; key++)
5136 if (strneq (cp, key->buffer, key->len))
5138 register char *p;
5139 int namelen, linelen;
5140 bool opgrp = FALSE;
5142 cp = skip_spaces (cp + key->len);
5143 if (*cp == TEX_opgrp)
5145 opgrp = TRUE;
5146 cp++;
5148 for (p = cp;
5149 (!iswhite (*p) && *p != '#' &&
5150 *p != TEX_opgrp && *p != TEX_clgrp);
5151 p++)
5152 continue;
5153 namelen = p - cp;
5154 linelen = lb.len;
5155 if (!opgrp || *p == TEX_clgrp)
5157 while (*p != '\0' && *p != TEX_opgrp && *p != TEX_clgrp)
5158 *p++;
5159 linelen = p - lb.buffer + 1;
5161 make_tag (cp, namelen, TRUE,
5162 lb.buffer, linelen, lineno, linecharno);
5163 goto tex_next_line; /* We only tag a line once */
5166 tex_next_line:
5171 #define TEX_LESC '\\'
5172 #define TEX_SESC '!'
5174 /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
5175 chars accordingly. */
5176 static void
5177 TEX_mode (inf)
5178 FILE *inf;
5180 int c;
5182 while ((c = getc (inf)) != EOF)
5184 /* Skip to next line if we hit the TeX comment char. */
5185 if (c == '%')
5186 while (c != '\n')
5187 c = getc (inf);
5188 else if (c == TEX_LESC || c == TEX_SESC )
5189 break;
5192 if (c == TEX_LESC)
5194 TEX_esc = TEX_LESC;
5195 TEX_opgrp = '{';
5196 TEX_clgrp = '}';
5198 else
5200 TEX_esc = TEX_SESC;
5201 TEX_opgrp = '<';
5202 TEX_clgrp = '>';
5204 /* If the input file is compressed, inf is a pipe, and rewind may fail.
5205 No attempt is made to correct the situation. */
5206 rewind (inf);
5209 /* Read environment and prepend it to the default string.
5210 Build token table. */
5211 static void
5212 TEX_decode_env (evarname, defenv)
5213 char *evarname;
5214 char *defenv;
5216 register char *env, *p;
5217 int i, len;
5219 /* Append default string to environment. */
5220 env = getenv (evarname);
5221 if (!env)
5222 env = defenv;
5223 else
5225 char *oldenv = env;
5226 env = concat (oldenv, defenv, "");
5229 /* Allocate a token table */
5230 for (len = 1, p = env; p;)
5231 if ((p = etags_strchr (p, ':')) && *++p != '\0')
5232 len++;
5233 TEX_toktab = xnew (len, linebuffer);
5235 /* Unpack environment string into token table. Be careful about */
5236 /* zero-length strings (leading ':', "::" and trailing ':') */
5237 for (i = 0; *env != '\0';)
5239 p = etags_strchr (env, ':');
5240 if (!p) /* End of environment string. */
5241 p = env + strlen (env);
5242 if (p - env > 0)
5243 { /* Only non-zero strings. */
5244 TEX_toktab[i].buffer = savenstr (env, p - env);
5245 TEX_toktab[i].len = p - env;
5246 i++;
5248 if (*p)
5249 env = p + 1;
5250 else
5252 TEX_toktab[i].buffer = NULL; /* Mark end of table. */
5253 TEX_toktab[i].len = 0;
5254 break;
5260 /* Texinfo support. Dave Love, Mar. 2000. */
5261 static void
5262 Texinfo_nodes (inf)
5263 FILE * inf;
5265 char *cp, *start;
5266 LOOP_ON_INPUT_LINES (inf, lb, cp)
5267 if (LOOKING_AT (cp, "@node"))
5269 start = cp;
5270 while (*cp != '\0' && *cp != ',')
5271 cp++;
5272 make_tag (start, cp - start, TRUE,
5273 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
5279 * HTML support.
5280 * Contents of <title>, <h1>, <h2>, <h3> are tags.
5281 * Contents of <a name=xxx> are tags with name xxx.
5283 * Francesco Potortì, 2002.
5285 static void
5286 HTML_labels (inf)
5287 FILE * inf;
5289 bool getnext = FALSE; /* next text outside of HTML tags is a tag */
5290 bool skiptag = FALSE; /* skip to the end of the current HTML tag */
5291 bool intag = FALSE; /* inside an html tag, looking for ID= */
5292 bool inanchor = FALSE; /* when INTAG, is an anchor, look for NAME= */
5293 char *end;
5296 linebuffer_setlen (&token_name, 0); /* no name in buffer */
5298 LOOP_ON_INPUT_LINES (inf, lb, dbp)
5299 for (;;) /* loop on the same line */
5301 if (skiptag) /* skip HTML tag */
5303 while (*dbp != '\0' && *dbp != '>')
5304 dbp++;
5305 if (*dbp == '>')
5307 dbp += 1;
5308 skiptag = FALSE;
5309 continue; /* look on the same line */
5311 break; /* go to next line */
5314 else if (intag) /* look for "name=" or "id=" */
5316 while (*dbp != '\0' && *dbp != '>'
5317 && lowcase (*dbp) != 'n' && lowcase (*dbp) != 'i')
5318 dbp++;
5319 if (*dbp == '\0')
5320 break; /* go to next line */
5321 if (*dbp == '>')
5323 dbp += 1;
5324 intag = FALSE;
5325 continue; /* look on the same line */
5327 if ((inanchor && LOOKING_AT_NOCASE (dbp, "name="))
5328 || LOOKING_AT_NOCASE (dbp, "id="))
5330 bool quoted = (dbp[0] == '"');
5332 if (quoted)
5333 for (end = ++dbp; *end != '\0' && *end != '"'; end++)
5334 continue;
5335 else
5336 for (end = dbp; *end != '\0' && intoken (*end); end++)
5337 continue;
5338 linebuffer_setlen (&token_name, end - dbp);
5339 strncpy (token_name.buffer, dbp, end - dbp);
5340 token_name.buffer[end - dbp] = '\0';
5342 dbp = end;
5343 intag = FALSE; /* we found what we looked for */
5344 skiptag = TRUE; /* skip to the end of the tag */
5345 getnext = TRUE; /* then grab the text */
5346 continue; /* look on the same line */
5348 dbp += 1;
5351 else if (getnext) /* grab next tokens and tag them */
5353 dbp = skip_spaces (dbp);
5354 if (*dbp == '\0')
5355 break; /* go to next line */
5356 if (*dbp == '<')
5358 intag = TRUE;
5359 inanchor = (lowcase (dbp[1]) == 'a' && !intoken (dbp[2]));
5360 continue; /* look on the same line */
5363 for (end = dbp + 1; *end != '\0' && *end != '<'; end++)
5364 continue;
5365 make_tag (token_name.buffer, token_name.len, TRUE,
5366 dbp, end - dbp, lineno, linecharno);
5367 linebuffer_setlen (&token_name, 0); /* no name in buffer */
5368 getnext = FALSE;
5369 break; /* go to next line */
5372 else /* look for an interesting HTML tag */
5374 while (*dbp != '\0' && *dbp != '<')
5375 dbp++;
5376 if (*dbp == '\0')
5377 break; /* go to next line */
5378 intag = TRUE;
5379 if (lowcase (dbp[1]) == 'a' && !intoken (dbp[2]))
5381 inanchor = TRUE;
5382 continue; /* look on the same line */
5384 else if (LOOKING_AT_NOCASE (dbp, "<title>")
5385 || LOOKING_AT_NOCASE (dbp, "<h1>")
5386 || LOOKING_AT_NOCASE (dbp, "<h2>")
5387 || LOOKING_AT_NOCASE (dbp, "<h3>"))
5389 intag = FALSE;
5390 getnext = TRUE;
5391 continue; /* look on the same line */
5393 dbp += 1;
5400 * Prolog support
5402 * Assumes that the predicate or rule starts at column 0.
