* lisp/files-x.el (modify-file-local-variable): Obey commenting conventions.
[emacs.git] / lib-src / etags.c
blobb739e8f00c55b0d4315a96fa63ed8d14b99a0942
1 /* Tags file maker to go with GNU Emacs -*- coding: latin-1 -*-
3 Copyright (C) 1984 The Regents of the University of California
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions are
7 met:
8 1. Redistributions of source code must retain the above copyright
9 notice, this list of conditions and the following disclaimer.
10 2. Redistributions in binary form must reproduce the above copyright
11 notice, this list of conditions and the following disclaimer in the
12 documentation and/or other materials provided with the
13 distribution.
14 3. Neither the name of the University nor the names of its
15 contributors may be used to endorse or promote products derived
16 from this software without specific prior written permission.
18 THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS''
19 AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
20 THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
21 PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS
22 BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24 SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
25 BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
26 WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
27 OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN
28 IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
31 Copyright (C) 1984, 1987-1989, 1993-1995, 1998-2011
32 Free Software Foundation, Inc.
34 This file is not considered part of GNU Emacs.
36 This program is free software: you can redistribute it and/or modify
37 it under the terms of the GNU General Public License as published by
38 the Free Software Foundation, either version 3 of the License, or
39 (at your option) any later version.
41 This program is distributed in the hope that it will be useful,
42 but WITHOUT ANY WARRANTY; without even the implied warranty of
43 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
44 GNU General Public License for more details.
46 You should have received a copy of the GNU General Public License
47 along with this program. If not, see <http://www.gnu.org/licenses/>. */
50 /* NB To comply with the above BSD license, copyright information is
51 reproduced in etc/ETAGS.README. That file should be updated when the
52 above notices are.
54 To the best of our knowledge, this code was originally based on the
55 ctags.c distributed with BSD4.2, which was copyrighted by the
56 University of California, as described above. */
60 * Authors:
61 * 1983 Ctags originally by Ken Arnold.
62 * 1984 Fortran added by Jim Kleckner.
63 * 1984 Ed Pelegri-Llopart added C typedefs.
64 * 1985 Emacs TAGS format by Richard Stallman.
65 * 1989 Sam Kendall added C++.
66 * 1992 Joseph B. Wells improved C and C++ parsing.
67 * 1993 Francesco Potortì reorganized C and C++.
68 * 1994 Line-by-line regexp tags by Tom Tromey.
69 * 2001 Nested classes by Francesco Potortì (concept by Mykola Dzyuba).
70 * 2002 #line directives by Francesco Potortì.
72 * Francesco Potortì <pot@gnu.org> has maintained and improved it since 1993.
76 * If you want to add support for a new language, start by looking at the LUA
77 * language, which is the simplest. Alternatively, consider distributing etags
78 * together with a configuration file containing regexp definitions for etags.
81 char pot_etags_version[] = "@(#) pot revision number is 17.38.1.4";
83 #define TRUE 1
84 #define FALSE 0
86 #ifdef DEBUG
87 # undef DEBUG
88 # define DEBUG TRUE
89 #else
90 # define DEBUG FALSE
91 # define NDEBUG /* disable assert */
92 #endif
94 #ifdef HAVE_CONFIG_H
95 # include <config.h>
96 /* This is probably not necessary any more. On some systems, config.h
97 used to define static as nothing for the sake of unexec. We don't
98 want that here since we don't use unexec. None of these systems
99 are supported any more, but the idea is still mentioned in
100 etc/PROBLEMS. */
101 # undef static
102 # ifndef PTR /* for XEmacs */
103 # define PTR void *
104 # endif
105 #else /* no config.h */
106 # if defined (__STDC__) && (__STDC__ || defined (__SUNPRO_C))
107 # define PTR void * /* for generic pointers */
108 # else /* not standard C */
109 # define const /* remove const for old compilers' sake */
110 # define PTR long * /* don't use void* */
111 # endif
112 #endif /* !HAVE_CONFIG_H */
114 #ifndef _GNU_SOURCE
115 # define _GNU_SOURCE 1 /* enables some compiler checks on GNU */
116 #endif
118 /* WIN32_NATIVE is for XEmacs.
119 MSDOS, WINDOWSNT, DOS_NT are for Emacs. */
120 #ifdef WIN32_NATIVE
121 # undef MSDOS
122 # undef WINDOWSNT
123 # define WINDOWSNT
124 #endif /* WIN32_NATIVE */
126 #ifdef MSDOS
127 # undef MSDOS
128 # define MSDOS TRUE
129 # include <fcntl.h>
130 # include <sys/param.h>
131 # include <io.h>
132 # ifndef HAVE_CONFIG_H
133 # define DOS_NT
134 # include <sys/config.h>
135 # endif
136 #else
137 # define MSDOS FALSE
138 #endif /* MSDOS */
140 #ifdef WINDOWSNT
141 # include <fcntl.h>
142 # include <direct.h>
143 # include <io.h>
144 # define MAXPATHLEN _MAX_PATH
145 # undef HAVE_NTGUI
146 # undef DOS_NT
147 # define DOS_NT
148 # ifndef HAVE_GETCWD
149 # define HAVE_GETCWD
150 # endif /* undef HAVE_GETCWD */
151 #else /* not WINDOWSNT */
152 #endif /* !WINDOWSNT */
154 #include <unistd.h>
155 #ifndef HAVE_UNISTD_H
156 # if defined (HAVE_GETCWD) && !defined (WINDOWSNT)
157 extern char *getcwd (char *buf, size_t size);
158 # endif
159 #endif /* HAVE_UNISTD_H */
161 #include <stdlib.h>
162 #include <string.h>
163 #include <stdio.h>
164 #include <ctype.h>
165 #include <errno.h>
166 #include <sys/types.h>
167 #include <sys/stat.h>
169 #include <assert.h>
170 #ifdef NDEBUG
171 # undef assert /* some systems have a buggy assert.h */
172 # define assert(x) ((void) 0)
173 #endif
175 #ifdef NO_LONG_OPTIONS /* define this if you don't have GNU getopt */
176 # define NO_LONG_OPTIONS TRUE
177 # define getopt_long(argc,argv,optstr,lopts,lind) getopt (argc, argv, optstr)
178 extern char *optarg;
179 extern int optind, opterr;
180 #else
181 # define NO_LONG_OPTIONS FALSE
182 # include <getopt.h>
183 #endif /* NO_LONG_OPTIONS */
185 #ifndef HAVE_CONFIG_H /* this is a standalone compilation */
186 # ifdef __CYGWIN__ /* compiling on Cygwin */
187 !!! NOTICE !!!
188 the regex.h distributed with Cygwin is not compatible with etags, alas!
189 If you want regular expression support, you should delete this notice and
190 arrange to use the GNU regex.h and regex.c.
191 # endif
192 #endif
193 #include <regex.h>
195 /* Define CTAGS to make the program "ctags" compatible with the usual one.
196 Leave it undefined to make the program "etags", which makes emacs-style
197 tag tables and tags typedefs, #defines and struct/union/enum by default. */
198 #ifdef CTAGS
199 # undef CTAGS
200 # define CTAGS TRUE
201 #else
202 # define CTAGS FALSE
203 #endif
205 #define streq(s,t) (assert ((s)!=NULL || (t)!=NULL), !strcmp (s, t))
206 #define strcaseeq(s,t) (assert ((s)!=NULL && (t)!=NULL), !etags_strcasecmp (s, t))
207 #define strneq(s,t,n) (assert ((s)!=NULL || (t)!=NULL), !strncmp (s, t, n))
208 #define strncaseeq(s,t,n) (assert ((s)!=NULL && (t)!=NULL), !etags_strncasecmp (s, t, n))
210 #define CHARS 256 /* 2^sizeof(char) */
211 #define CHAR(x) ((unsigned int)(x) & (CHARS - 1))
212 #define iswhite(c) (_wht[CHAR (c)]) /* c is white (see white) */
213 #define notinname(c) (_nin[CHAR (c)]) /* c is not in a name (see nonam) */
214 #define begtoken(c) (_btk[CHAR (c)]) /* c can start token (see begtk) */
215 #define intoken(c) (_itk[CHAR (c)]) /* c can be in token (see midtk) */
216 #define endtoken(c) (_etk[CHAR (c)]) /* c ends tokens (see endtk) */
218 #define ISALNUM(c) isalnum (CHAR (c))
219 #define ISALPHA(c) isalpha (CHAR (c))
220 #define ISDIGIT(c) isdigit (CHAR (c))
221 #define ISLOWER(c) islower (CHAR (c))
223 #define lowcase(c) tolower (CHAR (c))
227 * xnew, xrnew -- allocate, reallocate storage
229 * SYNOPSIS: Type *xnew (int n, Type);
230 * void xrnew (OldPointer, int n, Type);
232 #if DEBUG
233 # include "chkmalloc.h"
234 # define xnew(n,Type) ((Type *) trace_malloc (__FILE__, __LINE__, \
235 (n) * sizeof (Type)))
236 # define xrnew(op,n,Type) ((op) = (Type *) trace_realloc (__FILE__, __LINE__, \
237 (char *) (op), (n) * sizeof (Type)))
238 #else
239 # define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type)))
240 # define xrnew(op,n,Type) ((op) = (Type *) xrealloc ( \
241 (char *) (op), (n) * sizeof (Type)))
242 #endif
244 #define bool int
246 typedef void Lang_function (FILE *);
248 typedef struct
250 const char *suffix; /* file name suffix for this compressor */
251 const char *command; /* takes one arg and decompresses to stdout */
252 } compressor;
254 typedef struct
256 const char *name; /* language name */
257 const char *help; /* detailed help for the language */
258 Lang_function *function; /* parse function */
259 const char **suffixes; /* name suffixes of this language's files */
260 const char **filenames; /* names of this language's files */
261 const char **interpreters; /* interpreters for this language */
262 bool metasource; /* source used to generate other sources */
263 } language;
265 typedef struct fdesc
267 struct fdesc *next; /* for the linked list */
268 char *infname; /* uncompressed input file name */
269 char *infabsname; /* absolute uncompressed input file name */
270 char *infabsdir; /* absolute dir of input file */
271 char *taggedfname; /* file name to write in tagfile */
272 language *lang; /* language of file */
273 char *prop; /* file properties to write in tagfile */
274 bool usecharno; /* etags tags shall contain char number */
275 bool written; /* entry written in the tags file */
276 } fdesc;
278 typedef struct node_st
279 { /* sorting structure */
280 struct node_st *left, *right; /* left and right sons */
281 fdesc *fdp; /* description of file to whom tag belongs */
282 char *name; /* tag name */
283 char *regex; /* search regexp */
284 bool valid; /* write this tag on the tag file */
285 bool is_func; /* function tag: use regexp in CTAGS mode */
286 bool been_warned; /* warning already given for duplicated tag */
287 int lno; /* line number tag is on */
288 long cno; /* character number line starts on */
289 } node;
292 * A `linebuffer' is a structure which holds a line of text.
293 * `readline_internal' reads a line from a stream into a linebuffer
294 * and works regardless of the length of the line.
295 * SIZE is the size of BUFFER, LEN is the length of the string in
296 * BUFFER after readline reads it.
298 typedef struct
300 long size;
301 int len;
302 char *buffer;
303 } linebuffer;
305 /* Used to support mixing of --lang and file names. */
306 typedef struct
308 enum {
309 at_language, /* a language specification */
310 at_regexp, /* a regular expression */
311 at_filename, /* a file name */
312 at_stdin, /* read from stdin here */
313 at_end /* stop parsing the list */
314 } arg_type; /* argument type */
315 language *lang; /* language associated with the argument */
316 char *what; /* the argument itself */
317 } argument;
319 /* Structure defining a regular expression. */
320 typedef struct regexp
322 struct regexp *p_next; /* pointer to next in list */
323 language *lang; /* if set, use only for this language */
324 char *pattern; /* the regexp pattern */
325 char *name; /* tag name */
326 struct re_pattern_buffer *pat; /* the compiled pattern */
327 struct re_registers regs; /* re registers */
328 bool error_signaled; /* already signaled for this regexp */
329 bool force_explicit_name; /* do not allow implict tag name */
330 bool ignore_case; /* ignore case when matching */
331 bool multi_line; /* do a multi-line match on the whole file */
332 } regexp;
335 /* Many compilers barf on this:
336 Lang_function Ada_funcs;
337 so let's write it this way */
338 static void Ada_funcs (FILE *);
339 static void Asm_labels (FILE *);
340 static void C_entries (int c_ext, FILE *);
341 static void default_C_entries (FILE *);
342 static void plain_C_entries (FILE *);
343 static void Cjava_entries (FILE *);
344 static void Cobol_paragraphs (FILE *);
345 static void Cplusplus_entries (FILE *);
346 static void Cstar_entries (FILE *);
347 static void Erlang_functions (FILE *);
348 static void Forth_words (FILE *);
349 static void Fortran_functions (FILE *);
350 static void HTML_labels (FILE *);
351 static void Lisp_functions (FILE *);
352 static void Lua_functions (FILE *);
353 static void Makefile_targets (FILE *);
354 static void Pascal_functions (FILE *);
355 static void Perl_functions (FILE *);
356 static void PHP_functions (FILE *);
357 static void PS_functions (FILE *);
358 static void Prolog_functions (FILE *);
359 static void Python_functions (FILE *);
360 static void Scheme_functions (FILE *);
361 static void TeX_commands (FILE *);
362 static void Texinfo_nodes (FILE *);
363 static void Yacc_entries (FILE *);
364 static void just_read_file (FILE *);
366 static void print_language_names (void);
367 static void print_version (void);
368 static void print_help (argument *);
369 int main (int, char **);
371 static compressor *get_compressor_from_suffix (char *, char **);
372 static language *get_language_from_langname (const char *);
373 static language *get_language_from_interpreter (char *);
374 static language *get_language_from_filename (char *, bool);
375 static void readline (linebuffer *, FILE *);
376 static long readline_internal (linebuffer *, FILE *);
377 static bool nocase_tail (const char *);
378 static void get_tag (char *, char **);
380 static void analyse_regex (char *);
381 static void free_regexps (void);
382 static void regex_tag_multiline (void);
383 static void error (const char *, const char *);
384 static void suggest_asking_for_help (void) NO_RETURN;
385 void fatal (const char *, const char *) NO_RETURN;
386 static void pfatal (const char *) NO_RETURN;
387 static void add_node (node *, node **);
389 static void init (void);
390 static void process_file_name (char *, language *);
391 static void process_file (FILE *, char *, language *);
392 static void find_entries (FILE *);
393 static void free_tree (node *);
394 static void free_fdesc (fdesc *);
395 static void pfnote (char *, bool, char *, int, int, long);
396 static void make_tag (const char *, int, bool, char *, int, int, long);
397 static void invalidate_nodes (fdesc *, node **);
398 static void put_entries (node *);
400 static char *concat (const char *, const char *, const char *);
401 static char *skip_spaces (char *);
402 static char *skip_non_spaces (char *);
403 static char *savenstr (const char *, int);
404 static char *savestr (const char *);
405 static char *etags_strchr (const char *, int);
406 static char *etags_strrchr (const char *, int);
407 static int etags_strcasecmp (const char *, const char *);
408 static int etags_strncasecmp (const char *, const char *, int);
409 static char *etags_getcwd (void);
410 static char *relative_filename (char *, char *);
411 static char *absolute_filename (char *, char *);
412 static char *absolute_dirname (char *, char *);
413 static bool filename_is_absolute (char *f);
414 static void canonicalize_filename (char *);
415 static void linebuffer_init (linebuffer *);
416 static void linebuffer_setlen (linebuffer *, int);
417 static PTR xmalloc (size_t);
418 static PTR xrealloc (char *, size_t);
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 */
428 static ptrdiff_t whatlen_max; /* maximum length of any 'what' member */
430 static fdesc *fdhead; /* head of file description list */
431 static fdesc *curfdp; /* current file description */
432 static int lineno; /* line number of current line */
433 static long charno; /* current character number */
434 static long linecharno; /* charno of start of current line */
435 static char *dbp; /* pointer to start of current tag */
437 static const int invalidcharno = -1;
439 static node *nodehead; /* the head of the binary tree of tags */
440 static node *last_node; /* the last node created */
442 static linebuffer lb; /* the current line */
443 static linebuffer filebuf; /* a buffer containing the whole file */
444 static linebuffer token_name; /* a buffer containing a tag name */
446 /* boolean "functions" (see init) */
447 static bool _wht[CHARS], _nin[CHARS], _itk[CHARS], _btk[CHARS], _etk[CHARS];
448 static const char
449 /* white chars */
450 *white = " \f\t\n\r\v",
451 /* not in a name */
452 *nonam = " \f\t\n\r()=,;", /* look at make_tag before modifying! */
453 /* token ending chars */
454 *endtk = " \t\n\r\"'#()[]{}=-+%*/&|^~!<>;,.:?",
455 /* token starting chars */
456 *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~@",
457 /* valid in-token chars */
458 *midtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789";
460 static bool append_to_tagfile; /* -a: append to tags */
461 /* The next five default to TRUE in C and derived languages. */
462 static bool typedefs; /* -t: create tags for C and Ada typedefs */
463 static bool typedefs_or_cplusplus; /* -T: create tags for C typedefs, level */
464 /* 0 struct/enum/union decls, and C++ */
465 /* member functions. */
466 static bool constantypedefs; /* -d: create tags for C #define, enum */
467 /* constants and variables. */
468 /* -D: opposite of -d. Default under ctags. */
469 static bool globals; /* create tags for global variables */
470 static bool members; /* create tags for C member variables */
471 static bool declarations; /* --declarations: tag them and extern in C&Co*/
472 static bool no_line_directive; /* ignore #line directives (undocumented) */
473 static bool no_duplicates; /* no duplicate tags for ctags (undocumented) */
474 static bool update; /* -u: update tags */
475 static bool vgrind_style; /* -v: create vgrind style index output */
476 static bool no_warnings; /* -w: suppress warnings (undocumented) */
477 static bool cxref_style; /* -x: create cxref style output */
478 static bool cplusplus; /* .[hc] means C++, not C (undocumented) */
479 static bool ignoreindent; /* -I: ignore indentation in C */
480 static bool packages_only; /* --packages-only: in Ada, only tag packages*/
482 /* STDIN is defined in LynxOS system headers */
483 #ifdef STDIN
484 # undef STDIN
485 #endif
487 #define STDIN 0x1001 /* returned by getopt_long on --parse-stdin */
488 static bool parsing_stdin; /* --parse-stdin used */
490 static regexp *p_head; /* list of all regexps */
491 static bool need_filebuf; /* some regexes are multi-line */
493 static struct option longopts[] =
495 { "append", no_argument, NULL, 'a' },
496 { "packages-only", no_argument, &packages_only, TRUE },
497 { "c++", no_argument, NULL, 'C' },
498 { "declarations", no_argument, &declarations, TRUE },
499 { "no-line-directive", no_argument, &no_line_directive, TRUE },
500 { "no-duplicates", no_argument, &no_duplicates, TRUE },
501 { "help", no_argument, NULL, 'h' },
502 { "help", no_argument, NULL, 'H' },
503 { "ignore-indentation", no_argument, NULL, 'I' },
504 { "language", required_argument, NULL, 'l' },
505 { "members", no_argument, &members, TRUE },
506 { "no-members", no_argument, &members, FALSE },
507 { "output", required_argument, NULL, 'o' },
508 { "regex", required_argument, NULL, 'r' },
509 { "no-regex", no_argument, NULL, 'R' },
510 { "ignore-case-regex", required_argument, NULL, 'c' },
511 { "parse-stdin", required_argument, NULL, STDIN },
512 { "version", no_argument, NULL, 'V' },
514 #if CTAGS /* Ctags options */
515 { "backward-search", no_argument, NULL, 'B' },
516 { "cxref", no_argument, NULL, 'x' },
517 { "defines", no_argument, NULL, 'd' },
518 { "globals", no_argument, &globals, TRUE },
519 { "typedefs", no_argument, NULL, 't' },
520 { "typedefs-and-c++", no_argument, NULL, 'T' },
521 { "update", no_argument, NULL, 'u' },
522 { "vgrind", no_argument, NULL, 'v' },
523 { "no-warn", no_argument, NULL, 'w' },
525 #else /* Etags options */
526 { "no-defines", no_argument, NULL, 'D' },
527 { "no-globals", no_argument, &globals, FALSE },
528 { "include", required_argument, NULL, 'i' },
529 #endif
530 { NULL }
533 static compressor compressors[] =
535 { "z", "gzip -d -c"},
536 { "Z", "gzip -d -c"},
537 { "gz", "gzip -d -c"},
538 { "GZ", "gzip -d -c"},
539 { "bz2", "bzip2 -d -c" },
540 { "xz", "xz -d -c" },
541 { NULL }
545 * Language stuff.