5403 * Only the first clause of a predicate or rule is added.
5404 * Original code by Sunichirou Sugou (1989)
5405 * Rewritten by Anders Lindgren (1996)
5407 static int prolog_pr __P((char *, char *));
5408 static void prolog_skip_comment __P((linebuffer *, FILE *));
5409 static int prolog_atom __P((char *, int));
5411 static void
5412 Prolog_functions (inf)
5413 FILE *inf;
5415 char *cp, *last;
5416 int len;
5417 int allocated;
5419 allocated = 0;
5420 len = 0;
5421 last = NULL;
5423 LOOP_ON_INPUT_LINES (inf, lb, cp)
5425 if (cp[0] == '\0') /* Empty line */
5426 continue;
5427 else if (iswhite (cp[0])) /* Not a predicate */
5428 continue;
5429 else if (cp[0] == '/' && cp[1] == '*') /* comment. */
5430 prolog_skip_comment (&lb, inf);
5431 else if ((len = prolog_pr (cp, last)) > 0)
5433 /* Predicate or rule. Store the function name so that we
5434 only generate a tag for the first clause. */
5435 if (last == NULL)
5436 last = xnew(len + 1, char);
5437 else if (len + 1 > allocated)
5438 xrnew (last, len + 1, char);
5439 allocated = len + 1;
5440 strncpy (last, cp, len);
5441 last[len] = '\0';
5447 static void
5448 prolog_skip_comment (plb, inf)
5449 linebuffer *plb;
5450 FILE *inf;
5452 char *cp;
5456 for (cp = plb->buffer; *cp != '\0'; cp++)
5457 if (cp[0] == '*' && cp[1] == '/')
5458 return;
5459 readline (plb, inf);
5461 while (!feof(inf));
5465 * A predicate or rule definition is added if it matches:
5466 * <beginning of line><Prolog Atom><whitespace>(
5467 * or <beginning of line><Prolog Atom><whitespace>:-
5469 * It is added to the tags database if it doesn't match the
5470 * name of the previous clause header.
5472 * Return the size of the name of the predicate or rule, or 0 if no
5473 * header was found.
5475 static int
5476 prolog_pr (s, last)
5477 char *s;
5478 char *last; /* Name of last clause. */
5480 int pos;
5481 int len;
5483 pos = prolog_atom (s, 0);
5484 if (pos < 1)
5485 return 0;
5487 len = pos;
5488 pos = skip_spaces (s + pos) - s;
5490 if ((s[pos] == '.'
5491 || (s[pos] == '(' && (pos += 1))
5492 || (s[pos] == ':' && s[pos + 1] == '-' && (pos += 2)))
5493 && (last == NULL /* save only the first clause */
5494 || len != (int)strlen (last)
5495 || !strneq (s, last, len)))
5497 make_tag (s, len, TRUE, s, pos, lineno, linecharno);
5498 return len;
5500 else
5501 return 0;
5505 * Consume a Prolog atom.
5506 * Return the number of bytes consumed, or -1 if there was an error.
5508 * A prolog atom, in this context, could be one of:
5509 * - An alphanumeric sequence, starting with a lower case letter.
5510 * - A quoted arbitrary string. Single quotes can escape themselves.
5511 * Backslash quotes everything.
5513 static int
5514 prolog_atom (s, pos)
5515 char *s;
5516 int pos;
5518 int origpos;
5520 origpos = pos;
5522 if (ISLOWER(s[pos]) || (s[pos] == '_'))
5524 /* The atom is unquoted. */
5525 pos++;
5526 while (ISALNUM(s[pos]) || (s[pos] == '_'))
5528 pos++;
5530 return pos - origpos;
5532 else if (s[pos] == '\'')
5534 pos++;
5536 for (;;)
5538 if (s[pos] == '\'')
5540 pos++;
5541 if (s[pos] != '\'')
5542 break;
5543 pos++; /* A double quote */
5545 else if (s[pos] == '\0')
5546 /* Multiline quoted atoms are ignored. */
5547 return -1;
5548 else if (s[pos] == '\\')
5550 if (s[pos+1] == '\0')
5551 return -1;
5552 pos += 2;
5554 else
5555 pos++;
5557 return pos - origpos;
5559 else
5560 return -1;
5565 * Support for Erlang
5567 * Generates tags for functions, defines, and records.
5568 * Assumes that Erlang functions start at column 0.
5569 * Original code by Anders Lindgren (1996)
5571 static int erlang_func __P((char *, char *));
5572 static void erlang_attribute __P((char *));
5573 static int erlang_atom __P((char *));
5575 static void
5576 Erlang_functions (inf)
5577 FILE *inf;
5579 char *cp, *last;
5580 int len;
5581 int allocated;
5583 allocated = 0;
5584 len = 0;
5585 last = NULL;
5587 LOOP_ON_INPUT_LINES (inf, lb, cp)
5589 if (cp[0] == '\0') /* Empty line */
5590 continue;
5591 else if (iswhite (cp[0])) /* Not function nor attribute */
5592 continue;
5593 else if (cp[0] == '%') /* comment */
5594 continue;
5595 else if (cp[0] == '"') /* Sometimes, strings start in column one */
5596 continue;
5597 else if (cp[0] == '-') /* attribute, e.g. "-define" */
5599 erlang_attribute (cp);
5600 last = NULL;
5602 else if ((len = erlang_func (cp, last)) > 0)
5605 * Function. Store the function name so that we only
5606 * generates a tag for the first clause.
5608 if (last == NULL)
5609 last = xnew (len + 1, char);
5610 else if (len + 1 > allocated)
5611 xrnew (last, len + 1, char);
5612 allocated = len + 1;
5613 strncpy (last, cp, len);
5614 last[len] = '\0';
5621 * A function definition is added if it matches:
5622 * <beginning of line><Erlang Atom><whitespace>(
5624 * It is added to the tags database if it doesn't match the
5625 * name of the previous clause header.
5627 * Return the size of the name of the function, or 0 if no function
5628 * was found.
5630 static int
5631 erlang_func (s, last)
5632 char *s;
5633 char *last; /* Name of last clause. */
5635 int pos;
5636 int len;
5638 pos = erlang_atom (s);
5639 if (pos < 1)
5640 return 0;
5642 len = pos;
5643 pos = skip_spaces (s + pos) - s;
5645 /* Save only the first clause. */
5646 if (s[pos++] == '('
5647 && (last == NULL
5648 || len != (int)strlen (last)
5649 || !strneq (s, last, len)))
5651 make_tag (s, len, TRUE, s, pos, lineno, linecharno);
5652 return len;
5655 return 0;
5660 * Handle attributes. Currently, tags are generated for defines
5661 * and records.
5663 * They are on the form:
5664 * -define(foo, bar).
5665 * -define(Foo(M, N), M+N).
5666 * -record(graph, {vtab = notable, cyclic = true}).
5668 static void
5669 erlang_attribute (s)
5670 char *s;
5672 char *cp = s;
5674 if ((LOOKING_AT (cp, "-define") || LOOKING_AT (cp, "-record"))
5675 && *cp++ == '(')
5677 int len = erlang_atom (skip_spaces (cp));
5678 if (len > 0)
5679 make_tag (cp, len, TRUE, s, cp + len - s, lineno, linecharno);
5681 return;
5686 * Consume an Erlang atom (or variable).
5687 * Return the number of bytes consumed, or -1 if there was an error.
5689 static int
5690 erlang_atom (s)
5691 char *s;
5693 int pos = 0;
5695 if (ISALPHA (s[pos]) || s[pos] == '_')
5697 /* The atom is unquoted. */
5699 pos++;
5700 while (ISALNUM (s[pos]) || s[pos] == '_');
5702 else if (s[pos] == '\'')
5704 for (pos++; s[pos] != '\''; pos++)
5705 if (s[pos] == '\0' /* multiline quoted atoms are ignored */
5706 || (s[pos] == '\\' && s[++pos] == '\0'))
5707 return 0;
5708 pos++;
5711 return pos;
5715 #ifdef ETAGS_REGEXPS
5717 static char *scan_separators __P((char *));
5718 static void add_regex __P((char *, language *));
5719 static char *substitute __P((char *, char *, struct re_registers *));
5722 * Take a string like "/blah/" and turn it into "blah", verifying
5723 * that the first and last characters are the same, and handling
5724 * quoted separator characters. Actually, stops on the occurrence of
5725 * an unquoted separator. Also process \t, \n, etc. and turn into
5726 * appropriate characters. Works in place. Null terminates name string.
5727 * Returns pointer to terminating separator, or NULL for
5728 * unterminated regexps.
5730 static char *
5731 scan_separators (name)
5732 char *name;
5734 char sep = name[0];
5735 char *copyto = name;
5736 bool quoted = FALSE;
5738 for (++name; *name != '\0'; ++name)
5740 if (quoted)
5742 switch (*name)
5744 case 'a': *copyto++ = '\007'; break; /* BEL (bell) */
5745 case 'b': *copyto++ = '\b'; break; /* BS (back space) */
5746 case 'd': *copyto++ = 0177; break; /* DEL (delete) */
5747 case 'e': *copyto++ = 033; break; /* ESC (delete) */
5748 case 'f': *copyto++ = '\f'; break; /* FF (form feed) */
5749 case 'n': *copyto++ = '\n'; break; /* NL (new line) */
5750 case 'r': *copyto++ = '\r'; break; /* CR (carriage return) */
5751 case 't': *copyto++ = '\t'; break; /* TAB (horizontal tab) */
5752 case 'v': *copyto++ = '\v'; break; /* VT (vertical tab) */
5753 default:
5754 if (*name == sep)
5755 *copyto++ = sep;
5756 else
5758 /* Something else is quoted, so preserve the quote. */
5759 *copyto++ = '\\';
5760 *copyto++ = *name;
5762 break;
5764 quoted = FALSE;
5766 else if (*name == '\\')
5767 quoted = TRUE;
5768 else if (*name == sep)
5769 break;
5770 else
5771 *copyto++ = *name;
5773 if (*name != sep)
5774 name = NULL; /* signal unterminated regexp */
5776 /* Terminate copied string. */
5777 *copyto = '\0';
5778 return name;
5781 /* Look at the argument of --regex or --no-regex and do the right
5782 thing. Same for each line of a regexp file. */
5783 static void
5784 analyse_regex (regex_arg)
5785 char *regex_arg;
5787 if (regex_arg == NULL)
5789 free_regexps (); /* --no-regex: remove existing regexps */
5790 return;
5793 /* A real --regexp option or a line in a regexp file. */
5794 switch (regex_arg[0])
5796 /* Comments in regexp file or null arg to --regex. */
5797 case '\0':
5798 case ' ':
5799 case '\t':
5800 break;
5802 /* Read a regex file. This is recursive and may result in a
5803 loop, which will stop when the file descriptors are exhausted. */