548 /* Ada code */
549 static const char *Ada_suffixes [] =
550 { "ads", "adb", "ada", NULL };
551 static const char Ada_help [] =
552 "In Ada code, functions, procedures, packages, tasks and types are\n\
553 tags. Use the `--packages-only' option to create tags for\n\
554 packages only.\n\
555 Ada tag names have suffixes indicating the type of entity:\n\
556 Entity type: Qualifier:\n\
557 ------------ ----------\n\
558 function /f\n\
559 procedure /p\n\
560 package spec /s\n\
561 package body /b\n\
562 type /t\n\
563 task /k\n\
564 Thus, `M-x find-tag <RET> bidule/b <RET>' will go directly to the\n\
565 body of the package `bidule', while `M-x find-tag <RET> bidule <RET>'\n\
566 will just search for any tag `bidule'.";
568 /* Assembly code */
569 static const char *Asm_suffixes [] =
570 { "a", /* Unix assembler */
571 "asm", /* Microcontroller assembly */
572 "def", /* BSO/Tasking definition includes */
573 "inc", /* Microcontroller include files */
574 "ins", /* Microcontroller include files */
575 "s", "sa", /* Unix assembler */
576 "S", /* cpp-processed Unix assembler */
577 "src", /* BSO/Tasking C compiler output */
578 NULL
580 static const char Asm_help [] =
581 "In assembler code, labels appearing at the beginning of a line,\n\
582 followed by a colon, are tags.";
585 /* Note that .c and .h can be considered C++, if the --c++ flag was
586 given, or if the `class' or `template' keywords are met inside the file.
587 That is why default_C_entries is called for these. */
588 static const char *default_C_suffixes [] =
589 { "c", "h", NULL };
590 #if CTAGS /* C help for Ctags */
591 static const char default_C_help [] =
592 "In C code, any C function is a tag. Use -t to tag typedefs.\n\
593 Use -T to tag definitions of `struct', `union' and `enum'.\n\
594 Use -d to tag `#define' macro definitions and `enum' constants.\n\
595 Use --globals to tag global variables.\n\
596 You can tag function declarations and external variables by\n\
597 using `--declarations', and struct members by using `--members'.";
598 #else /* C help for Etags */
599 static const char default_C_help [] =
600 "In C code, any C function or typedef is a tag, and so are\n\
601 definitions of `struct', `union' and `enum'. `#define' macro\n\
602 definitions and `enum' constants are tags unless you specify\n\
603 `--no-defines'. Global variables are tags unless you specify\n\
604 `--no-globals' and so are struct members unless you specify\n\
605 `--no-members'. Use of `--no-globals', `--no-defines' and\n\
606 `--no-members' can make the tags table file much smaller.\n\
607 You can tag function declarations and external variables by\n\
608 using `--declarations'.";
609 #endif /* C help for Ctags and Etags */
611 static const char *Cplusplus_suffixes [] =
612 { "C", "c++", "cc", "cpp", "cxx", "H", "h++", "hh", "hpp", "hxx",
613 "M", /* Objective C++ */
614 "pdb", /* PostScript with C syntax */
615 NULL };
616 static const char Cplusplus_help [] =
617 "In C++ code, all the tag constructs of C code are tagged. (Use\n\
618 --help --lang=c --lang=c++ for full help.)\n\
619 In addition to C tags, member functions are also recognized. Member\n\
620 variables are recognized unless you use the `--no-members' option.\n\
621 Tags for variables and functions in classes are named `CLASS::VARIABLE'\n\
622 and `CLASS::FUNCTION'. `operator' definitions have tag names like\n\
623 `operator+'.";
625 static const char *Cjava_suffixes [] =
626 { "java", NULL };
627 static char Cjava_help [] =
628 "In Java code, all the tags constructs of C and C++ code are\n\
629 tagged. (Use --help --lang=c --lang=c++ --lang=java for full help.)";
632 static const char *Cobol_suffixes [] =
633 { "COB", "cob", NULL };
634 static char Cobol_help [] =
635 "In Cobol code, tags are paragraph names; that is, any word\n\
636 starting in column 8 and followed by a period.";
638 static const char *Cstar_suffixes [] =
639 { "cs", "hs", NULL };
641 static const char *Erlang_suffixes [] =
642 { "erl", "hrl", NULL };
643 static const char Erlang_help [] =
644 "In Erlang code, the tags are the functions, records and macros\n\
645 defined in the file.";
647 const char *Forth_suffixes [] =
648 { "fth", "tok", NULL };
649 static const char Forth_help [] =
650 "In Forth code, tags are words defined by `:',\n\
651 constant, code, create, defer, value, variable, buffer:, field.";
653 static const char *Fortran_suffixes [] =
654 { "F", "f", "f90", "for", NULL };
655 static const char Fortran_help [] =
656 "In Fortran code, functions, subroutines and block data are tags.";
658 static const char *HTML_suffixes [] =
659 { "htm", "html", "shtml", NULL };
660 static const char HTML_help [] =
661 "In HTML input files, the tags are the `title' and the `h1', `h2',\n\
662 `h3' headers. Also, tags are `name=' in anchors and all\n\
663 occurrences of `id='.";
665 static const char *Lisp_suffixes [] =
666 { "cl", "clisp", "el", "l", "lisp", "LSP", "lsp", "ml", NULL };
667 static const char Lisp_help [] =
668 "In Lisp code, any function defined with `defun', any variable\n\
669 defined with `defvar' or `defconst', and in general the first\n\
670 argument of any expression that starts with `(def' in column zero\n\
671 is a tag.";
673 static const char *Lua_suffixes [] =
674 { "lua", "LUA", NULL };
675 static const char Lua_help [] =
676 "In Lua scripts, all functions are tags.";
678 static const char *Makefile_filenames [] =
679 { "Makefile", "makefile", "GNUMakefile", "Makefile.in", "Makefile.am", NULL};
680 static const char Makefile_help [] =
681 "In makefiles, targets are tags; additionally, variables are tags\n\
682 unless you specify `--no-globals'.";
684 static const char *Objc_suffixes [] =
685 { "lm", /* Objective lex file */
686 "m", /* Objective C file */
687 NULL };
688 static const char Objc_help [] =
689 "In Objective C code, tags include Objective C definitions for classes,\n\
690 class categories, methods and protocols. Tags for variables and\n\
691 functions in classes are named `CLASS::VARIABLE' and `CLASS::FUNCTION'.\n\
692 (Use --help --lang=c --lang=objc --lang=java for full help.)";
694 static const char *Pascal_suffixes [] =
695 { "p", "pas", NULL };
696 static const char Pascal_help [] =
697 "In Pascal code, the tags are the functions and procedures defined\n\
698 in the file.";
699 /* " // this is for working around an Emacs highlighting bug... */
701 static const char *Perl_suffixes [] =
702 { "pl", "pm", NULL };
703 static const char *Perl_interpreters [] =
704 { "perl", "@PERL@", NULL };
705 static const char Perl_help [] =
706 "In Perl code, the tags are the packages, subroutines and variables\n\
707 defined by the `package', `sub', `my' and `local' keywords. Use\n\
708 `--globals' if you want to tag global variables. Tags for\n\
709 subroutines are named `PACKAGE::SUB'. The name for subroutines\n\
710 defined in the default package is `main::SUB'.";
712 static const char *PHP_suffixes [] =
713 { "php", "php3", "php4", NULL };
714 static const char PHP_help [] =
715 "In PHP code, tags are functions, classes and defines. Unless you use\n\
716 the `--no-members' option, vars are tags too.";
718 static const char *plain_C_suffixes [] =
719 { "pc", /* Pro*C file */
720 NULL };
722 static const char *PS_suffixes [] =
723 { "ps", "psw", NULL }; /* .psw is for PSWrap */
724 static const char PS_help [] =
725 "In PostScript code, the tags are the functions.";
727 static const char *Prolog_suffixes [] =
728 { "prolog", NULL };
729 static const char Prolog_help [] =
730 "In Prolog code, tags are predicates and rules at the beginning of\n\
731 line.";
733 static const char *Python_suffixes [] =
734 { "py", NULL };
735 static const char Python_help [] =
736 "In Python code, `def' or `class' at the beginning of a line\n\
737 generate a tag.";
739 /* Can't do the `SCM' or `scm' prefix with a version number. */
740 static const char *Scheme_suffixes [] =
741 { "oak", "sch", "scheme", "SCM", "scm", "SM", "sm", "ss", "t", NULL };
742 static const char Scheme_help [] =
743 "In Scheme code, tags include anything defined with `def' or with a\n\
744 construct whose name starts with `def'. They also include\n\
745 variables set with `set!' at top level in the file.";
747 static const char *TeX_suffixes [] =
748 { "bib", "clo", "cls", "ltx", "sty", "TeX", "tex", NULL };
749 static const char TeX_help [] =
750 "In LaTeX text, the argument of any of the commands `\\chapter',\n\
751 `\\section', `\\subsection', `\\subsubsection', `\\eqno', `\\label',\n\
752 `\\ref', `\\cite', `\\bibitem', `\\part', `\\appendix', `\\entry',\n\
753 `\\index', `\\def', `\\newcommand', `\\renewcommand',\n\
754 `\\newenvironment' or `\\renewenvironment' is a tag.\n\
756 Other commands can be specified by setting the environment variable\n\
757 `TEXTAGS' to a colon-separated list like, for example,\n\
758 TEXTAGS=\"mycommand:myothercommand\".";
761 static const char *Texinfo_suffixes [] =
762 { "texi", "texinfo", "txi", NULL };
763 static const char Texinfo_help [] =
764 "for texinfo files, lines starting with @node are tagged.";
766 static const char *Yacc_suffixes [] =
767 { "y", "y++", "ym", "yxx", "yy", NULL }; /* .ym is Objective yacc file */
768 static const char Yacc_help [] =
769 "In Bison or Yacc input files, each rule defines as a tag the\n\
770 nonterminal it constructs. The portions of the file that contain\n\
771 C code are parsed as C code (use --help --lang=c --lang=yacc\n\
772 for full help).";
774 static const char auto_help [] =
775 "`auto' is not a real language, it indicates to use\n\
776 a default language for files base on file name suffix and file contents.";
778 static const char none_help [] =
779 "`none' is not a real language, it indicates to only do\n\
780 regexp processing on files.";
782 static const char no_lang_help [] =
783 "No detailed help available for this language.";
787 * Table of languages.
789 * It is ok for a given function to be listed under more than one
790 * name. I just didn't.
793 static language lang_names [] =
795 { "ada", Ada_help, Ada_funcs, Ada_suffixes },
796 { "asm", Asm_help, Asm_labels, Asm_suffixes },
797 { "c", default_C_help, default_C_entries, default_C_suffixes },
798 { "c++", Cplusplus_help, Cplusplus_entries, Cplusplus_suffixes },
799 { "c*", no_lang_help, Cstar_entries, Cstar_suffixes },
800 { "cobol", Cobol_help, Cobol_paragraphs, Cobol_suffixes },
801 { "erlang", Erlang_help, Erlang_functions, Erlang_suffixes },
802 { "forth", Forth_help, Forth_words, Forth_suffixes },
803 { "fortran", Fortran_help, Fortran_functions, Fortran_suffixes },
804 { "html", HTML_help, HTML_labels, HTML_suffixes },
805 { "java", Cjava_help, Cjava_entries, Cjava_suffixes },
806 { "lisp", Lisp_help, Lisp_functions, Lisp_suffixes },
807 { "lua", Lua_help, Lua_functions, Lua_suffixes },
808 { "makefile", Makefile_help,Makefile_targets,NULL,Makefile_filenames},
809 { "objc", Objc_help, plain_C_entries, Objc_suffixes },
810 { "pascal", Pascal_help, Pascal_functions, Pascal_suffixes },
811 { "perl",Perl_help,Perl_functions,Perl_suffixes,NULL,Perl_interpreters},
812 { "php", PHP_help, PHP_functions, PHP_suffixes },
813 { "postscript",PS_help, PS_functions, PS_suffixes },
814 { "proc", no_lang_help, plain_C_entries, plain_C_suffixes },
815 { "prolog", Prolog_help, Prolog_functions, Prolog_suffixes },
816 { "python", Python_help, Python_functions, Python_suffixes },
817 { "scheme", Scheme_help, Scheme_functions, Scheme_suffixes },
818 { "tex", TeX_help, TeX_commands, TeX_suffixes },
819 { "texinfo", Texinfo_help, Texinfo_nodes, Texinfo_suffixes },
820 { "yacc", Yacc_help,Yacc_entries,Yacc_suffixes,NULL,NULL,TRUE},
821 { "auto", auto_help }, /* default guessing scheme */
822 { "none", none_help, just_read_file }, /* regexp matching only */
823 { NULL } /* end of list */
827 static void
828 print_language_names (void)
830 language *lang;
831 const char **name, **ext;
833 puts ("\nThese are the currently supported languages, along with the\n\
834 default file names and dot suffixes:");
835 for (lang = lang_names; lang->name != NULL; lang++)
837 printf (" %-*s", 10, lang->name);
838 if (lang->filenames != NULL)
839 for (name = lang->filenames; *name != NULL; name++)
840 printf (" %s", *name);
841 if (lang->suffixes != NULL)
842 for (ext = lang->suffixes; *ext != NULL; ext++)
843 printf (" .%s", *ext);
844 puts ("");
846 puts ("where `auto' means use default language for files based on file\n\
847 name suffix, and `none' means only do regexp processing on files.\n\
848 If no language is specified and no matching suffix is found,\n\
849 the first line of the file is read for a sharp-bang (#!) sequence\n\
850 followed by the name of an interpreter. If no such sequence is found,\n\
851 Fortran is tried first; if no tags are found, C is tried next.\n\
852 When parsing any C file, a \"class\" or \"template\" keyword\n\
853 switches to C++.");
854 puts ("Compressed files are supported using gzip, bzip2, and xz.\n\
856 For detailed help on a given language use, for example,\n\
857 etags --help --lang=ada.");
860 #ifndef EMACS_NAME
861 # define EMACS_NAME "standalone"
862 #endif
863 #ifndef VERSION
864 # define VERSION "17.38.1.4"
865 #endif
866 static void
867 print_version (void)
869 /* Makes it easier to update automatically. */
870 char emacs_copyright[] = "Copyright (C) 2011 Free Software Foundation, Inc.";
872 printf ("%s (%s %s)\n", (CTAGS) ? "ctags" : "etags", EMACS_NAME, VERSION);
873 puts (emacs_copyright);
874 puts ("This program is distributed under the terms in ETAGS.README");
876 exit (EXIT_SUCCESS);
879 #ifndef PRINT_UNDOCUMENTED_OPTIONS_HELP
880 # define PRINT_UNDOCUMENTED_OPTIONS_HELP FALSE
881 #endif
883 static void
884 print_help (argument *argbuffer)
886 bool help_for_lang = FALSE;
888 for (; argbuffer->arg_type != at_end; argbuffer++)
889 if (argbuffer->arg_type == at_language)
891 if (help_for_lang)
892 puts ("");
893 puts (argbuffer->lang->help);
894 help_for_lang = TRUE;
897 if (help_for_lang)
898 exit (EXIT_SUCCESS);
900 printf ("Usage: %s [options] [[regex-option ...] file-name] ...\n\
902 These are the options accepted by %s.\n", progname, progname);
903 if (NO_LONG_OPTIONS)
904 puts ("WARNING: long option names do not work with this executable,\n\
905 as it is not linked with GNU getopt.");
906 else
907 puts ("You may use unambiguous abbreviations for the long option names.");
908 puts (" A - as file name means read names from stdin (one per line).\n\
909 Absolute names are stored in the output file as they are.\n\
910 Relative ones are stored relative to the output file's directory.\n");
912 puts ("-a, --append\n\
913 Append tag entries to existing tags file.");
915 puts ("--packages-only\n\
916 For Ada files, only generate tags for packages.");
918 if (CTAGS)
919 puts ("-B, --backward-search\n\
920 Write the search commands for the tag entries using '?', the\n\
921 backward-search command instead of '/', the forward-search command.");
923 /* This option is mostly obsolete, because etags can now automatically
924 detect C++. Retained for backward compatibility and for debugging and
925 experimentation. In principle, we could want to tag as C++ even
926 before any "class" or "template" keyword.
927 puts ("-C, --c++\n\
928 Treat files whose name suffix defaults to C language as C++ files.");
931 puts ("--declarations\n\
932 In C and derived languages, create tags for function declarations,");
933 if (CTAGS)
934 puts ("\tand create tags for extern variables if --globals is used.");
935 else
936 puts
937 ("\tand create tags for extern variables unless --no-globals is used.");
939 if (CTAGS)
940 puts ("-d, --defines\n\
941 Create tag entries for C #define constants and enum constants, too.");
942 else
943 puts ("-D, --no-defines\n\
944 Don't create tag entries for C #define constants and enum constants.\n\
945 This makes the tags file smaller.");
947 if (!CTAGS)
948 puts ("-i FILE, --include=FILE\n\
949 Include a note in tag file indicating that, when searching for\n\
950 a tag, one should also consult the tags file FILE after\n\
951 checking the current file.");
953 puts ("-l LANG, --language=LANG\n\
954 Force the following files to be considered as written in the\n\
955 named language up to the next --language=LANG option.");
957 if (CTAGS)
958 puts ("--globals\n\
959 Create tag entries for global variables in some languages.");
960 else
961 puts ("--no-globals\n\
962 Do not create tag entries for global variables in some\n\
963 languages. This makes the tags file smaller.");
965 if (PRINT_UNDOCUMENTED_OPTIONS_HELP)
966 puts ("--no-line-directive\n\
967 Ignore #line preprocessor directives in C and derived languages.");
969 if (CTAGS)
970 puts ("--members\n\
971 Create tag entries for members of structures in some languages.");
972 else
973 puts ("--no-members\n\
974 Do not create tag entries for members of structures\n\
975 in some languages.");
977 puts ("-r REGEXP, --regex=REGEXP or --regex=@regexfile\n\
978 Make a tag for each line matching a regular expression pattern\n\
979 in the following files. {LANGUAGE}REGEXP uses REGEXP for LANGUAGE\n\
980 files only. REGEXFILE is a file containing one REGEXP per line.\n\
981 REGEXP takes the form /TAGREGEXP/TAGNAME/MODS, where TAGNAME/ is\n\
982 optional. The TAGREGEXP pattern is anchored (as if preceded by ^).");
983 puts (" If TAGNAME/ is present, the tags created are named.\n\
984 For example Tcl named tags can be created with:\n\
985 --regex=\"/proc[ \\t]+\\([^ \\t]+\\)/\\1/.\".\n\
986 MODS are optional one-letter modifiers: `i' means to ignore case,\n\
987 `m' means to allow multi-line matches, `s' implies `m' and\n\
988 causes dot to match any character, including newline.");
990 puts ("-R, --no-regex\n\
991 Don't create tags from regexps for the following files.");
993 puts ("-I, --ignore-indentation\n\
994 In C and C++ do not assume that a closing brace in the first\n\
995 column is the final brace of a function or structure definition.");
997 puts ("-o FILE, --output=FILE\n\
998 Write the tags to FILE.");
1000 puts ("--parse-stdin=NAME\n\
1001 Read from standard input and record tags as belonging to file NAME.");
1003 if (CTAGS)
1005 puts ("-t, --typedefs\n\
1006 Generate tag entries for C and Ada typedefs.");
1007 puts ("-T, --typedefs-and-c++\n\
1008 Generate tag entries for C typedefs, C struct/enum/union tags,\n\
1009 and C++ member functions.");
1012 if (CTAGS)
1013 puts ("-u, --update\n\
1014 Update the tag entries for the given files, leaving tag\n\
1015 entries for other files in place. Currently, this is\n\
1016 implemented by deleting the existing entries for the given\n\
1017 files and then rewriting the new entries at the end of the\n\
1018 tags file. It is often faster to simply rebuild the entire\n\
1019 tag file than to use this.");
1021 if (CTAGS)
1023 puts ("-v, --vgrind\n\
1024 Print on the standard output an index of items intended for\n\
1025 human consumption, similar to the output of vgrind. The index\n\
1026 is sorted, and gives the page number of each item.");
1028 if (PRINT_UNDOCUMENTED_OPTIONS_HELP)
1029 puts ("-w, --no-duplicates\n\
1030 Do not create duplicate tag entries, for compatibility with\n\
1031 traditional ctags.");
1033 if (PRINT_UNDOCUMENTED_OPTIONS_HELP)
1034 puts ("-w, --no-warn\n\
1035 Suppress warning messages about duplicate tag entries.");
1037 puts ("-x, --cxref\n\
1038 Like --vgrind, but in the style of cxref, rather than vgrind.\n\
1039 The output uses line numbers instead of page numbers, but\n\
1040 beyond that the differences are cosmetic; try both to see\n\
1041 which you like.");
1044 puts ("-V, --version\n\
1045 Print the version of the program.\n\
1046 -h, --help\n\
1047 Print this help message.\n\
1048 Followed by one or more `--language' options prints detailed\n\
1049 help about tag generation for the specified languages.");
1051 print_language_names ();
1053 puts ("");
1054 puts ("Report bugs to bug-gnu-emacs@gnu.org");
1056 exit (EXIT_SUCCESS);
1061 main (int argc, char **argv)
1063 int i;
1064 unsigned int nincluded_files;
1065 char **included_files;
1066 argument *argbuffer;
1067 int current_arg, file_count;
1068 linebuffer filename_lb;
1069 bool help_asked = FALSE;
1070 ptrdiff_t len;
1071 char *optstring;
1072 int opt;
1075 #ifdef DOS_NT
1076 _fmode = O_BINARY; /* all of files are treated as binary files */
1077 #endif /* DOS_NT */
1079 progname = argv[0];
1080 nincluded_files = 0;
1081 included_files = xnew (argc, char *);
1082 current_arg = 0;
1083 file_count = 0;
1085 /* Allocate enough no matter what happens. Overkill, but each one
1086 is small. */
1087 argbuffer = xnew (argc, argument);
1090 * Always find typedefs and structure tags.