5804 case '@':
5806 FILE *regexfp;
5807 linebuffer regexbuf;
5808 char *regexfile = regex_arg + 1;
5810 /* regexfile is a file containing regexps, one per line. */
5811 regexfp = fopen (regexfile, "r");
5812 if (regexfp == NULL)
5814 pfatal (regexfile);
5815 return;
5817 linebuffer_init (&regexbuf);
5818 while (readline_internal (&regexbuf, regexfp) > 0)
5819 analyse_regex (regexbuf.buffer);
5820 free (regexbuf.buffer);
5821 fclose (regexfp);
5823 break;
5825 /* Regexp to be used for a specific language only. */
5826 case '{':
5828 language *lang;
5829 char *lang_name = regex_arg + 1;
5830 char *cp;
5832 for (cp = lang_name; *cp != '}'; cp++)
5833 if (*cp == '\0')
5835 error ("unterminated language name in regex: %s", regex_arg);
5836 return;
5838 *cp++ = '\0';
5839 lang = get_language_from_langname (lang_name);
5840 if (lang == NULL)
5841 return;
5842 add_regex (cp, lang);
5844 break;
5846 /* Regexp to be used for any language. */
5847 default:
5848 add_regex (regex_arg, NULL);
5849 break;
5853 /* Separate the regexp pattern, compile it,
5854 and care for optional name and modifiers. */
5855 static void
5856 add_regex (regexp_pattern, lang)
5857 char *regexp_pattern;
5858 language *lang;
5860 static struct re_pattern_buffer zeropattern;
5861 char sep, *pat, *name, *modifiers;
5862 const char *err;
5863 struct re_pattern_buffer *patbuf;
5864 regexp *rp;
5865 bool
5866 force_explicit_name = TRUE, /* do not use implicit tag names */
5867 ignore_case = FALSE, /* case is significant */
5868 multi_line = FALSE, /* matches are done one line at a time */
5869 single_line = FALSE; /* dot does not match newline */
5872 if (strlen(regexp_pattern) < 3)
5874 error ("null regexp", (char *)NULL);
5875 return;
5877 sep = regexp_pattern[0];
5878 name = scan_separators (regexp_pattern);
5879 if (name == NULL)
5881 error ("%s: unterminated regexp", regexp_pattern);
5882 return;
5884 if (name[1] == sep)
5886 error ("null name for regexp \"%s\"", regexp_pattern);
5887 return;
5889 modifiers = scan_separators (name);
5890 if (modifiers == NULL) /* no terminating separator --> no name */
5892 modifiers = name;
5893 name = "";
5895 else
5896 modifiers += 1; /* skip separator */
5898 /* Parse regex modifiers. */
5899 for (; modifiers[0] != '\0'; modifiers++)
5900 switch (modifiers[0])
5902 case 'N':
5903 if (modifiers == name)
5904 error ("forcing explicit tag name but no name, ignoring", NULL);
5905 force_explicit_name = TRUE;
5906 break;
5907 case 'i':
5908 ignore_case = TRUE;
5909 break;
5910 case 's':
5911 single_line = TRUE;
5912 /* FALLTHRU */
5913 case 'm':
5914 multi_line = TRUE;
5915 need_filebuf = TRUE;
5916 break;
5917 default:
5919 char wrongmod [2];
5920 wrongmod[0] = modifiers[0];
5921 wrongmod[1] = '\0';
5922 error ("invalid regexp modifier `%s', ignoring", wrongmod);
5924 break;
5927 patbuf = xnew (1, struct re_pattern_buffer);
5928 *patbuf = zeropattern;
5929 if (ignore_case)
5931 static char lc_trans[CHARS];
5932 int i;
5933 for (i = 0; i < CHARS; i++)
5934 lc_trans[i] = lowcase (i);
5935 patbuf->translate = lc_trans; /* translation table to fold case */
5938 if (multi_line)
5939 pat = concat ("^", regexp_pattern, ""); /* anchor to beginning of line */
5940 else
5941 pat = regexp_pattern;
5943 if (single_line)
5944 re_set_syntax (RE_SYNTAX_EMACS | RE_DOT_NEWLINE);
5945 else
5946 re_set_syntax (RE_SYNTAX_EMACS);
5948 err = re_compile_pattern (pat, strlen (regexp_pattern), patbuf);
5949 if (multi_line)
5950 free (pat);
5951 if (err != NULL)
5953 error ("%s while compiling pattern", err);
5954 return;
5957 rp = p_head;
5958 p_head = xnew (1, regexp);
5959 p_head->pattern = savestr (regexp_pattern);
5960 p_head->p_next = rp;
5961 p_head->lang = lang;
5962 p_head->pat = patbuf;
5963 p_head->name = savestr (name);
5964 p_head->error_signaled = FALSE;
5965 p_head->force_explicit_name = force_explicit_name;
5966 p_head->ignore_case = ignore_case;
5967 p_head->multi_line = multi_line;
5971 * Do the substitutions indicated by the regular expression and
5972 * arguments.
5974 static char *
5975 substitute (in, out, regs)
5976 char *in, *out;
5977 struct re_registers *regs;
5979 char *result, *t;
5980 int size, dig, diglen;
5982 result = NULL;
5983 size = strlen (out);
5985 /* Pass 1: figure out how much to allocate by finding all \N strings. */
5986 if (out[size - 1] == '\\')
5987 fatal ("pattern error in \"%s\"", out);
5988 for (t = etags_strchr (out, '\\');
5989 t != NULL;
5990 t = etags_strchr (t + 2, '\\'))
5991 if (ISDIGIT (t[1]))
5993 dig = t[1] - '0';
5994 diglen = regs->end[dig] - regs->start[dig];
5995 size += diglen - 2;
5997 else
5998 size -= 1;
6000 /* Allocate space and do the substitutions. */
6001 assert (size >= 0);
6002 result = xnew (size + 1, char);
6004 for (t = result; *out != '\0'; out++)
6005 if (*out == '\\' && ISDIGIT (*++out))
6007 dig = *out - '0';
6008 diglen = regs->end[dig] - regs->start[dig];
6009 strncpy (t, in + regs->start[dig], diglen);
6010 t += diglen;
6012 else
6013 *t++ = *out;
6014 *t = '\0';
6016 assert (t <= result + size);
6017 assert (t - result == (int)strlen (result));
6019 return result;
6022 /* Deallocate all regexps. */
6023 static void
6024 free_regexps ()
6026 regexp *rp;
6027 while (p_head != NULL)
6029 rp = p_head->p_next;
6030 free (p_head->pattern);
6031 free (p_head->name);
6032 free (p_head);
6033 p_head = rp;
6035 return;
6039 * Reads the whole file as a single string from `filebuf' and looks for
6040 * multi-line regular expressions, creating tags on matches.