1091 * Also default to find macro constants, enum constants, struct
1092 * members and global variables. Do it for both etags and ctags.
1094 typedefs = typedefs_or_cplusplus = constantypedefs = TRUE;
1095 globals = members = TRUE;
1097 /* When the optstring begins with a '-' getopt_long does not rearrange the
1098 non-options arguments to be at the end, but leaves them alone. */
1099 optstring = concat (NO_LONG_OPTIONS ? "" : "-",
1100 "ac:Cf:Il:o:r:RSVhH",
1101 (CTAGS) ? "BxdtTuvw" : "Di:");
1103 while ((opt = getopt_long (argc, argv, optstring, longopts, NULL)) != EOF)
1104 switch (opt)
1106 case 0:
1107 /* If getopt returns 0, then it has already processed a
1108 long-named option. We should do nothing. */
1109 break;
1111 case 1:
1112 /* This means that a file name has been seen. Record it. */
1113 argbuffer[current_arg].arg_type = at_filename;
1114 argbuffer[current_arg].what = optarg;
1115 len = strlen (optarg);
1116 if (whatlen_max < len)
1117 whatlen_max = len;
1118 ++current_arg;
1119 ++file_count;
1120 break;
1122 case STDIN:
1123 /* Parse standard input. Idea by Vivek <vivek@etla.org>. */
1124 argbuffer[current_arg].arg_type = at_stdin;
1125 argbuffer[current_arg].what = optarg;
1126 len = strlen (optarg);
1127 if (whatlen_max < len)
1128 whatlen_max = len;
1129 ++current_arg;
1130 ++file_count;
1131 if (parsing_stdin)
1132 fatal ("cannot parse standard input more than once", (char *)NULL);
1133 parsing_stdin = TRUE;
1134 break;
1136 /* Common options. */
1137 case 'a': append_to_tagfile = TRUE; break;
1138 case 'C': cplusplus = TRUE; break;
1139 case 'f': /* for compatibility with old makefiles */
1140 case 'o':
1141 if (tagfile)
1143 error ("-o option may only be given once.", (char *)NULL);
1144 suggest_asking_for_help ();
1145 /* NOTREACHED */
1147 tagfile = optarg;
1148 break;
1149 case 'I':
1150 case 'S': /* for backward compatibility */
1151 ignoreindent = TRUE;
1152 break;
1153 case 'l':
1155 language *lang = get_language_from_langname (optarg);
1156 if (lang != NULL)
1158 argbuffer[current_arg].lang = lang;
1159 argbuffer[current_arg].arg_type = at_language;
1160 ++current_arg;
1163 break;
1164 case 'c':
1165 /* Backward compatibility: support obsolete --ignore-case-regexp. */
1166 optarg = concat (optarg, "i", ""); /* memory leak here */
1167 /* FALLTHRU */
1168 case 'r':
1169 argbuffer[current_arg].arg_type = at_regexp;
1170 argbuffer[current_arg].what = optarg;
1171 len = strlen (optarg);
1172 if (whatlen_max < len)
1173 whatlen_max = len;
1174 ++current_arg;
1175 break;
1176 case 'R':
1177 argbuffer[current_arg].arg_type = at_regexp;
1178 argbuffer[current_arg].what = NULL;
1179 ++current_arg;
1180 break;
1181 case 'V':
1182 print_version ();
1183 break;
1184 case 'h':
1185 case 'H':
1186 help_asked = TRUE;
1187 break;
1189 /* Etags options */
1190 case 'D': constantypedefs = FALSE; break;
1191 case 'i': included_files[nincluded_files++] = optarg; break;
1193 /* Ctags options. */
1194 case 'B': searchar = '?'; break;
1195 case 'd': constantypedefs = TRUE; break;
1196 case 't': typedefs = TRUE; break;
1197 case 'T': typedefs = typedefs_or_cplusplus = TRUE; break;
1198 case 'u': update = TRUE; break;
1199 case 'v': vgrind_style = TRUE; /*FALLTHRU*/
1200 case 'x': cxref_style = TRUE; break;
1201 case 'w': no_warnings = TRUE; break;
1202 default:
1203 suggest_asking_for_help ();
1204 /* NOTREACHED */
1207 /* No more options. Store the rest of arguments. */
1208 for (; optind < argc; optind++)
1210 argbuffer[current_arg].arg_type = at_filename;
1211 argbuffer[current_arg].what = argv[optind];
1212 len = strlen (argv[optind]);
1213 if (whatlen_max < len)
1214 whatlen_max = len;
1215 ++current_arg;
1216 ++file_count;
1219 argbuffer[current_arg].arg_type = at_end;
1221 if (help_asked)
1222 print_help (argbuffer);
1223 /* NOTREACHED */
1225 if (nincluded_files == 0 && file_count == 0)
1227 error ("no input files specified.", (char *)NULL);
1228 suggest_asking_for_help ();
1229 /* NOTREACHED */
1232 if (tagfile == NULL)
1233 tagfile = savestr (CTAGS ? "tags" : "TAGS");
1234 cwd = etags_getcwd (); /* the current working directory */
1235 if (cwd[strlen (cwd) - 1] != '/')
1237 char *oldcwd = cwd;
1238 cwd = concat (oldcwd, "/", "");
1239 free (oldcwd);
1242 /* Compute base directory for relative file names. */
1243 if (streq (tagfile, "-")
1244 || strneq (tagfile, "/dev/", 5))
1245 tagfiledir = cwd; /* relative file names are relative to cwd */
1246 else
1248 canonicalize_filename (tagfile);
1249 tagfiledir = absolute_dirname (tagfile, cwd);
1252 init (); /* set up boolean "functions" */
1254 linebuffer_init (&lb);
1255 linebuffer_init (&filename_lb);
1256 linebuffer_init (&filebuf);
1257 linebuffer_init (&token_name);
1259 if (!CTAGS)
1261 if (streq (tagfile, "-"))
1263 tagf = stdout;
1264 #ifdef DOS_NT
1265 /* Switch redirected `stdout' to binary mode (setting `_fmode'
1266 doesn't take effect until after `stdout' is already open). */
1267 if (!isatty (fileno (stdout)))
1268 setmode (fileno (stdout), O_BINARY);
1269 #endif /* DOS_NT */
1271 else
1272 tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1273 if (tagf == NULL)
1274 pfatal (tagfile);
1278 * Loop through files finding functions.
1280 for (i = 0; i < current_arg; i++)
1282 static language *lang; /* non-NULL if language is forced */
1283 char *this_file;
1285 switch (argbuffer[i].arg_type)
1287 case at_language:
1288 lang = argbuffer[i].lang;
1289 break;
1290 case at_regexp:
1291 analyse_regex (argbuffer[i].what);
1292 break;
1293 case at_filename:
1294 this_file = argbuffer[i].what;
1295 /* Input file named "-" means read file names from stdin
1296 (one per line) and use them. */
1297 if (streq (this_file, "-"))
1299 if (parsing_stdin)
1300 fatal ("cannot parse standard input AND read file names from it",
1301 (char *)NULL);
1302 while (readline_internal (&filename_lb, stdin) > 0)
1303 process_file_name (filename_lb.buffer, lang);
1305 else
1306 process_file_name (this_file, lang);
1307 break;
1308 case at_stdin:
1309 this_file = argbuffer[i].what;
1310 process_file (stdin, this_file, lang);
1311 break;
1315 free_regexps ();
1316 free (lb.buffer);
1317 free (filebuf.buffer);
1318 free (token_name.buffer);
1320 if (!CTAGS || cxref_style)
1322 /* Write the remaining tags to tagf (ETAGS) or stdout (CXREF). */
1323 put_entries (nodehead);
1324 free_tree (nodehead);
1325 nodehead = NULL;
1326 if (!CTAGS)
1328 fdesc *fdp;
1330 /* Output file entries that have no tags. */
1331 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
1332 if (!fdp->written)
1333 fprintf (tagf, "\f\n%s,0\n", fdp->taggedfname);
1335 while (nincluded_files-- > 0)
1336 fprintf (tagf, "\f\n%s,include\n", *included_files++);
1338 if (fclose (tagf) == EOF)
1339 pfatal (tagfile);
1342 exit (EXIT_SUCCESS);
1345 /* From here on, we are in (CTAGS && !cxref_style) */
1346 if (update)
1348 char *cmd =
1349 xmalloc (strlen (tagfile) + whatlen_max +
1350 sizeof "mv..OTAGS;fgrep -v '\t\t' OTAGS >;rm OTAGS");
1351 for (i = 0; i < current_arg; ++i)
1353 switch (argbuffer[i].arg_type)
1355 case at_filename:
1356 case at_stdin:
1357 break;
1358 default:
1359 continue; /* the for loop */
1361 strcpy (cmd, "mv ");
1362 strcat (cmd, tagfile);
1363 strcat (cmd, " OTAGS;fgrep -v '\t");
1364 strcat (cmd, argbuffer[i].what);
1365 strcat (cmd, "\t' OTAGS >");
1366 strcat (cmd, tagfile);
1367 strcat (cmd, ";rm OTAGS");
1368 if (system (cmd) != EXIT_SUCCESS)
1369 fatal ("failed to execute shell command", (char *)NULL);
1371 free (cmd);
1372 append_to_tagfile = TRUE;
1375 tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
1376 if (tagf == NULL)
1377 pfatal (tagfile);
1378 put_entries (nodehead); /* write all the tags (CTAGS) */
1379 free_tree (nodehead);
1380 nodehead = NULL;
1381 if (fclose (tagf) == EOF)
1382 pfatal (tagfile);
1384 if (CTAGS)
1385 if (append_to_tagfile || update)
1387 char *cmd = xmalloc (2 * strlen (tagfile) + sizeof "sort -u -o..");
1388 /* Maybe these should be used:
1389 setenv ("LC_COLLATE", "C", 1);
1390 setenv ("LC_ALL", "C", 1); */
1391 strcpy (cmd, "sort -u -o ");
1392 strcat (cmd, tagfile);
1393 strcat (cmd, " ");
1394 strcat (cmd, tagfile);
1395 exit (system (cmd));
1397 return EXIT_SUCCESS;
1402 * Return a compressor given the file name. If EXTPTR is non-zero,
1403 * return a pointer into FILE where the compressor-specific
1404 * extension begins. If no compressor is found, NULL is returned
1405 * and EXTPTR is not significant.
1406 * Idea by Vladimir Alexiev <vladimir@cs.ualberta.ca> (1998)
1408 static compressor *
1409 get_compressor_from_suffix (char *file, char **extptr)
1411 compressor *compr;
1412 char *slash, *suffix;
1414 /* File has been processed by canonicalize_filename,
1415 so we don't need to consider backslashes on DOS_NT. */
1416 slash = etags_strrchr (file, '/');
1417 suffix = etags_strrchr (file, '.');
1418 if (suffix == NULL || suffix < slash)
1419 return NULL;
1420 if (extptr != NULL)
1421 *extptr = suffix;
1422 suffix += 1;
1423 /* Let those poor souls who live with DOS 8+3 file name limits get
1424 some solace by treating foo.cgz as if it were foo.c.gz, etc.
1425 Only the first do loop is run if not MSDOS */
1428 for (compr = compressors; compr->suffix != NULL; compr++)
1429 if (streq (compr->suffix, suffix))
1430 return compr;
1431 if (!MSDOS)
1432 break; /* do it only once: not really a loop */
1433 if (extptr != NULL)
1434 *extptr = ++suffix;
1435 } while (*suffix != '\0');
1436 return NULL;
1442 * Return a language given the name.
1444 static language *
1445 get_language_from_langname (const char *name)
1447 language *lang;
1449 if (name == NULL)
1450 error ("empty language name", (char *)NULL);
1451 else
1453 for (lang = lang_names; lang->name != NULL; lang++)
1454 if (streq (name, lang->name))
1455 return lang;
1456 error ("unknown language \"%s\"", name);
1459 return NULL;
1464 * Return a language given the interpreter name.
1466 static language *
1467 get_language_from_interpreter (char *interpreter)
1469 language *lang;
1470 const char **iname;
1472 if (interpreter == NULL)
1473 return NULL;
1474 for (lang = lang_names; lang->name != NULL; lang++)
1475 if (lang->interpreters != NULL)
1476 for (iname = lang->interpreters; *iname != NULL; iname++)
1477 if (streq (*iname, interpreter))
1478 return lang;
1480 return NULL;
1486 * Return a language given the file name.
1488 static language *
1489 get_language_from_filename (char *file, int case_sensitive)
1491 language *lang;
1492 const char **name, **ext, *suffix;
1494 /* Try whole file name first. */
1495 for (lang = lang_names; lang->name != NULL; lang++)
1496 if (lang->filenames != NULL)
1497 for (name = lang->filenames; *name != NULL; name++)
1498 if ((case_sensitive)
1499 ? streq (*name, file)
1500 : strcaseeq (*name, file))
1501 return lang;
1503 /* If not found, try suffix after last dot. */
1504 suffix = etags_strrchr (file, '.');
1505 if (suffix == NULL)
1506 return NULL;
1507 suffix += 1;
1508 for (lang = lang_names; lang->name != NULL; lang++)
1509 if (lang->suffixes != NULL)
1510 for (ext = lang->suffixes; *ext != NULL; ext++)
1511 if ((case_sensitive)
1512 ? streq (*ext, suffix)
1513 : strcaseeq (*ext, suffix))
1514 return lang;
1515 return NULL;
1520 * This routine is called on each file argument.
1522 static void
1523 process_file_name (char *file, language *lang)
1525 struct stat stat_buf;
1526 FILE *inf;
1527 fdesc *fdp;
1528 compressor *compr;
1529 char *compressed_name, *uncompressed_name;
1530 char *ext, *real_name;
1531 int retval;
1533 canonicalize_filename (file);
1534 if (streq (file, tagfile) && !streq (tagfile, "-"))
1536 error ("skipping inclusion of %s in self.", file);
1537 return;
1539 if ((compr = get_compressor_from_suffix (file, &ext)) == NULL)
1541 compressed_name = NULL;
1542 real_name = uncompressed_name = savestr (file);
1544 else
1546 real_name = compressed_name = savestr (file);
1547 uncompressed_name = savenstr (file, ext - file);
1550 /* If the canonicalized uncompressed name
1551 has already been dealt with, skip it silently. */
1552 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
1554 assert (fdp->infname != NULL);
1555 if (streq (uncompressed_name, fdp->infname))
1556 goto cleanup;
1559 if (stat (real_name, &stat_buf) != 0)
1561 /* Reset real_name and try with a different name. */
1562 real_name = NULL;
1563 if (compressed_name != NULL) /* try with the given suffix */
1565 if (stat (uncompressed_name, &stat_buf) == 0)
1566 real_name = uncompressed_name;
1568 else /* try all possible suffixes */
1570 for (compr = compressors; compr->suffix != NULL; compr++)
1572 compressed_name = concat (file, ".", compr->suffix);
1573 if (stat (compressed_name, &stat_buf) != 0)
1575 if (MSDOS)
1577 char *suf = compressed_name + strlen (file);
1578 size_t suflen = strlen (compr->suffix) + 1;
1579 for ( ; suf[1]; suf++, suflen--)
1581 memmove (suf, suf + 1, suflen);
1582 if (stat (compressed_name, &stat_buf) == 0)
1584 real_name = compressed_name;
1585 break;
1588 if (real_name != NULL)
1589 break;
1590 } /* MSDOS */
1591 free (compressed_name);
1592 compressed_name = NULL;
1594 else
1596 real_name = compressed_name;
1597 break;
1601 if (real_name == NULL)
1603 perror (file);
1604 goto cleanup;
1606 } /* try with a different name */
1608 if (!S_ISREG (stat_buf.st_mode))
1610 error ("skipping %s: it is not a regular file.", real_name);
1611 goto cleanup;
1613 if (real_name == compressed_name)
1615 char *cmd = concat (compr->command, " ", real_name);
1616 inf = (FILE *) popen (cmd, "r");
1617 free (cmd);
1619 else
1620 inf = fopen (real_name, "r");
1621 if (inf == NULL)
1623 perror (real_name);
1624 goto cleanup;
1627 process_file (inf, uncompressed_name, lang);
1629 if (real_name == compressed_name)
1630 retval = pclose (inf);
1631 else
1632 retval = fclose (inf);
1633 if (retval < 0)
1634 pfatal (file);
1636 cleanup:
1637 free (compressed_name);
1638 free (uncompressed_name);
1639 last_node = NULL;
1640 curfdp = NULL;
1641 return;
1644 static void
1645 process_file (FILE *fh, char *fn, language *lang)
1647 static const fdesc emptyfdesc;
1648 fdesc *fdp;
1650 /* Create a new input file description entry. */
1651 fdp = xnew (1, fdesc);
1652 *fdp = emptyfdesc;
1653 fdp->next = fdhead;
1654 fdp->infname = savestr (fn);
1655 fdp->lang = lang;
1656 fdp->infabsname = absolute_filename (fn, cwd);
1657 fdp->infabsdir = absolute_dirname (fn, cwd);
1658 if (filename_is_absolute (fn))
1660 /* An absolute file name. Canonicalize it. */
1661 fdp->taggedfname = absolute_filename (fn, NULL);
1663 else
1665 /* A file name relative to cwd. Make it relative
1666 to the directory of the tags file. */
1667 fdp->taggedfname = relative_filename (fn, tagfiledir);
1669 fdp->usecharno = TRUE; /* use char position when making tags */
1670 fdp->prop = NULL;
1671 fdp->written = FALSE; /* not written on tags file yet */
1673 fdhead = fdp;
1674 curfdp = fdhead; /* the current file description */
1676 find_entries (fh);
1678 /* If not Ctags, and if this is not metasource and if it contained no #line
1679 directives, we can write the tags and free all nodes pointing to
1680 curfdp. */
1681 if (!CTAGS
1682 && curfdp->usecharno /* no #line directives in this file */
1683 && !curfdp->lang->metasource)
1685 node *np, *prev;
1687 /* Look for the head of the sublist relative to this file. See add_node
1688 for the structure of the node tree. */
1689 prev = NULL;
1690 for (np = nodehead; np != NULL; prev = np, np = np->left)
1691 if (np->fdp == curfdp)
1692 break;
1694 /* If we generated tags for this file, write and delete them. */
1695 if (np != NULL)
1697 /* This is the head of the last sublist, if any. The following
1698 instructions depend on this being true. */
1699 assert (np->left == NULL);
1701 assert (fdhead == curfdp);
1702 assert (last_node->fdp == curfdp);
1703 put_entries (np); /* write tags for file curfdp->taggedfname */
1704 free_tree (np); /* remove the written nodes */
1705 if (prev == NULL)
1706 nodehead = NULL; /* no nodes left */
1707 else
1708 prev->left = NULL; /* delete the pointer to the sublist */
1714 * This routine sets up the boolean pseudo-functions which work
1715 * by setting boolean flags dependent upon the corresponding character.
1716 * Every char which is NOT in that string is not a white char. Therefore,
1717 * all of the array "_wht" is set to FALSE, and then the elements
1718 * subscripted by the chars in "white" are set to TRUE. Thus "_wht"
1719 * of a char is TRUE if it is the string "white", else FALSE.
1721 static void
1722 init (void)
1724 register const char *sp;
1725 register int i;
1727 for (i = 0; i < CHARS; i++)
1728 iswhite (i) = notinname (i) = begtoken (i) = intoken (i) = endtoken (i) = FALSE;
1729 for (sp = white; *sp != '\0'; sp++) iswhite (*sp) = TRUE;
1730 for (sp = nonam; *sp != '\0'; sp++) notinname (*sp) = TRUE;
1731 notinname ('\0') = notinname ('\n');
1732 for (sp = begtk; *sp != '\0'; sp++) begtoken (*sp) = TRUE;
1733 begtoken ('\0') = begtoken ('\n');
1734 for (sp = midtk; *sp != '\0'; sp++) intoken (*sp) = TRUE;
1735 intoken ('\0') = intoken ('\n');
1736 for (sp = endtk; *sp != '\0'; sp++) endtoken (*sp) = TRUE;
1737 endtoken ('\0') = endtoken ('\n');
1741 * This routine opens the specified file and calls the function
1742 * which finds the function and type definitions.