6041 * readline already dealt with normal regexps.
6043 * Idea by Ben Wing <ben@666.com> (2002).
6045 static void
6046 regex_tag_multiline ()
6048 char *buffer = filebuf.buffer;
6049 regexp *rp;
6050 char *name;
6052 for (rp = p_head; rp != NULL; rp = rp->p_next)
6054 int match = 0;
6056 if (!rp->multi_line)
6057 continue; /* skip normal regexps */
6059 /* Generic initialisations before parsing file from memory. */
6060 lineno = 1; /* reset global line number */
6061 charno = 0; /* reset global char number */
6062 linecharno = 0; /* reset global char number of line start */
6064 /* Only use generic regexps or those for the current language. */
6065 if (rp->lang != NULL && rp->lang != curfdp->lang)
6066 continue;
6068 while (match >= 0 && match < filebuf.len)
6070 match = re_search (rp->pat, buffer, filebuf.len, charno,
6071 filebuf.len - match, &rp->regs);
6072 switch (match)
6074 case -2:
6075 /* Some error. */
6076 if (!rp->error_signaled)
6078 error ("regexp stack overflow while matching \"%s\"",
6079 rp->pattern);
6080 rp->error_signaled = TRUE;
6082 break;
6083 case -1:
6084 /* No match. */
6085 break;
6086 default:
6087 if (match == rp->regs.end[0])
6089 if (!rp->error_signaled)
6091 error ("regexp matches the empty string: \"%s\"",
6092 rp->pattern);
6093 rp->error_signaled = TRUE;
6095 match = -3; /* exit from while loop */
6096 break;
6099 /* Match occurred. Construct a tag. */
6100 while (charno < rp->regs.end[0])
6101 if (buffer[charno++] == '\n')
6102 lineno++, linecharno = charno;
6103 name = rp->name;
6104 if (name[0] == '\0')
6105 name = NULL;
6106 else /* make a named tag */
6107 name = substitute (buffer, rp->name, &rp->regs);
6108 if (rp->force_explicit_name)
6109 /* Force explicit tag name, if a name is there. */
6110 pfnote (name, TRUE, buffer + linecharno,
6111 charno - linecharno + 1, lineno, linecharno);
6112 else
6113 make_tag (name, strlen (name), TRUE, buffer + linecharno,
6114 charno - linecharno + 1, lineno, linecharno);
6115 break;
6121 #endif /* ETAGS_REGEXPS */
6124 static bool
6125 nocase_tail (cp)
6126 char *cp;
6128 register int len = 0;
6130 while (*cp != '\0' && lowcase (*cp) == lowcase (dbp[len]))
6131 cp++, len++;
6132 if (*cp == '\0' && !intoken (dbp[len]))
6134 dbp += len;
6135 return TRUE;
6137 return FALSE;
6140 static void
6141 get_tag (bp, namepp)
6142 register char *bp;
6143 char **namepp;
6145 register char *cp = bp;
6147 if (*bp != '\0')
6149 /* Go till you get to white space or a syntactic break */
6150 for (cp = bp + 1; !notinname (*cp); cp++)
6151 continue;
6152 make_tag (bp, cp - bp, TRUE,
6153 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
6156 if (namepp != NULL)
6157 *namepp = savenstr (bp, cp - bp);
6161 * Read a line of text from `stream' into `lbp', excluding the
6162 * newline or CR-NL, if any. Return the number of characters read from
6163 * `stream', which is the length of the line including the newline.
6165 * On DOS or Windows we do not count the CR character, if any before the
6166 * NL, in the returned length; this mirrors the behavior of Emacs on those
6167 * platforms (for text files, it translates CR-NL to NL as it reads in the
6168 * file).
6170 * If multi-line regular expressions are requested, each line read is
6171 * appended to `filebuf'.
6173 static long
6174 readline_internal (lbp, stream)
6175 linebuffer *lbp;
6176 register FILE *stream;
6178 char *buffer = lbp->buffer;
6179 register char *p = lbp->buffer;
6180 register char *pend;
6181 int chars_deleted;
6183 pend = p + lbp->size; /* Separate to avoid 386/IX compiler bug. */
6185 for (;;)
6187 register int c = getc (stream);
6188 if (p == pend)
6190 /* We're at the end of linebuffer: expand it. */
6191 lbp->size *= 2;
6192 xrnew (buffer, lbp->size, char);
6193 p += buffer - lbp->buffer;
6194 pend = buffer + lbp->size;
6195 lbp->buffer = buffer;
6197 if (c == EOF)
6199 *p = '\0';
6200 chars_deleted = 0;
6201 break;
6203 if (c == '\n')
6205 if (p > buffer && p[-1] == '\r')
6207 p -= 1;
6208 #ifdef DOS_NT
6209 /* Assume CRLF->LF translation will be performed by Emacs
6210 when loading this file, so CRs won't appear in the buffer.
6211 It would be cleaner to compensate within Emacs;
6212 however, Emacs does not know how many CRs were deleted
6213 before any given point in the file. */
6214 chars_deleted = 1;
6215 #else
6216 chars_deleted = 2;
6217 #endif
6219 else
6221 chars_deleted = 1;
6223 *p = '\0';
6224 break;
6226 *p++ = c;
6228 lbp->len = p - buffer;
6230 if (need_filebuf /* we need filebuf for multi-line regexps */
6231 && chars_deleted > 0) /* not at EOF */
6233 while (filebuf.size <= filebuf.len + lbp->len + 1) /* +1 for \n */
6235 /* Expand filebuf. */
6236 filebuf.size *= 2;
6237 xrnew (filebuf.buffer, filebuf.size, char);
6239 strncpy (filebuf.buffer + filebuf.len, lbp->buffer, lbp->len);
6240 filebuf.len += lbp->len;
6241 filebuf.buffer[filebuf.len++] = '\n';
6242 filebuf.buffer[filebuf.len] = '\0';
6245 return lbp->len + chars_deleted;
6249 * Like readline_internal, above, but in addition try to match the
6250 * input line against relevant regular expressions and manage #line
6251 * directives.
6253 static void
6254 readline (lbp, stream)
6255 linebuffer *lbp;
6256 FILE *stream;
6258 long result;
6260 linecharno = charno; /* update global char number of line start */
6261 result = readline_internal (lbp, stream); /* read line */
6262 lineno += 1; /* increment global line number */
6263 charno += result; /* increment global char number */
6265 /* Honour #line directives. */
6266 if (!no_line_directive)
6268 static bool discard_until_line_directive;
6270 /* Check whether this is a #line directive. */
6271 if (result > 12 && strneq (lbp->buffer, "#line ", 6))
6273 int start, lno;
6275 if (DEBUG) start = 0; /* shut up the compiler */
6276 if (sscanf (lbp->buffer, "#line %d \"%n", &lno, &start) == 1)
6278 char *endp = lbp->buffer + start;
6280 assert (start > 0);
6281 while ((endp = etags_strchr (endp, '"')) != NULL
6282 && endp[-1] == '\\')
6283 endp++;
6284 if (endp != NULL)
6285 /* Ok, this is a real #line directive. Let's deal with it. */
6287 char *taggedabsname; /* absolute name of original file */
6288 char *taggedfname; /* name of original file as given */
6289 char *name; /* temp var */
6291 discard_until_line_directive = FALSE; /* found it */
6292 name = lbp->buffer + start;
6293 *endp = '\0';
6294 canonicalize_filename (name); /* for DOS */
6295 taggedabsname = absolute_filename (name, curfdp->infabsdir);
6296 if (filename_is_absolute (name)
6297 || filename_is_absolute (curfdp->infname))
6298 taggedfname = savestr (taggedabsname);
6299 else
6300 taggedfname = relative_filename (taggedabsname,tagfiledir);
6302 if (streq (curfdp->taggedfname, taggedfname))
6303 /* The #line directive is only a line number change. We
6304 deal with this afterwards. */
6305 free (taggedfname);
6306 else
6307 /* The tags following this #line directive should be
6308 attributed to taggedfname. In order to do this, set
6309 curfdp accordingly. */
6311 fdesc *fdp; /* file description pointer */
6313 /* Go look for a file description already set up for the
6314 file indicated in the #line directive. If there is
6315 one, use it from now until the next #line
6316 directive. */
6317 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
6318 if (streq (fdp->infname, curfdp->infname)
6319 && streq (fdp->taggedfname, taggedfname))
6320 /* If we remove the second test above (after the &&)
6321 then all entries pertaining to the same file are
6322 coalesced in the tags file. If we use it, then
6323 entries pertaining to the same file but generated
6324 from different files (via #line directives) will
6325 go into separate sections in the tags file. These
6326 alternatives look equivalent. The first one
6327 destroys some apparently useless information. */
6329 curfdp = fdp;
6330 free (taggedfname);
6331 break;
6333 /* Else, if we already tagged the real file, skip all
6334 input lines until the next #line directive. */
6335 if (fdp == NULL) /* not found */
6336 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
6337 if (streq (fdp->infabsname, taggedabsname))
6339 discard_until_line_directive = TRUE;
6340 free (taggedfname);
6341 break;
6343 /* Else create a new file description and use that from
6344 now on, until the next #line directive. */
6345 if (fdp == NULL) /* not found */
6347 fdp = fdhead;
6348 fdhead = xnew (1, fdesc);
6349 *fdhead = *curfdp; /* copy curr. file description */
6350 fdhead->next = fdp;
6351 fdhead->infname = savestr (curfdp->infname);
6352 fdhead->infabsname = savestr (curfdp->infabsname);
6353 fdhead->infabsdir = savestr (curfdp->infabsdir);
6354 fdhead->taggedfname = taggedfname;
6355 fdhead->usecharno = FALSE;
6356 fdhead->prop = NULL;
6357 fdhead->written = FALSE;
6358 curfdp = fdhead;
6361 free (taggedabsname);
6362 lineno = lno - 1;
6363 readline (lbp, stream);
6364 return;
6365 } /* if a real #line directive */
6366 } /* if #line is followed by a a number */
6367 } /* if line begins with "#line " */
6369 /* If we are here, no #line directive was found. */
6370 if (discard_until_line_directive)
6372 if (result > 0)
6374 /* Do a tail recursion on ourselves, thus discarding the contents
6375 of the line buffer. */
6376 readline (lbp, stream);
6377 return;
6379 /* End of file. */
6380 discard_until_line_directive = FALSE;
6381 return;
6383 } /* if #line directives should be considered */
6385 #ifdef ETAGS_REGEXPS
6387 int match;
6388 regexp *rp;
6389 char *name;
6391 /* Match against relevant regexps. */
6392 if (lbp->len > 0)
6393 for (rp = p_head; rp != NULL; rp = rp->p_next)
6395 /* Only use generic regexps or those for the current language.
6396 Also do not use multiline regexps, which is the job of
6397 regex_tag_multiline. */
6398 if ((rp->lang != NULL && rp->lang != fdhead->lang)
6399 || rp->multi_line)
6400 continue;
6402 match = re_match (rp->pat, lbp->buffer, lbp->len, 0, &rp->regs);
6403 switch (match)
6405 case -2:
6406 /* Some error. */
6407 if (!rp->error_signaled)
6409 error ("regexp stack overflow while matching \"%s\"",
6410 rp->pattern);
6411 rp->error_signaled = TRUE;
6413 break;
6414 case -1:
6415 /* No match. */
6416 break;
6417 case 0:
6418 /* Empty string matched. */
6419 if (!rp->error_signaled)
6421 error ("regexp matches the empty string: \"%s\"", rp->pattern);
6422 rp->error_signaled = TRUE;
6424 break;
6425 default:
6426 /* Match occurred. Construct a tag. */
6427 name = rp->name;
6428 if (name[0] == '\0')
6429 name = NULL;
6430 else /* make a named tag */
6431 name = substitute (lbp->buffer, rp->name, &rp->regs);
6432 if (rp->force_explicit_name)
6433 /* Force explicit tag name, if a name is there. */
6434 pfnote (name, TRUE, lbp->buffer, match, lineno, linecharno);
6435 else
6436 make_tag (name, strlen (name), TRUE,
6437 lbp->buffer, match, lineno, linecharno);
6438 break;
6442 #endif /* ETAGS_REGEXPS */
6447 * Return a pointer to a space of size strlen(cp)+1 allocated
6448 * with xnew where the string CP has been copied.