1744 static void
1745 find_entries (FILE *inf)
1747 char *cp;
1748 language *lang = curfdp->lang;
1749 Lang_function *parser = NULL;
1751 /* If user specified a language, use it. */
1752 if (lang != NULL && lang->function != NULL)
1754 parser = lang->function;
1757 /* Else try to guess the language given the file name. */
1758 if (parser == NULL)
1760 lang = get_language_from_filename (curfdp->infname, TRUE);
1761 if (lang != NULL && lang->function != NULL)
1763 curfdp->lang = lang;
1764 parser = lang->function;
1768 /* Else look for sharp-bang as the first two characters. */
1769 if (parser == NULL
1770 && readline_internal (&lb, inf) > 0
1771 && lb.len >= 2
1772 && lb.buffer[0] == '#'
1773 && lb.buffer[1] == '!')
1775 char *lp;
1777 /* Set lp to point at the first char after the last slash in the
1778 line or, if no slashes, at the first nonblank. Then set cp to
1779 the first successive blank and terminate the string. */
1780 lp = etags_strrchr (lb.buffer+2, '/');
1781 if (lp != NULL)
1782 lp += 1;
1783 else
1784 lp = skip_spaces (lb.buffer + 2);
1785 cp = skip_non_spaces (lp);
1786 *cp = '\0';
1788 if (strlen (lp) > 0)
1790 lang = get_language_from_interpreter (lp);
1791 if (lang != NULL && lang->function != NULL)
1793 curfdp->lang = lang;
1794 parser = lang->function;
1799 /* We rewind here, even if inf may be a pipe. We fail if the
1800 length of the first line is longer than the pipe block size,
1801 which is unlikely. */
1802 rewind (inf);
1804 /* Else try to guess the language given the case insensitive file name. */
1805 if (parser == NULL)
1807 lang = get_language_from_filename (curfdp->infname, FALSE);
1808 if (lang != NULL && lang->function != NULL)
1810 curfdp->lang = lang;
1811 parser = lang->function;
1815 /* Else try Fortran or C. */
1816 if (parser == NULL)
1818 node *old_last_node = last_node;
1820 curfdp->lang = get_language_from_langname ("fortran");
1821 find_entries (inf);
1823 if (old_last_node == last_node)
1824 /* No Fortran entries found. Try C. */
1826 /* We do not tag if rewind fails.
1827 Only the file name will be recorded in the tags file. */
1828 rewind (inf);
1829 curfdp->lang = get_language_from_langname (cplusplus ? "c++" : "c");
1830 find_entries (inf);
1832 return;
1835 if (!no_line_directive
1836 && curfdp->lang != NULL && curfdp->lang->metasource)
1837 /* It may be that this is a bingo.y file, and we already parsed a bingo.c
1838 file, or anyway we parsed a file that is automatically generated from
1839 this one. If this is the case, the bingo.c file contained #line
1840 directives that generated tags pointing to this file. Let's delete
1841 them all before parsing this file, which is the real source. */
1843 fdesc **fdpp = &fdhead;
1844 while (*fdpp != NULL)
1845 if (*fdpp != curfdp
1846 && streq ((*fdpp)->taggedfname, curfdp->taggedfname))
1847 /* We found one of those! We must delete both the file description
1848 and all tags referring to it. */
1850 fdesc *badfdp = *fdpp;
1852 /* Delete the tags referring to badfdp->taggedfname
1853 that were obtained from badfdp->infname. */
1854 invalidate_nodes (badfdp, &nodehead);
1856 *fdpp = badfdp->next; /* remove the bad description from the list */
1857 free_fdesc (badfdp);
1859 else
1860 fdpp = &(*fdpp)->next; /* advance the list pointer */
1863 assert (parser != NULL);
1865 /* Generic initialisations before reading from file. */
1866 linebuffer_setlen (&filebuf, 0); /* reset the file buffer */
1868 /* Generic initialisations before parsing file with readline. */
1869 lineno = 0; /* reset global line number */
1870 charno = 0; /* reset global char number */
1871 linecharno = 0; /* reset global char number of line start */
1873 parser (inf);
1875 regex_tag_multiline ();
1880 * Check whether an implicitly named tag should be created,
1881 * then call `pfnote'.
1882 * NAME is a string that is internally copied by this function.
1884 * TAGS format specification
1885 * Idea by Sam Kendall <kendall@mv.mv.com> (1997)
1886 * The following is explained in some more detail in etc/ETAGS.EBNF.
1888 * make_tag creates tags with "implicit tag names" (unnamed tags)
1889 * if the following are all true, assuming NONAM=" \f\t\n\r()=,;":
1890 * 1. NAME does not contain any of the characters in NONAM;
1891 * 2. LINESTART contains name as either a rightmost, or rightmost but
1892 * one character, substring;
1893 * 3. the character, if any, immediately before NAME in LINESTART must
1894 * be a character in NONAM;
1895 * 4. the character, if any, immediately after NAME in LINESTART must
1896 * also be a character in NONAM.
1898 * The implementation uses the notinname() macro, which recognises the
1899 * characters stored in the string `nonam'.
1900 * etags.el needs to use the same characters that are in NONAM.
1902 static void
1903 make_tag (const char *name, /* tag name, or NULL if unnamed */
1904 int namelen, /* tag length */
1905 int is_func, /* tag is a function */
1906 char *linestart, /* start of the line where tag is */
1907 int linelen, /* length of the line where tag is */
1908 int lno, /* line number */
1909 long int cno) /* character number */
1911 bool named = (name != NULL && namelen > 0);
1912 char *nname = NULL;
1914 if (!CTAGS && named) /* maybe set named to false */
1915 /* Let's try to make an implicit tag name, that is, create an unnamed tag
1916 such that etags.el can guess a name from it. */
1918 int i;
1919 register const char *cp = name;
1921 for (i = 0; i < namelen; i++)
1922 if (notinname (*cp++))
1923 break;
1924 if (i == namelen) /* rule #1 */
1926 cp = linestart + linelen - namelen;
1927 if (notinname (linestart[linelen-1]))
1928 cp -= 1; /* rule #4 */
1929 if (cp >= linestart /* rule #2 */
1930 && (cp == linestart
1931 || notinname (cp[-1])) /* rule #3 */
1932 && strneq (name, cp, namelen)) /* rule #2 */
1933 named = FALSE; /* use implicit tag name */
1937 if (named)
1938 nname = savenstr (name, namelen);
1940 pfnote (nname, is_func, linestart, linelen, lno, cno);
1943 /* Record a tag. */
1944 static void
1945 pfnote (char *name, int is_func, char *linestart, int linelen, int lno, long int cno)
1946 /* tag name, or NULL if unnamed */
1947 /* tag is a function */
1948 /* start of the line where tag is */
1949 /* length of the line where tag is */
1950 /* line number */
1951 /* character number */
1953 register node *np;
1955 assert (name == NULL || name[0] != '\0');
1956 if (CTAGS && name == NULL)
1957 return;
1959 np = xnew (1, node);
1961 /* If ctags mode, change name "main" to M<thisfilename>. */
1962 if (CTAGS && !cxref_style && streq (name, "main"))
1964 register char *fp = etags_strrchr (curfdp->taggedfname, '/');
1965 np->name = concat ("M", fp == NULL ? curfdp->taggedfname : fp + 1, "");
1966 fp = etags_strrchr (np->name, '.');
1967 if (fp != NULL && fp[1] != '\0' && fp[2] == '\0')
1968 fp[0] = '\0';
1970 else
1971 np->name = name;
1972 np->valid = TRUE;
1973 np->been_warned = FALSE;
1974 np->fdp = curfdp;
1975 np->is_func = is_func;
1976 np->lno = lno;
1977 if (np->fdp->usecharno)
1978 /* Our char numbers are 0-base, because of C language tradition?
1979 ctags compatibility? old versions compatibility? I don't know.
1980 Anyway, since emacs's are 1-base we expect etags.el to take care
1981 of the difference. If we wanted to have 1-based numbers, we would
1982 uncomment the +1 below. */
1983 np->cno = cno /* + 1 */ ;
1984 else
1985 np->cno = invalidcharno;
1986 np->left = np->right = NULL;
1987 if (CTAGS && !cxref_style)
1989 if (strlen (linestart) < 50)
1990 np->regex = concat (linestart, "$", "");
1991 else
1992 np->regex = savenstr (linestart, 50);
1994 else
1995 np->regex = savenstr (linestart, linelen);
1997 add_node (np, &nodehead);
2001 * free_tree ()
2002 * recurse on left children, iterate on right children.
2004 static void
2005 free_tree (register node *np)
2007 while (np)
2009 register node *node_right = np->right;
2010 free_tree (np->left);
2011 free (np->name);
2012 free (np->regex);
2013 free (np);
2014 np = node_right;
2019 * free_fdesc ()
2020 * delete a file description
2022 static void
2023 free_fdesc (register fdesc *fdp)
2025 free (fdp->infname);
2026 free (fdp->infabsname);
2027 free (fdp->infabsdir);
2028 free (fdp->taggedfname);
2029 free (fdp->prop);
2030 free (fdp);
2034 * add_node ()
2035 * Adds a node to the tree of nodes. In etags mode, sort by file
2036 * name. In ctags mode, sort by tag name. Make no attempt at
2037 * balancing.
2039 * add_node is the only function allowed to add nodes, so it can
2040 * maintain state.
2042 static void
2043 add_node (node *np, node **cur_node_p)
2045 register int dif;
2046 register node *cur_node = *cur_node_p;
2048 if (cur_node == NULL)
2050 *cur_node_p = np;
2051 last_node = np;
2052 return;
2055 if (!CTAGS)
2056 /* Etags Mode */
2058 /* For each file name, tags are in a linked sublist on the right
2059 pointer. The first tags of different files are a linked list
2060 on the left pointer. last_node points to the end of the last
2061 used sublist. */
2062 if (last_node != NULL && last_node->fdp == np->fdp)
2064 /* Let's use the same sublist as the last added node. */
2065 assert (last_node->right == NULL);
2066 last_node->right = np;
2067 last_node = np;
2069 else if (cur_node->fdp == np->fdp)
2071 /* Scanning the list we found the head of a sublist which is
2072 good for us. Let's scan this sublist. */
2073 add_node (np, &cur_node->right);
2075 else
2076 /* The head of this sublist is not good for us. Let's try the
2077 next one. */
2078 add_node (np, &cur_node->left);
2079 } /* if ETAGS mode */
2081 else
2083 /* Ctags Mode */
2084 dif = strcmp (np->name, cur_node->name);
2087 * If this tag name matches an existing one, then
2088 * do not add the node, but maybe print a warning.
2090 if (no_duplicates && !dif)
2092 if (np->fdp == cur_node->fdp)
2094 if (!no_warnings)
2096 fprintf (stderr, "Duplicate entry in file %s, line %d: %s\n",
2097 np->fdp->infname, lineno, np->name);
2098 fprintf (stderr, "Second entry ignored\n");
2101 else if (!cur_node->been_warned && !no_warnings)
2103 fprintf
2104 (stderr,
2105 "Duplicate entry in files %s and %s: %s (Warning only)\n",
2106 np->fdp->infname, cur_node->fdp->infname, np->name);
2107 cur_node->been_warned = TRUE;
2109 return;
2112 /* Actually add the node */
2113 add_node (np, dif < 0 ? &cur_node->left : &cur_node->right);
2114 } /* if CTAGS mode */
2118 * invalidate_nodes ()
2119 * Scan the node tree and invalidate all nodes pointing to the
2120 * given file description (CTAGS case) or free them (ETAGS case).
2122 static void
2123 invalidate_nodes (fdesc *badfdp, node **npp)
2125 node *np = *npp;
2127 if (np == NULL)
2128 return;
2130 if (CTAGS)
2132 if (np->left != NULL)
2133 invalidate_nodes (badfdp, &np->left);
2134 if (np->fdp == badfdp)
2135 np->valid = FALSE;
2136 if (np->right != NULL)
2137 invalidate_nodes (badfdp, &np->right);
2139 else
2141 assert (np->fdp != NULL);
2142 if (np->fdp == badfdp)
2144 *npp = np->left; /* detach the sublist from the list */
2145 np->left = NULL; /* isolate it */
2146 free_tree (np); /* free it */
2147 invalidate_nodes (badfdp, npp);
2149 else
2150 invalidate_nodes (badfdp, &np->left);
2155 static int total_size_of_entries (node *);
2156 static int number_len (long);
2158 /* Length of a non-negative number's decimal representation. */
2159 static int
2160 number_len (long int num)
2162 int len = 1;
2163 while ((num /= 10) > 0)
2164 len += 1;
2165 return len;
2169 * Return total number of characters that put_entries will output for
2170 * the nodes in the linked list at the right of the specified node.
2171 * This count is irrelevant with etags.el since emacs 19.34 at least,
2172 * but is still supplied for backward compatibility.
2174 static int
2175 total_size_of_entries (register node *np)
2177 register int total = 0;
2179 for (; np != NULL; np = np->right)
2180 if (np->valid)
2182 total += strlen (np->regex) + 1; /* pat\177 */
2183 if (np->name != NULL)
2184 total += strlen (np->name) + 1; /* name\001 */
2185 total += number_len ((long) np->lno) + 1; /* lno, */
2186 if (np->cno != invalidcharno) /* cno */
2187 total += number_len (np->cno);
2188 total += 1; /* newline */
2191 return total;
2194 static void
2195 put_entries (register node *np)
2197 register char *sp;
2198 static fdesc *fdp = NULL;
2200 if (np == NULL)
2201 return;
2203 /* Output subentries that precede this one */
2204 if (CTAGS)
2205 put_entries (np->left);
2207 /* Output this entry */
2208 if (np->valid)
2210 if (!CTAGS)
2212 /* Etags mode */
2213 if (fdp != np->fdp)
2215 fdp = np->fdp;
2216 fprintf (tagf, "\f\n%s,%d\n",
2217 fdp->taggedfname, total_size_of_entries (np));
2218 fdp->written = TRUE;
2220 fputs (np->regex, tagf);
2221 fputc ('\177', tagf);
2222 if (np->name != NULL)
2224 fputs (np->name, tagf);
2225 fputc ('\001', tagf);
2227 fprintf (tagf, "%d,", np->lno);
2228 if (np->cno != invalidcharno)
2229 fprintf (tagf, "%ld", np->cno);
2230 fputs ("\n", tagf);
2232 else
2234 /* Ctags mode */
2235 if (np->name == NULL)
2236 error ("internal error: NULL name in ctags mode.", (char *)NULL);
2238 if (cxref_style)
2240 if (vgrind_style)
2241 fprintf (stdout, "%s %s %d\n",
2242 np->name, np->fdp->taggedfname, (np->lno + 63) / 64);
2243 else
2244 fprintf (stdout, "%-16s %3d %-16s %s\n",
2245 np->name, np->lno, np->fdp->taggedfname, np->regex);
2247 else
2249 fprintf (tagf, "%s\t%s\t", np->name, np->fdp->taggedfname);
2251 if (np->is_func)
2252 { /* function or #define macro with args */
2253 putc (searchar, tagf);
2254 putc ('^', tagf);
2256 for (sp = np->regex; *sp; sp++)
2258 if (*sp == '\\' || *sp == searchar)
2259 putc ('\\', tagf);
2260 putc (*sp, tagf);
2262 putc (searchar, tagf);
2264 else
2265 { /* anything else; text pattern inadequate */
2266 fprintf (tagf, "%d", np->lno);
2268 putc ('\n', tagf);
2271 } /* if this node contains a valid tag */
2273 /* Output subentries that follow this one */
2274 put_entries (np->right);
2275 if (!CTAGS)
2276 put_entries (np->left);
2280 /* C extensions. */
2281 #define C_EXT 0x00fff /* C extensions */
2282 #define C_PLAIN 0x00000 /* C */
2283 #define C_PLPL 0x00001 /* C++ */
2284 #define C_STAR 0x00003 /* C* */
2285 #define C_JAVA 0x00005 /* JAVA */
2286 #define C_AUTO 0x01000 /* C, but switch to C++ if `class' is met */
2287 #define YACC 0x10000 /* yacc file */
2290 * The C symbol tables.
2292 enum sym_type
2294 st_none,
2295 st_C_objprot, st_C_objimpl, st_C_objend,
2296 st_C_gnumacro,
2297 st_C_ignore, st_C_attribute,
2298 st_C_javastruct,
2299 st_C_operator,
2300 st_C_class, st_C_template,
2301 st_C_struct, st_C_extern, st_C_enum, st_C_define, st_C_typedef
2304 static unsigned int hash (const char *, unsigned int);
2305 static struct C_stab_entry * in_word_set (const char *, unsigned int);
2306 static enum sym_type C_symtype (char *, int, int);
2308 /* Feed stuff between (but not including) %[ and %] lines to:
2309 gperf -m 5
2311 %compare-strncmp
2312 %enum
2313 %struct-type
2314 struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
2316 if, 0, st_C_ignore
2317 for, 0, st_C_ignore
2318 while, 0, st_C_ignore
2319 switch, 0, st_C_ignore
2320 return, 0, st_C_ignore
2321 __attribute__, 0, st_C_attribute
2322 GTY, 0, st_C_attribute
2323 @interface, 0, st_C_objprot
2324 @protocol, 0, st_C_objprot
2325 @implementation,0, st_C_objimpl
2326 @end, 0, st_C_objend
2327 import, (C_JAVA & ~C_PLPL), st_C_ignore
2328 package, (C_JAVA & ~C_PLPL), st_C_ignore
2329 friend, C_PLPL, st_C_ignore
2330 extends, (C_JAVA & ~C_PLPL), st_C_javastruct
2331 implements, (C_JAVA & ~C_PLPL), st_C_javastruct
2332 interface, (C_JAVA & ~C_PLPL), st_C_struct
2333 class, 0, st_C_class
2334 namespace, C_PLPL, st_C_struct
2335 domain, C_STAR, st_C_struct
2336 union, 0, st_C_struct
2337 struct, 0, st_C_struct
2338 extern, 0, st_C_extern
2339 enum, 0, st_C_enum
2340 typedef, 0, st_C_typedef
2341 define, 0, st_C_define
2342 undef, 0, st_C_define
2343 operator, C_PLPL, st_C_operator
2344 template, 0, st_C_template
2345 # DEFUN used in emacs, the next three used in glibc (SYSCALL only for mach).
2346 DEFUN, 0, st_C_gnumacro
2347 SYSCALL, 0, st_C_gnumacro
2348 ENTRY, 0, st_C_gnumacro
2349 PSEUDO, 0, st_C_gnumacro
2350 # These are defined inside C functions, so currently they are not met.
2351 # EXFUN used in glibc, DEFVAR_* in emacs.