6450 static char *
6451 savestr (cp)
6452 char *cp;
6454 return savenstr (cp, strlen (cp));
6458 * Return a pointer to a space of size LEN+1 allocated with xnew where
6459 * the string CP has been copied for at most the first LEN characters.
6461 static char *
6462 savenstr (cp, len)
6463 char *cp;
6464 int len;
6466 register char *dp;
6468 dp = xnew (len + 1, char);
6469 strncpy (dp, cp, len);
6470 dp[len] = '\0';
6471 return dp;
6475 * Return the ptr in sp at which the character c last
6476 * appears; NULL if not found
6478 * Identical to POSIX strrchr, included for portability.
6480 static char *
6481 etags_strrchr (sp, c)
6482 register const char *sp;
6483 register int c;
6485 register const char *r;
6487 r = NULL;
6490 if (*sp == c)
6491 r = sp;
6492 } while (*sp++);
6493 return (char *)r;
6497 * Return the ptr in sp at which the character c first
6498 * appears; NULL if not found
6500 * Identical to POSIX strchr, included for portability.
6502 static char *
6503 etags_strchr (sp, c)
6504 register const char *sp;
6505 register int c;
6509 if (*sp == c)
6510 return (char *)sp;
6511 } while (*sp++);
6512 return NULL;
6516 * Compare two strings, ignoring case for alphabetic characters.
6518 * Same as BSD's strcasecmp, included for portability.
6520 static int
6521 etags_strcasecmp (s1, s2)
6522 register const char *s1;
6523 register const char *s2;
6525 while (*s1 != '\0'
6526 && (ISALPHA (*s1) && ISALPHA (*s2)
6527 ? lowcase (*s1) == lowcase (*s2)
6528 : *s1 == *s2))
6529 s1++, s2++;
6531 return (ISALPHA (*s1) && ISALPHA (*s2)
6532 ? lowcase (*s1) - lowcase (*s2)
6533 : *s1 - *s2);
6537 * Compare two strings, ignoring case for alphabetic characters.
6538 * Stop after a given number of characters
6540 * Same as BSD's strncasecmp, included for portability.
6542 static int
6543 etags_strncasecmp (s1, s2, n)
6544 register const char *s1;
6545 register const char *s2;
6546 register int n;
6548 while (*s1 != '\0' && n-- > 0
6549 && (ISALPHA (*s1) && ISALPHA (*s2)
6550 ? lowcase (*s1) == lowcase (*s2)
6551 : *s1 == *s2))
6552 s1++, s2++;
6554 if (n < 0)
6555 return 0;
6556 else
6557 return (ISALPHA (*s1) && ISALPHA (*s2)
6558 ? lowcase (*s1) - lowcase (*s2)
6559 : *s1 - *s2);
6562 /* Skip spaces (end of string is not space), return new pointer. */
6563 static char *
6564 skip_spaces (cp)
6565 char *cp;
6567 while (iswhite (*cp))
6568 cp++;
6569 return cp;
6572 /* Skip non spaces, except end of string, return new pointer. */
6573 static char *
6574 skip_non_spaces (cp)
6575 char *cp;
6577 while (*cp != '\0' && !iswhite (*cp))
6578 cp++;
6579 return cp;
6582 /* Print error message and exit. */
6583 void
6584 fatal (s1, s2)
6585 char *s1, *s2;
6587 error (s1, s2);
6588 exit (EXIT_FAILURE);
6591 static void
6592 pfatal (s1)
6593 char *s1;
6595 perror (s1);
6596 exit (EXIT_FAILURE);
6599 static void
6600 suggest_asking_for_help ()
6602 fprintf (stderr, "\tTry `%s %s' for a complete list of options.\n",
6603 progname, LONG_OPTIONS ? "--help" : "-h");
6604 exit (EXIT_FAILURE);
6607 /* Print error message. `s1' is printf control string, `s2' is arg for it. */
6608 static void
6609 error (s1, s2)
6610 const char *s1, *s2;
6612 fprintf (stderr, "%s: ", progname);
6613 fprintf (stderr, s1, s2);
6614 fprintf (stderr, "\n");
6617 /* Return a newly-allocated string whose contents
6618 concatenate those of s1, s2, s3. */
6619 static char *
6620 concat (s1, s2, s3)
6621 char *s1, *s2, *s3;
6623 int len1 = strlen (s1), len2 = strlen (s2), len3 = strlen (s3);
6624 char *result = xnew (len1 + len2 + len3 + 1, char);
6626 strcpy (result, s1);
6627 strcpy (result + len1, s2);
6628 strcpy (result + len1 + len2, s3);
6629 result[len1 + len2 + len3] = '\0';
6631 return result;
6635 /* Does the same work as the system V getcwd, but does not need to
6636 guess the buffer size in advance. */
6637 static char *
6638 etags_getcwd ()
6640 #ifdef HAVE_GETCWD
6641 int bufsize = 200;
6642 char *path = xnew (bufsize, char);
6644 while (getcwd (path, bufsize) == NULL)
6646 if (errno != ERANGE)
6647 pfatal ("getcwd");
6648 bufsize *= 2;
6649 free (path);
6650 path = xnew (bufsize, char);
6653 canonicalize_filename (path);
6654 return path;
6656 #else /* not HAVE_GETCWD */
6657 #if MSDOS
6659 char *p, path[MAXPATHLEN + 1]; /* Fixed size is safe on MSDOS. */
6661 getwd (path);
6663 for (p = path; *p != '\0'; p++)
6664 if (*p == '\\')
6665 *p = '/';
6666 else
6667 *p = lowcase (*p);
6669 return strdup (path);
6670 #else /* not MSDOS */
6671 linebuffer path;
6672 FILE *pipe;
6674 linebuffer_init (&path);
6675 pipe = (FILE *) popen ("pwd 2>/dev/null", "r");
6676 if (pipe == NULL || readline_internal (&path, pipe) == 0)
6677 pfatal ("pwd");
6678 pclose (pipe);
6680 return path.buffer;
6681 #endif /* not MSDOS */
6682 #endif /* not HAVE_GETCWD */
6685 /* Return a newly allocated string containing the file name of FILE
6686 relative to the absolute directory DIR (which should end with a slash). */
6687 static char *
6688 relative_filename (file, dir)
6689 char *file, *dir;
6691 char *fp, *dp, *afn, *res;
6692 int i;