2352 #EXFUN, 0, st_C_gnumacro
2353 #DEFVAR_, 0, st_C_gnumacro
2355 and replace lines between %< and %> with its output, then:
2356 - remove the #if characterset check
2357 - make in_word_set static and not inline. */
2358 /*%<*/
2359 /* C code produced by gperf version 3.0.1 */
2360 /* Command-line: gperf -m 5 */
2361 /* Computed positions: -k'2-3' */
2363 struct C_stab_entry { const char *name; int c_ext; enum sym_type type; };
2364 /* maximum key range = 33, duplicates = 0 */
2366 static inline unsigned int
2367 hash (register const char *str, register unsigned int len)
2369 static unsigned char asso_values[] =
2371 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2372 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2373 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2374 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2375 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2376 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2377 35, 35, 35, 35, 35, 35, 35, 35, 35, 3,
2378 26, 35, 35, 35, 35, 35, 35, 35, 27, 35,
2379 35, 35, 35, 24, 0, 35, 35, 35, 35, 0,
2380 35, 35, 35, 35, 35, 1, 35, 16, 35, 6,
2381 23, 0, 0, 35, 22, 0, 35, 35, 5, 0,
2382 0, 15, 1, 35, 6, 35, 8, 19, 35, 16,
2383 4, 5, 35, 35, 35, 35, 35, 35, 35, 35,
2384 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2385 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2386 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2387 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2388 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2389 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2390 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2391 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2392 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2393 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2394 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2395 35, 35, 35, 35, 35, 35, 35, 35, 35, 35,
2396 35, 35, 35, 35, 35, 35
2398 register int hval = len;
2400 switch (hval)
2402 default:
2403 hval += asso_values[(unsigned char)str[2]];
2404 /*FALLTHROUGH*/
2405 case 2:
2406 hval += asso_values[(unsigned char)str[1]];
2407 break;
2409 return hval;
2412 static struct C_stab_entry *
2413 in_word_set (register const char *str, register unsigned int len)
2415 enum
2417 TOTAL_KEYWORDS = 33,
2418 MIN_WORD_LENGTH = 2,
2419 MAX_WORD_LENGTH = 15,
2420 MIN_HASH_VALUE = 2,
2421 MAX_HASH_VALUE = 34
2424 static struct C_stab_entry wordlist[] =
2426 {""}, {""},
2427 {"if", 0, st_C_ignore},
2428 {"GTY", 0, st_C_attribute},
2429 {"@end", 0, st_C_objend},
2430 {"union", 0, st_C_struct},
2431 {"define", 0, st_C_define},
2432 {"import", (C_JAVA & ~C_PLPL), st_C_ignore},
2433 {"template", 0, st_C_template},
2434 {"operator", C_PLPL, st_C_operator},
2435 {"@interface", 0, st_C_objprot},
2436 {"implements", (C_JAVA & ~C_PLPL), st_C_javastruct},
2437 {"friend", C_PLPL, st_C_ignore},
2438 {"typedef", 0, st_C_typedef},
2439 {"return", 0, st_C_ignore},
2440 {"@implementation",0, st_C_objimpl},
2441 {"@protocol", 0, st_C_objprot},
2442 {"interface", (C_JAVA & ~C_PLPL), st_C_struct},
2443 {"extern", 0, st_C_extern},
2444 {"extends", (C_JAVA & ~C_PLPL), st_C_javastruct},
2445 {"struct", 0, st_C_struct},
2446 {"domain", C_STAR, st_C_struct},
2447 {"switch", 0, st_C_ignore},
2448 {"enum", 0, st_C_enum},
2449 {"for", 0, st_C_ignore},
2450 {"namespace", C_PLPL, st_C_struct},
2451 {"class", 0, st_C_class},
2452 {"while", 0, st_C_ignore},
2453 {"undef", 0, st_C_define},
2454 {"package", (C_JAVA & ~C_PLPL), st_C_ignore},
2455 {"__attribute__", 0, st_C_attribute},
2456 {"SYSCALL", 0, st_C_gnumacro},
2457 {"ENTRY", 0, st_C_gnumacro},
2458 {"PSEUDO", 0, st_C_gnumacro},
2459 {"DEFUN", 0, st_C_gnumacro}
2462 if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH)
2464 register int key = hash (str, len);
2466 if (key <= MAX_HASH_VALUE && key >= 0)
2468 register const char *s = wordlist[key].name;
2470 if (*str == *s && !strncmp (str + 1, s + 1, len - 1) && s[len] == '\0')
2471 return &wordlist[key];
2474 return 0;
2476 /*%>*/
2478 static enum sym_type
2479 C_symtype (char *str, int len, int c_ext)
2481 register struct C_stab_entry *se = in_word_set (str, len);
2483 if (se == NULL || (se->c_ext && !(c_ext & se->c_ext)))
2484 return st_none;
2485 return se->type;
2490 * Ignoring __attribute__ ((list))
2492 static bool inattribute; /* looking at an __attribute__ construct */
2495 * C functions and variables are recognized using a simple
2496 * finite automaton. fvdef is its state variable.
2498 static enum
2500 fvnone, /* nothing seen */
2501 fdefunkey, /* Emacs DEFUN keyword seen */
2502 fdefunname, /* Emacs DEFUN name seen */
2503 foperator, /* func: operator keyword seen (cplpl) */
2504 fvnameseen, /* function or variable name seen */
2505 fstartlist, /* func: just after open parenthesis */
2506 finlist, /* func: in parameter list */
2507 flistseen, /* func: after parameter list */
2508 fignore, /* func: before open brace */
2509 vignore /* var-like: ignore until ';' */
2510 } fvdef;
2512 static bool fvextern; /* func or var: extern keyword seen; */
2515 * typedefs are recognized using a simple finite automaton.
2516 * typdef is its state variable.
2518 static enum
2520 tnone, /* nothing seen */
2521 tkeyseen, /* typedef keyword seen */
2522 ttypeseen, /* defined type seen */
2523 tinbody, /* inside typedef body */
2524 tend, /* just before typedef tag */
2525 tignore /* junk after typedef tag */
2526 } typdef;
2529 * struct-like structures (enum, struct and union) are recognized
2530 * using another simple finite automaton. `structdef' is its state
2531 * variable.
2533 static enum
2535 snone, /* nothing seen yet,
2536 or in struct body if bracelev > 0 */
2537 skeyseen, /* struct-like keyword seen */
2538 stagseen, /* struct-like tag seen */
2539 scolonseen /* colon seen after struct-like tag */
2540 } structdef;
2543 * When objdef is different from onone, objtag is the name of the class.
2545 static const char *objtag = "<uninited>";
2548 * Yet another little state machine to deal with preprocessor lines.
2550 static enum
2552 dnone, /* nothing seen */
2553 dsharpseen, /* '#' seen as first char on line */
2554 ddefineseen, /* '#' and 'define' seen */
2555 dignorerest /* ignore rest of line */
2556 } definedef;
2559 * State machine for Objective C protocols and implementations.
2560 * Idea by Tom R.Hageman <tom@basil.icce.rug.nl> (1995)
2562 static enum
2564 onone, /* nothing seen */
2565 oprotocol, /* @interface or @protocol seen */
2566 oimplementation, /* @implementations seen */
2567 otagseen, /* class name seen */
2568 oparenseen, /* parenthesis before category seen */
2569 ocatseen, /* category name seen */
2570 oinbody, /* in @implementation body */
2571 omethodsign, /* in @implementation body, after +/- */
2572 omethodtag, /* after method name */
2573 omethodcolon, /* after method colon */
2574 omethodparm, /* after method parameter */
2575 oignore /* wait for @end */
2576 } objdef;
2580 * Use this structure to keep info about the token read, and how it
2581 * should be tagged. Used by the make_C_tag function to build a tag.
2583 static struct tok
2585 char *line; /* string containing the token */
2586 int offset; /* where the token starts in LINE */
2587 int length; /* token length */
2589 The previous members can be used to pass strings around for generic
2590 purposes. The following ones specifically refer to creating tags. In this
2591 case the token contained here is the pattern that will be used to create a
2592 tag.
2594 bool valid; /* do not create a tag; the token should be
2595 invalidated whenever a state machine is
2596 reset prematurely */
2597 bool named; /* create a named tag */
2598 int lineno; /* source line number of tag */
2599 long linepos; /* source char number of tag */
2600 } token; /* latest token read */
2603 * Variables and functions for dealing with nested structures.
2604 * Idea by Mykola Dzyuba <mdzyuba@yahoo.com> (2001)
2606 static void pushclass_above (int, char *, int);
2607 static void popclass_above (int);
2608 static void write_classname (linebuffer *, const char *qualifier);
2610 static struct {
2611 char **cname; /* nested class names */
2612 int *bracelev; /* nested class brace level */
2613 int nl; /* class nesting level (elements used) */
2614 int size; /* length of the array */
2615 } cstack; /* stack for nested declaration tags */
2616 /* Current struct nesting depth (namespace, class, struct, union, enum). */
2617 #define nestlev (cstack.nl)
2618 /* After struct keyword or in struct body, not inside a nested function. */
2619 #define instruct (structdef == snone && nestlev > 0 \
2620 && bracelev == cstack.bracelev[nestlev-1] + 1)
2622 static void
2623 pushclass_above (int bracelev, char *str, int len)
2625 int nl;
2627 popclass_above (bracelev);
2628 nl = cstack.nl;
2629 if (nl >= cstack.size)
2631 int size = cstack.size *= 2;
2632 xrnew (cstack.cname, size, char *);
2633 xrnew (cstack.bracelev, size, int);
2635 assert (nl == 0 || cstack.bracelev[nl-1] < bracelev);
2636 cstack.cname[nl] = (str == NULL) ? NULL : savenstr (str, len);
2637 cstack.bracelev[nl] = bracelev;
2638 cstack.nl = nl + 1;
2641 static void
2642 popclass_above (int bracelev)
2644 int nl;
2646 for (nl = cstack.nl - 1;
2647 nl >= 0 && cstack.bracelev[nl] >= bracelev;
2648 nl--)
2650 free (cstack.cname[nl]);
2651 cstack.nl = nl;
2655 static void
2656 write_classname (linebuffer *cn, const char *qualifier)
2658 int i, len;
2659 int qlen = strlen (qualifier);
2661 if (cstack.nl == 0 || cstack.cname[0] == NULL)
2663 len = 0;
2664 cn->len = 0;
2665 cn->buffer[0] = '\0';
2667 else
2669 len = strlen (cstack.cname[0]);
2670 linebuffer_setlen (cn, len);
2671 strcpy (cn->buffer, cstack.cname[0]);
2673 for (i = 1; i < cstack.nl; i++)
2675 char *s;
2676 int slen;
2678 s = cstack.cname[i];
2679 if (s == NULL)
2680 continue;
2681 slen = strlen (s);
2682 len += slen + qlen;
2683 linebuffer_setlen (cn, len);
2684 strncat (cn->buffer, qualifier, qlen);
2685 strncat (cn->buffer, s, slen);
2690 static bool consider_token (char *, int, int, int *, int, int, bool *);
2691 static void make_C_tag (bool);
2694 * consider_token ()
2695 * checks to see if the current token is at the start of a
2696 * function or variable, or corresponds to a typedef, or
2697 * is a struct/union/enum tag, or #define, or an enum constant.
2699 * *IS_FUNC gets TRUE if the token is a function or #define macro
2700 * with args. C_EXTP points to which language we are looking at.
2702 * Globals
2703 * fvdef IN OUT
2704 * structdef IN OUT
2705 * definedef IN OUT
2706 * typdef IN OUT
2707 * objdef IN OUT
2710 static bool
2711 consider_token (register char *str, register int len, register int c, int *c_extp, int bracelev, int parlev, int *is_func_or_var)
2712 /* IN: token pointer */
2713 /* IN: token length */
2714 /* IN: first char after the token */
2715 /* IN, OUT: C extensions mask */
2716 /* IN: brace level */
2717 /* IN: parenthesis level */
2718 /* OUT: function or variable found */
2720 /* When structdef is stagseen, scolonseen, or snone with bracelev > 0,
2721 structtype is the type of the preceding struct-like keyword, and
2722 structbracelev is the brace level where it has been seen. */
2723 static enum sym_type structtype;
2724 static int structbracelev;
2725 static enum sym_type toktype;
2728 toktype = C_symtype (str, len, *c_extp);
2731 * Skip __attribute__
2733 if (toktype == st_C_attribute)
2735 inattribute = TRUE;
2736 return FALSE;
2740 * Advance the definedef state machine.
2742 switch (definedef)
2744 case dnone:
2745 /* We're not on a preprocessor line. */
2746 if (toktype == st_C_gnumacro)
2748 fvdef = fdefunkey;
2749 return FALSE;
2751 break;
2752 case dsharpseen:
2753 if (toktype == st_C_define)
2755 definedef = ddefineseen;
2757 else
2759 definedef = dignorerest;
2761 return FALSE;
2762 case ddefineseen:
2764 * Make a tag for any macro, unless it is a constant
2765 * and constantypedefs is FALSE.
2767 definedef = dignorerest;
2768 *is_func_or_var = (c == '(');
2769 if (!*is_func_or_var && !constantypedefs)
2770 return FALSE;
2771 else
2772 return TRUE;
2773 case dignorerest:
2774 return FALSE;
2775 default:
2776 error ("internal error: definedef value.", (char *)NULL);
2780 * Now typedefs
2782 switch (typdef)
2784 case tnone:
2785 if (toktype == st_C_typedef)
2787 if (typedefs)
2788 typdef = tkeyseen;
2789 fvextern = FALSE;
2790 fvdef = fvnone;
2791 return FALSE;
2793 break;
2794 case tkeyseen:
2795 switch (toktype)
2797 case st_none:
2798 case st_C_class:
2799 case st_C_struct:
2800 case st_C_enum:
2801 typdef = ttypeseen;
2803 break;
2804 case ttypeseen:
2805 if (structdef == snone && fvdef == fvnone)
2807 fvdef = fvnameseen;
2808 return TRUE;
2810 break;
2811 case tend:
2812 switch (toktype)
2814 case st_C_class:
2815 case st_C_struct:
2816 case st_C_enum:
2817 return FALSE;
2819 return TRUE;
2822 switch (toktype)
2824 case st_C_javastruct:
2825 if (structdef == stagseen)
2826 structdef = scolonseen;
2827 return FALSE;
2828 case st_C_template:
2829 case st_C_class:
2830 if ((*c_extp & C_AUTO) /* automatic detection of C++ language */
2831 && bracelev == 0
2832 && definedef == dnone && structdef == snone
2833 && typdef == tnone && fvdef == fvnone)
2834 *c_extp = (*c_extp | C_PLPL) & ~C_AUTO;
2835 if (toktype == st_C_template)
2836 break;
2837 /* FALLTHRU */
2838 case st_C_struct:
2839 case st_C_enum:
2840 if (parlev == 0
2841 && fvdef != vignore
2842 && (typdef == tkeyseen
2843 || (typedefs_or_cplusplus && structdef == snone)))
2845 structdef = skeyseen;
2846 structtype = toktype;
2847 structbracelev = bracelev;
2848 if (fvdef == fvnameseen)
2849 fvdef = fvnone;
2851 return FALSE;
2854 if (structdef == skeyseen)
2856 structdef = stagseen;
2857 return TRUE;
2860 if (typdef != tnone)
2861 definedef = dnone;
2863 /* Detect Objective C constructs. */
2864 switch (objdef)
2866 case onone:
2867 switch (toktype)
2869 case st_C_objprot:
2870 objdef = oprotocol;
2871 return FALSE;
2872 case st_C_objimpl:
2873 objdef = oimplementation;
2874 return FALSE;
2876 break;
2877 case oimplementation:
2878 /* Save the class tag for functions or variables defined inside. */
2879 objtag = savenstr (str, len);
2880 objdef = oinbody;
2881 return FALSE;
2882 case oprotocol:
2883 /* Save the class tag for categories. */
2884 objtag = savenstr (str, len);
2885 objdef = otagseen;
2886 *is_func_or_var = TRUE;
2887 return TRUE;
2888 case oparenseen:
2889 objdef = ocatseen;
2890 *is_func_or_var = TRUE;
2891 return TRUE;
2892 case oinbody:
2893 break;
2894 case omethodsign:
2895 if (parlev == 0)
2897 fvdef = fvnone;
2898 objdef = omethodtag;
2899 linebuffer_setlen (&token_name, len);
2900 strncpy (token_name.buffer, str, len);
2901 token_name.buffer[len] = '\0';
2902 return TRUE;
2904 return FALSE;
2905 case omethodcolon:
2906 if (parlev == 0)
2907 objdef = omethodparm;
2908 return FALSE;
2909 case omethodparm:
2910 if (parlev == 0)
2912 fvdef = fvnone;
2913 objdef = omethodtag;
2914 linebuffer_setlen (&token_name, token_name.len + len);
2915 strncat (token_name.buffer, str, len);
2916 return TRUE;
2918 return FALSE;
2919 case oignore:
2920 if (toktype == st_C_objend)
2922 /* Memory leakage here: the string pointed by objtag is
2923 never released, because many tests would be needed to
2924 avoid breaking on incorrect input code. The amount of
2925 memory leaked here is the sum of the lengths of the
2926 class tags.
2927 free (objtag); */
2928 objdef = onone;
2930 return FALSE;
2933 /* A function, variable or enum constant? */
2934 switch (toktype)
2936 case st_C_extern:
2937 fvextern = TRUE;
2938 switch (fvdef)
2940 case finlist:
2941 case flistseen:
2942 case fignore:
2943 case vignore:
2944 break;
2945 default:
2946 fvdef = fvnone;
2948 return FALSE;
2949 case st_C_ignore:
2950 fvextern = FALSE;
2951 fvdef = vignore;
2952 return FALSE;
2953 case st_C_operator:
2954 fvdef = foperator;
2955 *is_func_or_var = TRUE;
2956 return TRUE;
2957 case st_none:
2958 if (constantypedefs
2959 && structdef == snone
2960 && structtype == st_C_enum && bracelev > structbracelev)
2961 return TRUE; /* enum constant */
2962 switch (fvdef)
2964 case fdefunkey:
2965 if (bracelev > 0)
2966 break;
2967 fvdef = fdefunname; /* GNU macro */
2968 *is_func_or_var = TRUE;
2969 return TRUE;
2970 case fvnone:
2971 switch (typdef)
2973 case ttypeseen:
2974 return FALSE;
2975 case tnone:
2976 if ((strneq (str, "asm", 3) && endtoken (str[3]))
2977 || (strneq (str, "__asm__", 7) && endtoken (str[7])))
2979 fvdef = vignore;
2980 return FALSE;
2982 break;
2984 /* FALLTHRU */
2985 case fvnameseen:
2986 if (len >= 10 && strneq (str+len-10, "::operator", 10))
2988 if (*c_extp & C_AUTO) /* automatic detection of C++ */
2989 *c_extp = (*c_extp | C_PLPL) & ~C_AUTO;
2990 fvdef = foperator;
2991 *is_func_or_var = TRUE;
2992 return TRUE;
2994 if (bracelev > 0 && !instruct)
2995 break;
2996 fvdef = fvnameseen; /* function or variable */
2997 *is_func_or_var = TRUE;
2998 return TRUE;
3000 break;
3003 return FALSE;
3008 * C_entries often keeps pointers to tokens or lines which are older than
3009 * the line currently read. By keeping two line buffers, and switching
3010 * them at end of line, it is possible to use those pointers.
3012 static struct
3014 long linepos;
3015 linebuffer lb;
3016 } lbs[2];
3018 #define current_lb_is_new (newndx == curndx)
3019 #define switch_line_buffers() (curndx = 1 - curndx)
3021 #define curlb (lbs[curndx].lb)
3022 #define newlb (lbs[newndx].lb)
3023 #define curlinepos (lbs[curndx].linepos)
3024 #define newlinepos (lbs[newndx].linepos)
3026 #define plainc ((c_ext & C_EXT) == C_PLAIN)
3027 #define cplpl (c_ext & C_PLPL)
3028 #define cjava ((c_ext & C_JAVA) == C_JAVA)
3030 #define CNL_SAVE_DEFINEDEF() \
3031 do { \
3032 curlinepos = charno; \
3033 readline (&curlb, inf); \
3034 lp = curlb.buffer; \
3035 quotednl = FALSE; \
3036 newndx = curndx; \
3037 } while (0)
3039 #define CNL() \
3040 do { \
3041 CNL_SAVE_DEFINEDEF(); \
3042 if (savetoken.valid) \
3044 token = savetoken; \
3045 savetoken.valid = FALSE; \
3047 definedef = dnone; \
3048 } while (0)
3051 static void
3052 make_C_tag (int isfun)
3054 /* This function is never called when token.valid is FALSE, but
3055 we must protect against invalid input or internal errors. */
3056 if (token.valid)
3057 make_tag (token_name.buffer, token_name.len, isfun, token.line,
3058 token.offset+token.length+1, token.lineno, token.linepos);
3059 else if (DEBUG)
3060 { /* this branch is optimised away if !DEBUG */
3061 make_tag (concat ("INVALID TOKEN:-->", token_name.buffer, ""),
3062 token_name.len + 17, isfun, token.line,
3063 token.offset+token.length+1, token.lineno, token.linepos);
3064 error ("INVALID TOKEN", NULL);
3067 token.valid = FALSE;
3072 * C_entries ()
3073 * This routine finds functions, variables, typedefs,
3074 * #define's, enum constants and struct/union/enum definitions in
3075 * C syntax and adds them to the list.
3077 static void
3078 C_entries (int c_ext, FILE *inf)
3079 /* extension of C */
3080 /* input file */
3082 register char c; /* latest char read; '\0' for end of line */
3083 register char *lp; /* pointer one beyond the character `c' */
3084 int curndx, newndx; /* indices for current and new lb */
3085 register int tokoff; /* offset in line of start of current token */
3086 register int toklen; /* length of current token */
3087 const char *qualifier; /* string used to qualify names */
3088 int qlen; /* length of qualifier */
3089 int bracelev; /* current brace level */
3090 int bracketlev; /* current bracket level */
3091 int parlev; /* current parenthesis level */
3092 int attrparlev; /* __attribute__ parenthesis level */
3093 int templatelev; /* current template level */
3094 int typdefbracelev; /* bracelev where a typedef struct body begun */
3095 bool incomm, inquote, inchar, quotednl, midtoken;
3096 bool yacc_rules; /* in the rules part of a yacc file */
3097 struct tok savetoken = {0}; /* token saved during preprocessor handling */
3100 linebuffer_init (&lbs[0].lb);
3101 linebuffer_init (&lbs[1].lb);
3102 if (cstack.size == 0)
3104 cstack.size = (DEBUG) ? 1 : 4;
3105 cstack.nl = 0;
3106 cstack.cname = xnew (cstack.size, char *);
3107 cstack.bracelev = xnew (cstack.size, int);
3110 tokoff = toklen = typdefbracelev = 0; /* keep compiler quiet */
3111 curndx = newndx = 0;
3112 lp = curlb.buffer;
3113 *lp = 0;
3115 fvdef = fvnone; fvextern = FALSE; typdef = tnone;
3116 structdef = snone; definedef = dnone; objdef = onone;
3117 yacc_rules = FALSE;
3118 midtoken = inquote = inchar = incomm = quotednl = FALSE;
3119 token.valid = savetoken.valid = FALSE;
3120 bracelev = bracketlev = parlev = attrparlev = templatelev = 0;
3121 if (cjava)
3122 { qualifier = "."; qlen = 1; }
3123 else
3124 { qualifier = "::"; qlen = 2; }
3127 while (!feof (inf))
3129 c = *lp++;
3130 if (c == '\\')
3132 /* If we are at the end of the line, the next character is a
3133 '\0'; do not skip it, because it is what tells us
3134 to read the next line. */
3135 if (*lp == '\0')
3137 quotednl = TRUE;
3138 continue;
3140 lp++;
3141 c = ' ';
3143 else if (incomm)
3145 switch (c)
3147 case '*':
3148 if (*lp == '/')
3150 c = *lp++;
3151 incomm = FALSE;
3153 break;
3154 case '\0':
3155 /* Newlines inside comments do not end macro definitions in
3156 traditional cpp. */
3157 CNL_SAVE_DEFINEDEF ();
3158 break;
3160 continue;
3162 else if (inquote)
3164 switch (c)
3166 case '"':
3167 inquote = FALSE;
3168 break;
3169 case '\0':
3170 /* Newlines inside strings do not end macro definitions
3171 in traditional cpp, even though compilers don't
3172 usually accept them. */
3173 CNL_SAVE_DEFINEDEF ();
3174 break;
3176 continue;
3178 else if (inchar)
3180 switch (c)
3182 case '\0':
3183 /* Hmmm, something went wrong. */
3184 CNL ();
3185 /* FALLTHRU */
3186 case '\'':
3187 inchar = FALSE;
3188 break;
3190 continue;
3192 else if (bracketlev > 0)
3194 switch (c)
3196 case ']':
3197 if (--bracketlev > 0)
3198 continue;
3199 break;
3200 case '\0':
3201 CNL_SAVE_DEFINEDEF ();
3202 break;
3204 continue;
3206 else switch (c)
3208 case '"':
3209 inquote = TRUE;
3210 if (inattribute)
3211 break;
3212 switch (fvdef)
3214 case fdefunkey:
3215 case fstartlist:
3216 case finlist:
3217 case fignore:
3218 case vignore:
3219 break;
3220 default:
3221 fvextern = FALSE;
3222 fvdef = fvnone;
3224 continue;
3225 case '\'':
3226 inchar = TRUE;
3227 if (inattribute)
3228 break;
3229 if (fvdef != finlist && fvdef != fignore && fvdef !=vignore)
3231 fvextern = FALSE;
3232 fvdef = fvnone;
3234 continue;
3235 case '/':
3236 if (*lp == '*')
3238 incomm = TRUE;
3239 lp++;
3240 c = ' ';
3242 else if (/* cplpl && */ *lp == '/')
3244 c = '\0';
3246 break;
3247 case '%':
3248 if ((c_ext & YACC) && *lp == '%')
3250 /* Entering or exiting rules section in yacc file. */
3251 lp++;
3252 definedef = dnone; fvdef = fvnone; fvextern = FALSE;
3253 typdef = tnone; structdef = snone;
3254 midtoken = inquote = inchar = incomm = quotednl = FALSE;
3255 bracelev = 0;
3256 yacc_rules = !yacc_rules;
3257 continue;
3259 else
3260 break;
3261 case '#':
3262 if (definedef == dnone)
3264 char *cp;
3265 bool cpptoken = TRUE;
3267 /* Look back on this line. If all blanks, or nonblanks
3268 followed by an end of comment, this is a preprocessor
3269 token. */
3270 for (cp = newlb.buffer; cp < lp-1; cp++)
3271 if (!iswhite (*cp))
3273 if (*cp == '*' && *(cp+1) == '/')
3275 cp++;
3276 cpptoken = TRUE;
3278 else
3279 cpptoken = FALSE;
3281 if (cpptoken)
3282 definedef = dsharpseen;
3283 } /* if (definedef == dnone) */
3284 continue;
3285 case '[':
3286 bracketlev++;
3287 continue;
3288 } /* switch (c) */
3291 /* Consider token only if some involved conditions are satisfied. */
3292 if (typdef != tignore
3293 && definedef != dignorerest
3294 && fvdef != finlist
3295 && templatelev == 0
3296 && (definedef != dnone
3297 || structdef != scolonseen)
3298 && !inattribute)
3300 if (midtoken)
3302 if (endtoken (c))
3304 if (c == ':' && *lp == ':' && begtoken (lp[1]))
3305 /* This handles :: in the middle,
3306 but not at the beginning of an identifier.
3307 Also, space-separated :: is not recognised. */
3309 if (c_ext & C_AUTO) /* automatic detection of C++ */
3310 c_ext = (c_ext | C_PLPL) & ~C_AUTO;
3311 lp += 2;
3312 toklen += 2;
3313 c = lp[-1];
3314 goto still_in_token;
3316 else
3318 bool funorvar = FALSE;
3320 if (yacc_rules
3321 || consider_token (newlb.buffer + tokoff, toklen, c,
3322 &c_ext, bracelev, parlev,
3323 &funorvar))
3325 if (fvdef == foperator)
3327 char *oldlp = lp;
3328 lp = skip_spaces (lp-1);
3329 if (*lp != '\0')
3330 lp += 1;
3331 while (*lp != '\0'
3332 && !iswhite (*lp) && *lp != '(')
3333 lp += 1;
3334 c = *lp++;
3335 toklen += lp - oldlp;
3337 token.named = FALSE;
3338 if (!plainc
3339 && nestlev > 0 && definedef == dnone)
3340 /* in struct body */
3342 write_classname (&token_name, qualifier);
3343 linebuffer_setlen (&token_name,
3344 token_name.len+qlen+toklen);
3345 strcat (token_name.buffer, qualifier);
3346 strncat (token_name.buffer,
3347 newlb.buffer + tokoff, toklen);
3348 token.named = TRUE;
3350 else if (objdef == ocatseen)
3351 /* Objective C category */
3353 int len = strlen (objtag) + 2 + toklen;
3354 linebuffer_setlen (&token_name, len);
3355 strcpy (token_name.buffer, objtag);
3356 strcat (token_name.buffer, "(");
3357 strncat (token_name.buffer,
3358 newlb.buffer + tokoff, toklen);
3359 strcat (token_name.buffer, ")");
3360 token.named = TRUE;
3362 else if (objdef == omethodtag
3363 || objdef == omethodparm)
3364 /* Objective C method */
3366 token.named = TRUE;
3368 else if (fvdef == fdefunname)
3369 /* GNU DEFUN and similar macros */
3371 bool defun = (newlb.buffer[tokoff] == 'F');
3372 int off = tokoff;
3373 int len = toklen;
3375 /* Rewrite the tag so that emacs lisp DEFUNs
3376 can be found by their elisp name */
3377 if (defun)
3379 off += 1;
3380 len -= 1;
3382 linebuffer_setlen (&token_name, len);
3383 strncpy (token_name.buffer,
3384 newlb.buffer + off, len);
3385 token_name.buffer[len] = '\0';
3386 if (defun)
3387 while (--len >= 0)
3388 if (token_name.buffer[len] == '_')
3389 token_name.buffer[len] = '-';
3390 token.named = defun;
3392 else
3394 linebuffer_setlen (&token_name, toklen);
3395 strncpy (token_name.buffer,
3396 newlb.buffer + tokoff, toklen);
3397 token_name.buffer[toklen] = '\0';
3398 /* Name macros and members. */
3399 token.named = (structdef == stagseen
3400 || typdef == ttypeseen
3401 || typdef == tend
3402 || (funorvar
3403 && definedef == dignorerest)
3404 || (funorvar
3405 && definedef == dnone
3406 && structdef == snone
3407 && bracelev > 0));
3409 token.lineno = lineno;
3410 token.offset = tokoff;
3411 token.length = toklen;
3412 token.line = newlb.buffer;
3413 token.linepos = newlinepos;
3414 token.valid = TRUE;
3416 if (definedef == dnone
3417 && (fvdef == fvnameseen
3418 || fvdef == foperator
3419 || structdef == stagseen
3420 || typdef == tend
3421 || typdef == ttypeseen
3422 || objdef != onone))
3424 if (current_lb_is_new)
3425 switch_line_buffers ();
3427 else if (definedef != dnone
3428 || fvdef == fdefunname
3429 || instruct)
3430 make_C_tag (funorvar);
3432 else /* not yacc and consider_token failed */
3434 if (inattribute && fvdef == fignore)
3436 /* We have just met __attribute__ after a
3437 function parameter list: do not tag the
3438 function again. */
3439 fvdef = fvnone;
3442 midtoken = FALSE;
3444 } /* if (endtoken (c)) */
3445 else if (intoken (c))
3446 still_in_token:
3448 toklen++;
3449 continue;
3451 } /* if (midtoken) */
3452 else if (begtoken (c))
3454 switch (definedef)
3456 case dnone:
3457 switch (fvdef)
3459 case fstartlist:
3460 /* This prevents tagging fb in
3461 void (__attribute__((noreturn)) *fb) (void);
3462 Fixing this is not easy and not very important. */
3463 fvdef = finlist;
3464 continue;
3465 case flistseen:
3466 if (plainc || declarations)
3468 make_C_tag (TRUE); /* a function */
3469 fvdef = fignore;
3471 break;
3473 if (structdef == stagseen && !cjava)
3475 popclass_above (bracelev);
3476 structdef = snone;
3478 break;
3479 case dsharpseen:
3480 savetoken = token;
3481 break;
3483 if (!yacc_rules || lp == newlb.buffer + 1)
3485 tokoff = lp - 1 - newlb.buffer;
3486 toklen = 1;
3487 midtoken = TRUE;
3489 continue;
3490 } /* if (begtoken) */
3491 } /* if must look at token */
3494 /* Detect end of line, colon, comma, semicolon and various braces
3495 after having handled a token.*/
3496 switch (c)
3498 case ':':
3499 if (inattribute)
3500 break;
3501 if (yacc_rules && token.offset == 0 && token.valid)
3503 make_C_tag (FALSE); /* a yacc function */
3504 break;
3506 if (definedef != dnone)
3507 break;
3508 switch (objdef)
3510 case otagseen:
3511 objdef = oignore;
3512 make_C_tag (TRUE); /* an Objective C class */
3513 break;
3514 case omethodtag:
3515 case omethodparm:
3516 objdef = omethodcolon;
3517 linebuffer_setlen (&token_name, token_name.len + 1);
3518 strcat (token_name.buffer, ":");
3519 break;
3521 if (structdef == stagseen)
3523 structdef = scolonseen;
3524 break;
3526 /* Should be useless, but may be work as a safety net. */
3527 if (cplpl && fvdef == flistseen)
3529 make_C_tag (TRUE); /* a function */
3530 fvdef = fignore;
3531 break;
3533 break;
3534 case ';':
3535 if (definedef != dnone || inattribute)
3536 break;
3537 switch (typdef)
3539 case tend:
3540 case ttypeseen:
3541 make_C_tag (FALSE); /* a typedef */
3542 typdef = tnone;
3543 fvdef = fvnone;
3544 break;
3545 case tnone:
3546 case tinbody:
3547 case tignore:
3548 switch (fvdef)
3550 case fignore:
3551 if (typdef == tignore || cplpl)
3552 fvdef = fvnone;
3553 break;
3554 case fvnameseen:
3555 if ((globals && bracelev == 0 && (!fvextern || declarations))
3556 || (members && instruct))
3557 make_C_tag (FALSE); /* a variable */
3558 fvextern = FALSE;
3559 fvdef = fvnone;
3560 token.valid = FALSE;
3561 break;
3562 case flistseen:
3563 if ((declarations
3564 && (cplpl || !instruct)
3565 && (typdef == tnone || (typdef != tignore && instruct)))
3566 || (members
3567 && plainc && instruct))
3568 make_C_tag (TRUE); /* a function */
3569 /* FALLTHRU */
3570 default:
3571 fvextern = FALSE;
3572 fvdef = fvnone;
3573 if (declarations
3574 && cplpl && structdef == stagseen)
3575 make_C_tag (FALSE); /* forward declaration */
3576 else
3577 token.valid = FALSE;
3578 } /* switch (fvdef) */
3579 /* FALLTHRU */
3580 default:
3581 if (!instruct)
3582 typdef = tnone;
3584 if (structdef == stagseen)
3585 structdef = snone;
3586 break;
3587 case ',':
3588 if (definedef != dnone || inattribute)
3589 break;
3590 switch (objdef)
3592 case omethodtag:
3593 case omethodparm:
3594 make_C_tag (TRUE); /* an Objective C method */
3595 objdef = oinbody;
3596 break;
3598 switch (fvdef)
3600 case fdefunkey:
3601 case foperator:
3602 case fstartlist:
3603 case finlist:
3604 case fignore:
3605 case vignore:
3606 break;
3607 case fdefunname:
3608 fvdef = fignore;
3609 break;
3610 case fvnameseen:
3611 if (parlev == 0
3612 && ((globals
3613 && bracelev == 0
3614 && templatelev == 0
3615 && (!fvextern || declarations))
3616 || (members && instruct)))
3617 make_C_tag (FALSE); /* a variable */
3618 break;
3619 case flistseen:
3620 if ((declarations && typdef == tnone && !instruct)
3621 || (members && typdef != tignore && instruct))
3623 make_C_tag (TRUE); /* a function */
3624 fvdef = fvnameseen;
3626 else if (!declarations)
3627 fvdef = fvnone;
3628 token.valid = FALSE;
3629 break;
3630 default:
3631 fvdef = fvnone;
3633 if (structdef == stagseen)
3634 structdef = snone;
3635 break;
3636 case ']':
3637 if (definedef != dnone || inattribute)
3638 break;
3639 if (structdef == stagseen)
3640 structdef = snone;
3641 switch (typdef)
3643 case ttypeseen:
3644 case tend:
3645 typdef = tignore;
3646 make_C_tag (FALSE); /* a typedef */
3647 break;
3648 case tnone:
3649 case tinbody:
3650 switch (fvdef)
3652 case foperator:
3653 case finlist:
3654 case fignore:
3655 case vignore:
3656 break;
3657 case fvnameseen:
3658 if ((members && bracelev == 1)
3659 || (globals && bracelev == 0
3660 && (!fvextern || declarations)))
3661 make_C_tag (FALSE); /* a variable */
3662 /* FALLTHRU */
3663 default:
3664 fvdef = fvnone;
3666 break;
3668 break;
3669 case '(':
3670 if (inattribute)
3672 attrparlev++;
3673 break;
3675 if (definedef != dnone)
3676 break;
3677 if (objdef == otagseen && parlev == 0)
3678 objdef = oparenseen;
3679 switch (fvdef)
3681 case fvnameseen:
3682 if (typdef == ttypeseen
3683 && *lp != '*'
3684 && !instruct)
3686 /* This handles constructs like:
3687 typedef void OperatorFun (int fun); */
3688 make_C_tag (FALSE);
3689 typdef = tignore;
3690 fvdef = fignore;
3691 break;
3693 /* FALLTHRU */
3694 case foperator:
3695 fvdef = fstartlist;
3696 break;
3697 case flistseen:
3698 fvdef = finlist;
3699 break;
3701 parlev++;
3702 break;
3703 case ')':
3704 if (inattribute)
3706 if (--attrparlev == 0)
3707 inattribute = FALSE;
3708 break;
3710 if (definedef != dnone)
3711 break;
3712 if (objdef == ocatseen && parlev == 1)
3714 make_C_tag (TRUE); /* an Objective C category */
3715 objdef = oignore;
3717 if (--parlev == 0)
3719 switch (fvdef)
3721 case fstartlist:
3722 case finlist:
3723 fvdef = flistseen;
3724 break;
3726 if (!instruct
3727 && (typdef == tend
3728 || typdef == ttypeseen))
3730 typdef = tignore;
3731 make_C_tag (FALSE); /* a typedef */
3734 else if (parlev < 0) /* can happen due to ill-conceived #if's. */
3735 parlev = 0;
3736 break;
3737 case '{':
3738 if (definedef != dnone)
3739 break;
3740 if (typdef == ttypeseen)
3742 /* Whenever typdef is set to tinbody (currently only
3743 here), typdefbracelev should be set to bracelev. */
3744 typdef = tinbody;
3745 typdefbracelev = bracelev;
3747 switch (fvdef)
3749 case flistseen:
3750 make_C_tag (TRUE); /* a function */
3751 /* FALLTHRU */
3752 case fignore:
3753 fvdef = fvnone;
3754 break;
3755 case fvnone:
3756 switch (objdef)
3758 case otagseen:
3759 make_C_tag (TRUE); /* an Objective C class */
3760 objdef = oignore;
3761 break;
3762 case omethodtag:
3763 case omethodparm:
3764 make_C_tag (TRUE); /* an Objective C method */
3765 objdef = oinbody;
3766 break;
3767 default:
3768 /* Neutralize `extern "C" {' grot. */
3769 if (bracelev == 0 && structdef == snone && nestlev == 0
3770 && typdef == tnone)
3771 bracelev = -1;
3773 break;
3775 switch (structdef)
3777 case skeyseen: /* unnamed struct */
3778 pushclass_above (bracelev, NULL, 0);
3779 structdef = snone;
3780 break;
3781 case stagseen: /* named struct or enum */
3782 case scolonseen: /* a class */
3783 pushclass_above (bracelev,token.line+token.offset, token.length);
3784 structdef = snone;
3785 make_C_tag (FALSE); /* a struct or enum */
3786 break;
3788 bracelev += 1;
3789 break;
3790 case '*':
3791 if (definedef != dnone)
3792 break;
3793 if (fvdef == fstartlist)
3795 fvdef = fvnone; /* avoid tagging `foo' in `foo (*bar()) ()' */
3796 token.valid = FALSE;
3798 break;
3799 case '}':
3800 if (definedef != dnone)
3801 break;
3802 bracelev -= 1;
3803 if (!ignoreindent && lp == newlb.buffer + 1)
3805 if (bracelev != 0)
3806 token.valid = FALSE; /* unexpected value, token unreliable */
3807 bracelev = 0; /* reset brace level if first column */
3808 parlev = 0; /* also reset paren level, just in case... */
3810 else if (bracelev < 0)
3812 token.valid = FALSE; /* something gone amiss, token unreliable */
3813 bracelev = 0;
3815 if (bracelev == 0 && fvdef == vignore)
3816 fvdef = fvnone; /* end of function */
3817 popclass_above (bracelev);
3818 structdef = snone;
3819 /* Only if typdef == tinbody is typdefbracelev significant. */
3820 if (typdef == tinbody && bracelev <= typdefbracelev)
3822 assert (bracelev == typdefbracelev);
3823 typdef = tend;
3825 break;
3826 case '=':
3827 if (definedef != dnone)
3828 break;
3829 switch (fvdef)
3831 case foperator:
3832 case finlist:
3833 case fignore:
3834 case vignore:
3835 break;
3836 case fvnameseen:
3837 if ((members && bracelev == 1)
3838 || (globals && bracelev == 0 && (!fvextern || declarations)))
3839 make_C_tag (FALSE); /* a variable */
3840 /* FALLTHRU */
3841 default:
3842 fvdef = vignore;
3844 break;
3845 case '<':
3846 if (cplpl
3847 && (structdef == stagseen || fvdef == fvnameseen))
3849 templatelev++;
3850 break;
3852 goto resetfvdef;
3853 case '>':
3854 if (templatelev > 0)
3856 templatelev--;
3857 break;
3859 goto resetfvdef;
3860 case '+':
3861 case '-':
3862 if (objdef == oinbody && bracelev == 0)
3864 objdef = omethodsign;
3865 break;
3867 /* FALLTHRU */
3868 resetfvdef:
3869 case '#': case '~': case '&': case '%': case '/':
3870 case '|': case '^': case '!': case '.': case '?':
3871 if (definedef != dnone)
3872 break;
3873 /* These surely cannot follow a function tag in C. */
3874 switch (fvdef)
3876 case foperator:
3877 case finlist:
3878 case fignore:
3879 case vignore:
3880 break;
3881 default:
3882 fvdef = fvnone;
3884 break;
3885 case '\0':
3886 if (objdef == otagseen)
3888 make_C_tag (TRUE); /* an Objective C class */
3889 objdef = oignore;
3891 /* If a macro spans multiple lines don't reset its state. */
3892 if (quotednl)
3893 CNL_SAVE_DEFINEDEF ();
3894 else
3895 CNL ();
3896 break;
3897 } /* switch (c) */
3899 } /* while not eof */
3901 free (lbs[0].lb.buffer);
3902 free (lbs[1].lb.buffer);
3906 * Process either a C++ file or a C file depending on the setting
3907 * of a global flag.