6694 /* Find the common root of file and dir (with a trailing slash). */
6695 afn = absolute_filename (file, cwd);
6696 fp = afn;
6697 dp = dir;
6698 while (*fp++ == *dp++)
6699 continue;
6700 fp--, dp--; /* back to the first differing char */
6701 #ifdef DOS_NT
6702 if (fp == afn && afn[0] != '/') /* cannot build a relative name */
6703 return afn;
6704 #endif
6705 do /* look at the equal chars until '/' */
6706 fp--, dp--;
6707 while (*fp != '/');
6709 /* Build a sequence of "../" strings for the resulting relative file name. */
6710 i = 0;
6711 while ((dp = etags_strchr (dp + 1, '/')) != NULL)
6712 i += 1;
6713 res = xnew (3*i + strlen (fp + 1) + 1, char);
6714 res[0] = '\0';
6715 while (i-- > 0)
6716 strcat (res, "../");
6718 /* Add the file name relative to the common root of file and dir. */
6719 strcat (res, fp + 1);
6720 free (afn);
6722 return res;
6725 /* Return a newly allocated string containing the absolute file name
6726 of FILE given DIR (which should end with a slash). */
6727 static char *
6728 absolute_filename (file, dir)
6729 char *file, *dir;
6731 char *slashp, *cp, *res;
6733 if (filename_is_absolute (file))
6734 res = savestr (file);
6735 #ifdef DOS_NT
6736 /* We don't support non-absolute file names with a drive
6737 letter, like `d:NAME' (it's too much hassle). */
6738 else if (file[1] == ':')
6739 fatal ("%s: relative file names with drive letters not supported", file);
6740 #endif
6741 else
6742 res = concat (dir, file, "");
6744 /* Delete the "/dirname/.." and "/." substrings. */
6745 slashp = etags_strchr (res, '/');
6746 while (slashp != NULL && slashp[0] != '\0')
6748 if (slashp[1] == '.')
6750 if (slashp[2] == '.'
6751 && (slashp[3] == '/' || slashp[3] == '\0'))
6753 cp = slashp;
6755 cp--;
6756 while (cp >= res && !filename_is_absolute (cp));
6757 if (cp < res)
6758 cp = slashp; /* the absolute name begins with "/.." */
6759 #ifdef DOS_NT
6760 /* Under MSDOS and NT we get `d:/NAME' as absolute
6761 file name, so the luser could say `d:/../NAME'.
6762 We silently treat this as `d:/NAME'. */
6763 else if (cp[0] != '/')
6764 cp = slashp;
6765 #endif
6766 strcpy (cp, slashp + 3);
6767 slashp = cp;
6768 continue;
6770 else if (slashp[2] == '/' || slashp[2] == '\0')
6772 strcpy (slashp, slashp + 2);
6773 continue;
6777 slashp = etags_strchr (slashp + 1, '/');
6780 if (res[0] == '\0')
6781 return savestr ("/");
6782 else
6783 return res;
6786 /* Return a newly allocated string containing the absolute
6787 file name of dir where FILE resides given DIR (which should
6788 end with a slash). */
6789 static char *
6790 absolute_dirname (file, dir)
6791 char *file, *dir;
6793 char *slashp, *res;
6794 char save;
6796 canonicalize_filename (file);
6797 slashp = etags_strrchr (file, '/');
6798 if (slashp == NULL)
6799 return savestr (dir);
6800 save = slashp[1];
6801 slashp[1] = '\0';
6802 res = absolute_filename (file, dir);
6803 slashp[1] = save;
6805 return res;
6808 /* Whether the argument string is an absolute file name. The argument
6809 string must have been canonicalized with canonicalize_filename. */
6810 static bool
6811 filename_is_absolute (fn)
6812 char *fn;
6814 return (fn[0] == '/'
6815 #ifdef DOS_NT
6816 || (ISALPHA(fn[0]) && fn[1] == ':' && fn[2] == '/')
6817 #endif
6821 /* Translate backslashes into slashes. Works in place. */
6822 static void
6823 canonicalize_filename (fn)
6824 register char *fn;
6826 #ifdef DOS_NT
6827 /* Canonicalize drive letter case. */
6828 if (fn[0] != '\0' && fn[1] == ':' && ISLOWER (fn[0]))
6829 fn[0] = upcase (fn[0]);
6830 /* Convert backslashes to slashes. */
6831 for (; *fn != '\0'; fn++)
6832 if (*fn == '\\')
6833 *fn = '/';
6834 #else
6835 /* No action. */
6836 fn = NULL; /* shut up the compiler */
6837 #endif
6841 /* Initialize a linebuffer for use */
6842 static void
6843 linebuffer_init (lbp)
6844 linebuffer *lbp;
6846 lbp->size = (DEBUG) ? 3 : 200;
6847 lbp->buffer = xnew (lbp->size, char);
6848 lbp->buffer[0] = '\0';
6849 lbp->len = 0;
6852 /* Set the minimum size of a string contained in a linebuffer. */
6853 static void
6854 linebuffer_setlen (lbp, toksize)
6855 linebuffer *lbp;
6856 int toksize;
6858 while (lbp->size <= toksize)
6860 lbp->size *= 2;
6861 xrnew (lbp->buffer, lbp->size, char);
6863 lbp->len = toksize;
6866 /* Like malloc but get fatal error if memory is exhausted. */
6867 static PTR
6868 xmalloc (size)
6869 unsigned int size;
6871 PTR result = (PTR) malloc (size);
6872 if (result == NULL)
6873 fatal ("virtual memory exhausted", (char *)NULL);
6874 return result;
6877 static PTR
6878 xrealloc (ptr, size)
6879 char *ptr;
6880 unsigned int size;
6882 PTR result = (PTR) realloc (ptr, size);
6883 if (result == NULL)
6884 fatal ("virtual memory exhausted", (char *)NULL);
6885 return result;
6889 * Local Variables:
6890 * c-indentation-style: gnu
6891 * indent-tabs-mode: t
6892 * tab-width: 8
6893 * fill-column: 79
6894 * c-font-lock-extra-types: ("FILE" "bool" "language" "linebuffer" "fdesc" "node" "regexp")
6895 * End:
6898 /* arch-tag: 8a9b748d-390c-4922-99db-2eeefa921051
6899 (do not change this comment) */
6901 /* etags.c ends here */