3909 static void
3910 default_C_entries (FILE *inf)
3912 C_entries (cplusplus ? C_PLPL : C_AUTO, inf);
3915 /* Always do plain C. */
3916 static void
3917 plain_C_entries (FILE *inf)
3919 C_entries (0, inf);
3922 /* Always do C++. */
3923 static void
3924 Cplusplus_entries (FILE *inf)
3926 C_entries (C_PLPL, inf);
3929 /* Always do Java. */
3930 static void
3931 Cjava_entries (FILE *inf)
3933 C_entries (C_JAVA, inf);
3936 /* Always do C*. */
3937 static void
3938 Cstar_entries (FILE *inf)
3940 C_entries (C_STAR, inf);
3943 /* Always do Yacc. */
3944 static void
3945 Yacc_entries (FILE *inf)
3947 C_entries (YACC, inf);
3951 /* Useful macros. */
3952 #define LOOP_ON_INPUT_LINES(file_pointer, line_buffer, char_pointer) \
3953 for (; /* loop initialization */ \
3954 !feof (file_pointer) /* loop test */ \
3955 && /* instructions at start of loop */ \
3956 (readline (&line_buffer, file_pointer), \
3957 char_pointer = line_buffer.buffer, \
3958 TRUE); \
3961 #define LOOKING_AT(cp, kw) /* kw is the keyword, a literal string */ \
3962 ((assert ("" kw), TRUE) /* syntax error if not a literal string */ \
3963 && strneq ((cp), kw, sizeof (kw)-1) /* cp points at kw */ \
3964 && notinname ((cp)[sizeof (kw)-1]) /* end of kw */ \
3965 && ((cp) = skip_spaces ((cp)+sizeof (kw)-1))) /* skip spaces */
3967 /* Similar to LOOKING_AT but does not use notinname, does not skip */
3968 #define LOOKING_AT_NOCASE(cp, kw) /* the keyword is a literal string */ \
3969 ((assert ("" kw), TRUE) /* syntax error if not a literal string */ \
3970 && strncaseeq ((cp), kw, sizeof (kw)-1) /* cp points at kw */ \
3971 && ((cp) += sizeof (kw)-1)) /* skip spaces */
3974 * Read a file, but do no processing. This is used to do regexp
3975 * matching on files that have no language defined.
3977 static void
3978 just_read_file (FILE *inf)
3980 while (!feof (inf))
3981 readline (&lb, inf);
3985 /* Fortran parsing */
3987 static void F_takeprec (void);
3988 static void F_getit (FILE *);
3990 static void
3991 F_takeprec (void)
3993 dbp = skip_spaces (dbp);
3994 if (*dbp != '*')
3995 return;
3996 dbp++;
3997 dbp = skip_spaces (dbp);
3998 if (strneq (dbp, "(*)", 3))
4000 dbp += 3;
4001 return;
4003 if (!ISDIGIT (*dbp))
4005 --dbp; /* force failure */
4006 return;
4009 dbp++;
4010 while (ISDIGIT (*dbp));
4013 static void
4014 F_getit (FILE *inf)
4016 register char *cp;
4018 dbp = skip_spaces (dbp);
4019 if (*dbp == '\0')
4021 readline (&lb, inf);
4022 dbp = lb.buffer;
4023 if (dbp[5] != '&')
4024 return;
4025 dbp += 6;
4026 dbp = skip_spaces (dbp);
4028 if (!ISALPHA (*dbp) && *dbp != '_' && *dbp != '$')
4029 return;
4030 for (cp = dbp + 1; *cp != '\0' && intoken (*cp); cp++)
4031 continue;
4032 make_tag (dbp, cp-dbp, TRUE,
4033 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4037 static void
4038 Fortran_functions (FILE *inf)
4040 LOOP_ON_INPUT_LINES (inf, lb, dbp)
4042 if (*dbp == '%')
4043 dbp++; /* Ratfor escape to fortran */
4044 dbp = skip_spaces (dbp);
4045 if (*dbp == '\0')
4046 continue;
4048 if (LOOKING_AT_NOCASE (dbp, "recursive"))
4049 dbp = skip_spaces (dbp);
4051 if (LOOKING_AT_NOCASE (dbp, "pure"))
4052 dbp = skip_spaces (dbp);
4054 if (LOOKING_AT_NOCASE (dbp, "elemental"))
4055 dbp = skip_spaces (dbp);
4057 switch (lowcase (*dbp))
4059 case 'i':
4060 if (nocase_tail ("integer"))
4061 F_takeprec ();
4062 break;
4063 case 'r':
4064 if (nocase_tail ("real"))
4065 F_takeprec ();
4066 break;
4067 case 'l':
4068 if (nocase_tail ("logical"))
4069 F_takeprec ();
4070 break;
4071 case 'c':
4072 if (nocase_tail ("complex") || nocase_tail ("character"))
4073 F_takeprec ();
4074 break;
4075 case 'd':
4076 if (nocase_tail ("double"))
4078 dbp = skip_spaces (dbp);
4079 if (*dbp == '\0')
4080 continue;
4081 if (nocase_tail ("precision"))
4082 break;
4083 continue;
4085 break;
4087 dbp = skip_spaces (dbp);
4088 if (*dbp == '\0')
4089 continue;
4090 switch (lowcase (*dbp))
4092 case 'f':
4093 if (nocase_tail ("function"))
4094 F_getit (inf);
4095 continue;
4096 case 's':
4097 if (nocase_tail ("subroutine"))
4098 F_getit (inf);
4099 continue;
4100 case 'e':
4101 if (nocase_tail ("entry"))
4102 F_getit (inf);
4103 continue;
4104 case 'b':
4105 if (nocase_tail ("blockdata") || nocase_tail ("block data"))
4107 dbp = skip_spaces (dbp);
4108 if (*dbp == '\0') /* assume un-named */
4109 make_tag ("blockdata", 9, TRUE,
4110 lb.buffer, dbp - lb.buffer, lineno, linecharno);
4111 else
4112 F_getit (inf); /* look for name */
4114 continue;
4121 * Ada parsing
4122 * Original code by
4123 * Philippe Waroquiers (1998)
4126 /* Once we are positioned after an "interesting" keyword, let's get
4127 the real tag value necessary. */
4128 static void
4129 Ada_getit (FILE *inf, const char *name_qualifier)
4131 register char *cp;
4132 char *name;
4133 char c;
4135 while (!feof (inf))
4137 dbp = skip_spaces (dbp);
4138 if (*dbp == '\0'
4139 || (dbp[0] == '-' && dbp[1] == '-'))
4141 readline (&lb, inf);
4142 dbp = lb.buffer;
4144 switch (lowcase (*dbp))
4146 case 'b':
4147 if (nocase_tail ("body"))
4149 /* Skipping body of procedure body or package body or ....
4150 resetting qualifier to body instead of spec. */
4151 name_qualifier = "/b";
4152 continue;
4154 break;
4155 case 't':
4156 /* Skipping type of task type or protected type ... */
4157 if (nocase_tail ("type"))
4158 continue;
4159 break;
4161 if (*dbp == '"')
4163 dbp += 1;
4164 for (cp = dbp; *cp != '\0' && *cp != '"'; cp++)
4165 continue;
4167 else
4169 dbp = skip_spaces (dbp);
4170 for (cp = dbp;
4171 (*cp != '\0'
4172 && (ISALPHA (*cp) || ISDIGIT (*cp) || *cp == '_' || *cp == '.'));
4173 cp++)
4174 continue;
4175 if (cp == dbp)
4176 return;
4178 c = *cp;
4179 *cp = '\0';
4180 name = concat (dbp, name_qualifier, "");
4181 *cp = c;
4182 make_tag (name, strlen (name), TRUE,
4183 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4184 free (name);
4185 if (c == '"')
4186 dbp = cp + 1;
4187 return;
4191 static void
4192 Ada_funcs (FILE *inf)
4194 bool inquote = FALSE;
4195 bool skip_till_semicolumn = FALSE;
4197 LOOP_ON_INPUT_LINES (inf, lb, dbp)
4199 while (*dbp != '\0')
4201 /* Skip a string i.e. "abcd". */
4202 if (inquote || (*dbp == '"'))
4204 dbp = etags_strchr (dbp + !inquote, '"');
4205 if (dbp != NULL)
4207 inquote = FALSE;
4208 dbp += 1;
4209 continue; /* advance char */
4211 else
4213 inquote = TRUE;
4214 break; /* advance line */
4218 /* Skip comments. */
4219 if (dbp[0] == '-' && dbp[1] == '-')
4220 break; /* advance line */
4222 /* Skip character enclosed in single quote i.e. 'a'
4223 and skip single quote starting an attribute i.e. 'Image. */
4224 if (*dbp == '\'')
4226 dbp++ ;
4227 if (*dbp != '\0')
4228 dbp++;
4229 continue;
4232 if (skip_till_semicolumn)
4234 if (*dbp == ';')
4235 skip_till_semicolumn = FALSE;
4236 dbp++;
4237 continue; /* advance char */
4240 /* Search for beginning of a token. */
4241 if (!begtoken (*dbp))
4243 dbp++;
4244 continue; /* advance char */
4247 /* We are at the beginning of a token. */
4248 switch (lowcase (*dbp))
4250 case 'f':
4251 if (!packages_only && nocase_tail ("function"))
4252 Ada_getit (inf, "/f");
4253 else
4254 break; /* from switch */
4255 continue; /* advance char */
4256 case 'p':
4257 if (!packages_only && nocase_tail ("procedure"))
4258 Ada_getit (inf, "/p");
4259 else if (nocase_tail ("package"))
4260 Ada_getit (inf, "/s");
4261 else if (nocase_tail ("protected")) /* protected type */
4262 Ada_getit (inf, "/t");
4263 else
4264 break; /* from switch */
4265 continue; /* advance char */
4267 case 'u':
4268 if (typedefs && !packages_only && nocase_tail ("use"))
4270 /* when tagging types, avoid tagging use type Pack.Typename;
4271 for this, we will skip everything till a ; */
4272 skip_till_semicolumn = TRUE;
4273 continue; /* advance char */
4276 case 't':
4277 if (!packages_only && nocase_tail ("task"))
4278 Ada_getit (inf, "/k");
4279 else if (typedefs && !packages_only && nocase_tail ("type"))
4281 Ada_getit (inf, "/t");
4282 while (*dbp != '\0')
4283 dbp += 1;
4285 else
4286 break; /* from switch */
4287 continue; /* advance char */
4290 /* Look for the end of the token. */
4291 while (!endtoken (*dbp))
4292 dbp++;
4294 } /* advance char */
4295 } /* advance line */
4300 * Unix and microcontroller assembly tag handling
4301 * Labels: /^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]/
4302 * Idea by Bob Weiner, Motorola Inc. (1994)
4304 static void
4305 Asm_labels (FILE *inf)
4307 register char *cp;
4309 LOOP_ON_INPUT_LINES (inf, lb, cp)
4311 /* If first char is alphabetic or one of [_.$], test for colon
4312 following identifier. */
4313 if (ISALPHA (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
4315 /* Read past label. */
4316 cp++;
4317 while (ISALNUM (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
4318 cp++;
4319 if (*cp == ':' || iswhite (*cp))
4320 /* Found end of label, so copy it and add it to the table. */
4321 make_tag (lb.buffer, cp - lb.buffer, TRUE,
4322 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4329 * Perl support
4330 * Perl sub names: /^sub[ \t\n]+[^ \t\n{]+/
4331 * Perl variable names: /^(my|local).../
4332 * Original code by Bart Robinson <lomew@cs.utah.edu> (1995)
4333 * Additions by Michael Ernst <mernst@alum.mit.edu> (1997)
4334 * Ideas by Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> (2001)
4336 static void
4337 Perl_functions (FILE *inf)
4339 char *package = savestr ("main"); /* current package name */
4340 register char *cp;
4342 LOOP_ON_INPUT_LINES (inf, lb, cp)
4344 cp = skip_spaces (cp);
4346 if (LOOKING_AT (cp, "package"))
4348 free (package);
4349 get_tag (cp, &package);
4351 else if (LOOKING_AT (cp, "sub"))
4353 char *pos;
4354 char *sp = cp;
4356 while (!notinname (*cp))
4357 cp++;
4358 if (cp == sp)
4359 continue; /* nothing found */
4360 if ((pos = etags_strchr (sp, ':')) != NULL
4361 && pos < cp && pos[1] == ':')
4362 /* The name is already qualified. */
4363 make_tag (sp, cp - sp, TRUE,
4364 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4365 else
4366 /* Qualify it. */
4368 char savechar, *name;
4370 savechar = *cp;
4371 *cp = '\0';
4372 name = concat (package, "::", sp);
4373 *cp = savechar;
4374 make_tag (name, strlen (name), TRUE,
4375 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4376 free (name);
4379 else if (globals) /* only if we are tagging global vars */
4381 /* Skip a qualifier, if any. */
4382 bool qual = LOOKING_AT (cp, "my") || LOOKING_AT (cp, "local");
4383 /* After "my" or "local", but before any following paren or space. */
4384 char *varstart = cp;
4386 if (qual /* should this be removed? If yes, how? */
4387 && (*cp == '$' || *cp == '@' || *cp == '%'))
4389 varstart += 1;
4391 cp++;
4392 while (ISALNUM (*cp) || *cp == '_');
4394 else if (qual)
4396 /* Should be examining a variable list at this point;
4397 could insist on seeing an open parenthesis. */
4398 while (*cp != '\0' && *cp != ';' && *cp != '=' && *cp != ')')
4399 cp++;
4401 else
4402 continue;
4404 make_tag (varstart, cp - varstart, FALSE,
4405 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4408 free (package);
4413 * Python support
4414 * Look for /^[\t]*def[ \t\n]+[^ \t\n(:]+/ or /^class[ \t\n]+[^ \t\n(:]+/
4415 * Idea by Eric S. Raymond <esr@thyrsus.com> (1997)
4416 * More ideas by seb bacon <seb@jamkit.com> (2002)
4418 static void
4419 Python_functions (FILE *inf)
4421 register char *cp;
4423 LOOP_ON_INPUT_LINES (inf, lb, cp)
4425 cp = skip_spaces (cp);
4426 if (LOOKING_AT (cp, "def") || LOOKING_AT (cp, "class"))
4428 char *name = cp;
4429 while (!notinname (*cp) && *cp != ':')
4430 cp++;
4431 make_tag (name, cp - name, TRUE,
4432 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4439 * PHP support
4440 * Look for:
4441 * - /^[ \t]*function[ \t\n]+[^ \t\n(]+/
4442 * - /^[ \t]*class[ \t\n]+[^ \t\n]+/
4443 * - /^[ \t]*define\(\"[^\"]+/
4444 * Only with --members:
4445 * - /^[ \t]*var[ \t\n]+\$[^ \t\n=;]/
4446 * Idea by Diez B. Roggisch (2001)
4448 static void
4449 PHP_functions (FILE *inf)
4451 register char *cp, *name;
4452 bool search_identifier = FALSE;
4454 LOOP_ON_INPUT_LINES (inf, lb, cp)
4456 cp = skip_spaces (cp);
4457 name = cp;
4458 if (search_identifier
4459 && *cp != '\0')
4461 while (!notinname (*cp))
4462 cp++;
4463 make_tag (name, cp - name, TRUE,
4464 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4465 search_identifier = FALSE;
4467 else if (LOOKING_AT (cp, "function"))
4469 if (*cp == '&')
4470 cp = skip_spaces (cp+1);
4471 if (*cp != '\0')
4473 name = cp;
4474 while (!notinname (*cp))
4475 cp++;
4476 make_tag (name, cp - name, TRUE,
4477 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4479 else
4480 search_identifier = TRUE;
4482 else if (LOOKING_AT (cp, "class"))
4484 if (*cp != '\0')
4486 name = cp;
4487 while (*cp != '\0' && !iswhite (*cp))
4488 cp++;
4489 make_tag (name, cp - name, FALSE,
4490 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4492 else
4493 search_identifier = TRUE;
4495 else if (strneq (cp, "define", 6)
4496 && (cp = skip_spaces (cp+6))
4497 && *cp++ == '('
4498 && (*cp == '"' || *cp == '\''))
4500 char quote = *cp++;
4501 name = cp;
4502 while (*cp != quote && *cp != '\0')
4503 cp++;
4504 make_tag (name, cp - name, FALSE,
4505 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4507 else if (members
4508 && LOOKING_AT (cp, "var")
4509 && *cp == '$')
4511 name = cp;
4512 while (!notinname (*cp))
4513 cp++;
4514 make_tag (name, cp - name, FALSE,
4515 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
4522 * Cobol tag functions
4523 * We could look for anything that could be a paragraph name.
4524 * i.e. anything that starts in column 8 is one word and ends in a full stop.
4525 * Idea by Corny de Souza (1993)
4527 static void
4528 Cobol_paragraphs (FILE *inf)
4530 register char *bp, *ep;
4532 LOOP_ON_INPUT_LINES (inf, lb, bp)
4534 if (lb.len < 9)
4535 continue;
4536 bp += 8;
4538 /* If eoln, compiler option or comment ignore whole line. */
4539 if (bp[-1] != ' ' || !ISALNUM (bp[0]))
4540 continue;
4542 for (ep = bp; ISALNUM (*ep) || *ep == '-'; ep++)
4543 continue;
4544 if (*ep++ == '.')
4545 make_tag (bp, ep - bp, TRUE,
4546 lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
4552 * Makefile support
4553 * Ideas by Assar Westerlund <assar@sics.se> (2001)
4555 static void
4556 Makefile_targets (FILE *inf)
4558 register char *bp;
4560 LOOP_ON_INPUT_LINES (inf, lb, bp)
4562 if (*bp == '\t' || *bp == '#')
4563 continue;
4564 while (*bp != '\0' && *bp != '=' && *bp != ':')
4565 bp++;
4566 if (*bp == ':' || (globals && *bp == '='))
4568 /* We should detect if there is more than one tag, but we do not.
4569 We just skip initial and final spaces. */
4570 char * namestart = skip_spaces (lb.buffer);
4571 while (--bp > namestart)
4572 if (!notinname (*bp))
4573 break;
4574 make_tag (namestart, bp - namestart + 1, TRUE,
4575 lb.buffer, bp - lb.buffer + 2, lineno, linecharno);
4582 * Pascal parsing
4583 * Original code by Mosur K. Mohan (1989)
4585 * Locates tags for procedures & functions. Doesn't do any type- or
4586 * var-definitions. It does look for the keyword "extern" or
4587 * "forward" immediately following the procedure statement; if found,
4588 * the tag is skipped.
4590 static void
4591 Pascal_functions (FILE *inf)
4593 linebuffer tline; /* mostly copied from C_entries */
4594 long save_lcno;
4595 int save_lineno, namelen, taglen;
4596 char c, *name;
4598 bool /* each of these flags is TRUE if: */
4599 incomment, /* point is inside a comment */
4600 inquote, /* point is inside '..' string */
4601 get_tagname, /* point is after PROCEDURE/FUNCTION
4602 keyword, so next item = potential tag */
4603 found_tag, /* point is after a potential tag */
4604 inparms, /* point is within parameter-list */
4605 verify_tag; /* point has passed the parm-list, so the
4606 next token will determine whether this
4607 is a FORWARD/EXTERN to be ignored, or
4608 whether it is a real tag */
4610 save_lcno = save_lineno = namelen = taglen = 0; /* keep compiler quiet */
4611 name = NULL; /* keep compiler quiet */
4612 dbp = lb.buffer;
4613 *dbp = '\0';
4614 linebuffer_init (&tline);
4616 incomment = inquote = FALSE;
4617 found_tag = FALSE; /* have a proc name; check if extern */
4618 get_tagname = FALSE; /* found "procedure" keyword */
4619 inparms = FALSE; /* found '(' after "proc" */
4620 verify_tag = FALSE; /* check if "extern" is ahead */
4623 while (!feof (inf)) /* long main loop to get next char */
4625 c = *dbp++;
4626 if (c == '\0') /* if end of line */
4628 readline (&lb, inf);
4629 dbp = lb.buffer;
4630 if (*dbp == '\0')
4631 continue;
4632 if (!((found_tag && verify_tag)
4633 || get_tagname))
4634 c = *dbp++; /* only if don't need *dbp pointing
4635 to the beginning of the name of
4636 the procedure or function */
4638 if (incomment)
4640 if (c == '}') /* within { } comments */
4641 incomment = FALSE;
4642 else if (c == '*' && *dbp == ')') /* within (* *) comments */
4644 dbp++;
4645 incomment = FALSE;
4647 continue;
4649 else if (inquote)
4651 if (c == '\'')
4652 inquote = FALSE;
4653 continue;
4655 else
4656 switch (c)
4658 case '\'':
4659 inquote = TRUE; /* found first quote */
4660 continue;
4661 case '{': /* found open { comment */
4662 incomment = TRUE;
4663 continue;
4664 case '(':
4665 if (*dbp == '*') /* found open (* comment */
4667 incomment = TRUE;
4668 dbp++;
4670 else if (found_tag) /* found '(' after tag, i.e., parm-list */
4671 inparms = TRUE;
4672 continue;
4673 case ')': /* end of parms list */
4674 if (inparms)
4675 inparms = FALSE;
4676 continue;
4677 case ';':
4678 if (found_tag && !inparms) /* end of proc or fn stmt */
4680 verify_tag = TRUE;
4681 break;
4683 continue;
4685 if (found_tag && verify_tag && (*dbp != ' '))
4687 /* Check if this is an "extern" declaration. */
4688 if (*dbp == '\0')
4689 continue;
4690 if (lowcase (*dbp == 'e'))
4692 if (nocase_tail ("extern")) /* superfluous, really! */
4694 found_tag = FALSE;
4695 verify_tag = FALSE;
4698 else if (lowcase (*dbp) == 'f')
4700 if (nocase_tail ("forward")) /* check for forward reference */
4702 found_tag = FALSE;
4703 verify_tag = FALSE;
4706 if (found_tag && verify_tag) /* not external proc, so make tag */
4708 found_tag = FALSE;
4709 verify_tag = FALSE;
4710 make_tag (name, namelen, TRUE,
4711 tline.buffer, taglen, save_lineno, save_lcno);
4712 continue;
4715 if (get_tagname) /* grab name of proc or fn */
4717 char *cp;
4719 if (*dbp == '\0')
4720 continue;
4722 /* Find block name. */
4723 for (cp = dbp + 1; *cp != '\0' && !endtoken (*cp); cp++)
4724 continue;
4726 /* Save all values for later tagging. */
4727 linebuffer_setlen (&tline, lb.len);
4728 strcpy (tline.buffer, lb.buffer);
4729 save_lineno = lineno;
4730 save_lcno = linecharno;
4731 name = tline.buffer + (dbp - lb.buffer);
4732 namelen = cp - dbp;
4733 taglen = cp - lb.buffer + 1;
4735 dbp = cp; /* set dbp to e-o-token */
4736 get_tagname = FALSE;
4737 found_tag = TRUE;
4738 continue;
4740 /* And proceed to check for "extern". */
4742 else if (!incomment && !inquote && !found_tag)
4744 /* Check for proc/fn keywords. */
4745 switch (lowcase (c))
4747 case 'p':
4748 if (nocase_tail ("rocedure")) /* c = 'p', dbp has advanced */
4749 get_tagname = TRUE;
4750 continue;
4751 case 'f':
4752 if (nocase_tail ("unction"))
4753 get_tagname = TRUE;
4754 continue;
4757 } /* while not eof */
4759 free (tline.buffer);
4764 * Lisp tag functions
4765 * look for (def or (DEF, quote or QUOTE
4768 static void L_getit (void);
4770 static void
4771 L_getit (void)
4773 if (*dbp == '\'') /* Skip prefix quote */
4774 dbp++;
4775 else if (*dbp == '(')
4777 dbp++;
4778 /* Try to skip "(quote " */
4779 if (!LOOKING_AT (dbp, "quote") && !LOOKING_AT (dbp, "QUOTE"))
4780 /* Ok, then skip "(" before name in (defstruct (foo)) */
4781 dbp = skip_spaces (dbp);
4783 get_tag (dbp, NULL);
4786 static void
4787 Lisp_functions (FILE *inf)
4789 LOOP_ON_INPUT_LINES (inf, lb, dbp)
4791 if (dbp[0] != '(')
4792 continue;
4794 if (strneq (dbp+1, "def", 3) || strneq (dbp+1, "DEF", 3))
4796 dbp = skip_non_spaces (dbp);
4797 dbp = skip_spaces (dbp);
4798 L_getit ();
4800 else
4802 /* Check for (foo::defmumble name-defined ... */
4804 dbp++;
4805 while (!notinname (*dbp) && *dbp != ':');
4806 if (*dbp == ':')
4809 dbp++;
4810 while (*dbp == ':');
4812 if (strneq (dbp, "def", 3) || strneq (dbp, "DEF", 3))
4814 dbp = skip_non_spaces (dbp);
4815 dbp = skip_spaces (dbp);
4816 L_getit ();
4825 * Lua script language parsing
4826 * Original code by David A. Capello <dacap@users.sourceforge.net> (2004)
4828 * "function" and "local function" are tags if they start at column 1.
4830 static void
4831 Lua_functions (FILE *inf)
4833 register char *bp;
4835 LOOP_ON_INPUT_LINES (inf, lb, bp)
4837 if (bp[0] != 'f' && bp[0] != 'l')
4838 continue;
4840 (void)LOOKING_AT (bp, "local"); /* skip possible "local" */
4842 if (LOOKING_AT (bp, "function"))
4843 get_tag (bp, NULL);
4849 * PostScript tags
4850 * Just look for lines where the first character is '/'
4851 * Also look at "defineps" for PSWrap
4852 * Ideas by:
4853 * Richard Mlynarik <mly@adoc.xerox.com> (1997)
4854 * Masatake Yamato <masata-y@is.aist-nara.ac.jp> (1999)
4856 static void
4857 PS_functions (FILE *inf)
4859 register char *bp, *ep;
4861 LOOP_ON_INPUT_LINES (inf, lb, bp)
4863 if (bp[0] == '/')
4865 for (ep = bp+1;
4866 *ep != '\0' && *ep != ' ' && *ep != '{';
4867 ep++)
4868 continue;
4869 make_tag (bp, ep - bp, TRUE,
4870 lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
4872 else if (LOOKING_AT (bp, "defineps"))
4873 get_tag (bp, NULL);
4879 * Forth tags
4880 * Ignore anything after \ followed by space or in ( )
4881 * Look for words defined by :
4882 * Look for constant, code, create, defer, value, and variable
4883 * OBP extensions: Look for buffer:, field,
4884 * Ideas by Eduardo Horvath <eeh@netbsd.org> (2004)
4886 static void
4887 Forth_words (FILE *inf)
4889 register char *bp;
4891 LOOP_ON_INPUT_LINES (inf, lb, bp)
4892 while ((bp = skip_spaces (bp))[0] != '\0')
4893 if (bp[0] == '\\' && iswhite (bp[1]))
4894 break; /* read next line */
4895 else if (bp[0] == '(' && iswhite (bp[1]))
4896 do /* skip to ) or eol */
4897 bp++;
4898 while (*bp != ')' && *bp != '\0');
4899 else if ((bp[0] == ':' && iswhite (bp[1]) && bp++)
4900 || LOOKING_AT_NOCASE (bp, "constant")
4901 || LOOKING_AT_NOCASE (bp, "code")
4902 || LOOKING_AT_NOCASE (bp, "create")
4903 || LOOKING_AT_NOCASE (bp, "defer")
4904 || LOOKING_AT_NOCASE (bp, "value")
4905 || LOOKING_AT_NOCASE (bp, "variable")
4906 || LOOKING_AT_NOCASE (bp, "buffer:")
4907 || LOOKING_AT_NOCASE (bp, "field"))
4908 get_tag (skip_spaces (bp), NULL); /* Yay! A definition! */
4909 else
4910 bp = skip_non_spaces (bp);
4915 * Scheme tag functions
4916 * look for (def... xyzzy
4917 * (def... (xyzzy
4918 * (def ... ((...(xyzzy ....
4919 * (set! xyzzy
4920 * Original code by Ken Haase (1985?)
4922 static void
4923 Scheme_functions (FILE *inf)
4925 register char *bp;
4927 LOOP_ON_INPUT_LINES (inf, lb, bp)
4929 if (strneq (bp, "(def", 4) || strneq (bp, "(DEF", 4))
4931 bp = skip_non_spaces (bp+4);
4932 /* Skip over open parens and white space. Don't continue past
4933 '\0'. */
4934 while (*bp && notinname (*bp))
4935 bp++;
4936 get_tag (bp, NULL);
4938 if (LOOKING_AT (bp, "(SET!") || LOOKING_AT (bp, "(set!"))
4939 get_tag (bp, NULL);
4944 /* Find tags in TeX and LaTeX input files. */
4946 /* TEX_toktab is a table of TeX control sequences that define tags.
4947 * Each entry records one such control sequence.
4949 * Original code from who knows whom.
4950 * Ideas by:
4951 * Stefan Monnier (2002)
4954 static linebuffer *TEX_toktab = NULL; /* Table with tag tokens */
4956 /* Default set of control sequences to put into TEX_toktab.
4957 The value of environment var TEXTAGS is prepended to this. */
4958 static const char *TEX_defenv = "\
4959 :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem\
4960 :part:appendix:entry:index:def\
4961 :newcommand:renewcommand:newenvironment:renewenvironment";
4963 static void TEX_mode (FILE *);
4964 static void TEX_decode_env (const char *, const char *);
4966 static char TEX_esc = '\\';
4967 static char TEX_opgrp = '{';
4968 static char TEX_clgrp = '}';
4971 * TeX/LaTeX scanning loop.
4973 static void
4974 TeX_commands (FILE *inf)
4976 char *cp;
4977 linebuffer *key;
4979 /* Select either \ or ! as escape character. */
4980 TEX_mode (inf);
4982 /* Initialize token table once from environment. */
4983 if (TEX_toktab == NULL)
4984 TEX_decode_env ("TEXTAGS", TEX_defenv);
4986 LOOP_ON_INPUT_LINES (inf, lb, cp)
4988 /* Look at each TEX keyword in line. */
4989 for (;;)
4991 /* Look for a TEX escape. */
4992 while (*cp++ != TEX_esc)
4993 if (cp[-1] == '\0' || cp[-1] == '%')
4994 goto tex_next_line;
4996 for (key = TEX_toktab; key->buffer != NULL; key++)
4997 if (strneq (cp, key->buffer, key->len))
4999 register char *p;
5000 int namelen, linelen;
5001 bool opgrp = FALSE;
5003 cp = skip_spaces (cp + key->len);
5004 if (*cp == TEX_opgrp)
5006 opgrp = TRUE;
5007 cp++;
5009 for (p = cp;
5010 (!iswhite (*p) && *p != '#' &&
5011 *p != TEX_opgrp && *p != TEX_clgrp);
5012 p++)
5013 continue;
5014 namelen = p - cp;
5015 linelen = lb.len;
5016 if (!opgrp || *p == TEX_clgrp)
5018 while (*p != '\0' && *p != TEX_opgrp && *p != TEX_clgrp)
5019 p++;
5020 linelen = p - lb.buffer + 1;
5022 make_tag (cp, namelen, TRUE,
5023 lb.buffer, linelen, lineno, linecharno);
5024 goto tex_next_line; /* We only tag a line once */
5027 tex_next_line:
5032 #define TEX_LESC '\\'
5033 #define TEX_SESC '!'
5035 /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
5036 chars accordingly. */
5037 static void
5038 TEX_mode (FILE *inf)
5040 int c;
5042 while ((c = getc (inf)) != EOF)
5044 /* Skip to next line if we hit the TeX comment char. */
5045 if (c == '%')
5046 while (c != '\n' && c != EOF)
5047 c = getc (inf);
5048 else if (c == TEX_LESC || c == TEX_SESC )
5049 break;
5052 if (c == TEX_LESC)
5054 TEX_esc = TEX_LESC;
5055 TEX_opgrp = '{';
5056 TEX_clgrp = '}';
5058 else
5060 TEX_esc = TEX_SESC;
5061 TEX_opgrp = '<';
5062 TEX_clgrp = '>';
5064 /* If the input file is compressed, inf is a pipe, and rewind may fail.
5065 No attempt is made to correct the situation. */
5066 rewind (inf);
5069 /* Read environment and prepend it to the default string.
5070 Build token table. */
5071 static void
5072 TEX_decode_env (const char *evarname, const char *defenv)
5074 register const char *env, *p;
5075 int i, len;
5077 /* Append default string to environment. */
5078 env = getenv (evarname);
5079 if (!env)
5080 env = defenv;
5081 else
5082 env = concat (env, defenv, "");
5084 /* Allocate a token table */
5085 for (len = 1, p = env; p;)
5086 if ((p = etags_strchr (p, ':')) && *++p != '\0')
5087 len++;
5088 TEX_toktab = xnew (len, linebuffer);
5090 /* Unpack environment string into token table. Be careful about */
5091 /* zero-length strings (leading ':', "::" and trailing ':') */
5092 for (i = 0; *env != '\0';)
5094 p = etags_strchr (env, ':');
5095 if (!p) /* End of environment string. */
5096 p = env + strlen (env);
5097 if (p - env > 0)
5098 { /* Only non-zero strings. */
5099 TEX_toktab[i].buffer = savenstr (env, p - env);
5100 TEX_toktab[i].len = p - env;
5101 i++;
5103 if (*p)
5104 env = p + 1;
5105 else
5107 TEX_toktab[i].buffer = NULL; /* Mark end of table. */
5108 TEX_toktab[i].len = 0;
5109 break;
5115 /* Texinfo support. Dave Love, Mar. 2000. */
5116 static void
5117 Texinfo_nodes (FILE *inf)
5119 char *cp, *start;
5120 LOOP_ON_INPUT_LINES (inf, lb, cp)
5121 if (LOOKING_AT (cp, "@node"))
5123 start = cp;
5124 while (*cp != '\0' && *cp != ',')
5125 cp++;
5126 make_tag (start, cp - start, TRUE,
5127 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
5133 * HTML support.
5134 * Contents of <title>, <h1>, <h2>, <h3> are tags.
5135 * Contents of <a name=xxx> are tags with name xxx.
5137 * Francesco Potortì, 2002.
5139 static void
5140 HTML_labels (FILE *inf)
5142 bool getnext = FALSE; /* next text outside of HTML tags is a tag */
5143 bool skiptag = FALSE; /* skip to the end of the current HTML tag */
5144 bool intag = FALSE; /* inside an html tag, looking for ID= */
5145 bool inanchor = FALSE; /* when INTAG, is an anchor, look for NAME= */
5146 char *end;
5149 linebuffer_setlen (&token_name, 0); /* no name in buffer */
5151 LOOP_ON_INPUT_LINES (inf, lb, dbp)
5152 for (;;) /* loop on the same line */
5154 if (skiptag) /* skip HTML tag */
5156 while (*dbp != '\0' && *dbp != '>')
5157 dbp++;
5158 if (*dbp == '>')
5160 dbp += 1;
5161 skiptag = FALSE;
5162 continue; /* look on the same line */
5164 break; /* go to next line */
5167 else if (intag) /* look for "name=" or "id=" */
5169 while (*dbp != '\0' && *dbp != '>'
5170 && lowcase (*dbp) != 'n' && lowcase (*dbp) != 'i')
5171 dbp++;
5172 if (*dbp == '\0')
5173 break; /* go to next line */
5174 if (*dbp == '>')
5176 dbp += 1;
5177 intag = FALSE;
5178 continue; /* look on the same line */
5180 if ((inanchor && LOOKING_AT_NOCASE (dbp, "name="))
5181 || LOOKING_AT_NOCASE (dbp, "id="))
5183 bool quoted = (dbp[0] == '"');
5185 if (quoted)
5186 for (end = ++dbp; *end != '\0' && *end != '"'; end++)
5187 continue;
5188 else
5189 for (end = dbp; *end != '\0' && intoken (*end); end++)
5190 continue;
5191 linebuffer_setlen (&token_name, end - dbp);
5192 strncpy (token_name.buffer, dbp, end - dbp);
5193 token_name.buffer[end - dbp] = '\0';
5195 dbp = end;
5196 intag = FALSE; /* we found what we looked for */
5197 skiptag = TRUE; /* skip to the end of the tag */
5198 getnext = TRUE; /* then grab the text */
5199 continue; /* look on the same line */
5201 dbp += 1;
5204 else if (getnext) /* grab next tokens and tag them */
5206 dbp = skip_spaces (dbp);
5207 if (*dbp == '\0')
5208 break; /* go to next line */
5209 if (*dbp == '<')
5211 intag = TRUE;
5212 inanchor = (lowcase (dbp[1]) == 'a' && !intoken (dbp[2]));
5213 continue; /* look on the same line */
5216 for (end = dbp + 1; *end != '\0' && *end != '<'; end++)
5217 continue;
5218 make_tag (token_name.buffer, token_name.len, TRUE,
5219 dbp, end - dbp, lineno, linecharno);
5220 linebuffer_setlen (&token_name, 0); /* no name in buffer */
5221 getnext = FALSE;
5222 break; /* go to next line */
5225 else /* look for an interesting HTML tag */
5227 while (*dbp != '\0' && *dbp != '<')
5228 dbp++;
5229 if (*dbp == '\0')
5230 break; /* go to next line */
5231 intag = TRUE;
5232 if (lowcase (dbp[1]) == 'a' && !intoken (dbp[2]))
5234 inanchor = TRUE;
5235 continue; /* look on the same line */
5237 else if (LOOKING_AT_NOCASE (dbp, "<title>")
5238 || LOOKING_AT_NOCASE (dbp, "<h1>")
5239 || LOOKING_AT_NOCASE (dbp, "<h2>")
5240 || LOOKING_AT_NOCASE (dbp, "<h3>"))
5242 intag = FALSE;
5243 getnext = TRUE;
5244 continue; /* look on the same line */
5246 dbp += 1;
5253 * Prolog support
5255 * Assumes that the predicate or rule starts at column 0.
5256 * Only the first clause of a predicate or rule is added.
5257 * Original code by Sunichirou Sugou (1989)
5258 * Rewritten by Anders Lindgren (1996)
5260 static size_t prolog_pr (char *, char *);
5261 static void prolog_skip_comment (linebuffer *, FILE *);
5262 static size_t prolog_atom (char *, size_t);
5264 static void
5265 Prolog_functions (FILE *inf)
5267 char *cp, *last;
5268 size_t len;
5269 size_t allocated;
5271 allocated = 0;
5272 len = 0;
5273 last = NULL;
5275 LOOP_ON_INPUT_LINES (inf, lb, cp)
5277 if (cp[0] == '\0') /* Empty line */
5278 continue;
5279 else if (iswhite (cp[0])) /* Not a predicate */
5280 continue;
5281 else if (cp[0] == '/' && cp[1] == '*') /* comment. */
5282 prolog_skip_comment (&lb, inf);
5283 else if ((len = prolog_pr (cp, last)) > 0)
5285 /* Predicate or rule. Store the function name so that we
5286 only generate a tag for the first clause. */
5287 if (last == NULL)
5288 last = xnew (len + 1, char);
5289 else if (len + 1 > allocated)
5290 xrnew (last, len + 1, char);
5291 allocated = len + 1;
5292 strncpy (last, cp, len);
5293 last[len] = '\0';
5296 free (last);
5300 static void
5301 prolog_skip_comment (linebuffer *plb, FILE *inf)
5303 char *cp;
5307 for (cp = plb->buffer; *cp != '\0'; cp++)
5308 if (cp[0] == '*' && cp[1] == '/')
5309 return;
5310 readline (plb, inf);
5312 while (!feof (inf));
5316 * A predicate or rule definition is added if it matches:
5317 * <beginning of line><Prolog Atom><whitespace>(
5318 * or <beginning of line><Prolog Atom><whitespace>:-
5320 * It is added to the tags database if it doesn't match the
5321 * name of the previous clause header.
5323 * Return the size of the name of the predicate or rule, or 0 if no
5324 * header was found.
5326 static size_t
5327 prolog_pr (char *s, char *last)
5329 /* Name of last clause. */
5331 size_t pos;
5332 size_t len;
5334 pos = prolog_atom (s, 0);
5335 if (! pos)
5336 return 0;
5338 len = pos;
5339 pos = skip_spaces (s + pos) - s;
5341 if ((s[pos] == '.'
5342 || (s[pos] == '(' && (pos += 1))
5343 || (s[pos] == ':' && s[pos + 1] == '-' && (pos += 2)))
5344 && (last == NULL /* save only the first clause */
5345 || len != strlen (last)
5346 || !strneq (s, last, len)))
5348 make_tag (s, len, TRUE, s, pos, lineno, linecharno);
5349 return len;
5351 else
5352 return 0;
5356 * Consume a Prolog atom.
5357 * Return the number of bytes consumed, or 0 if there was an error.
5359 * A prolog atom, in this context, could be one of:
5360 * - An alphanumeric sequence, starting with a lower case letter.
5361 * - A quoted arbitrary string. Single quotes can escape themselves.
5362 * Backslash quotes everything.
5364 static size_t
5365 prolog_atom (char *s, size_t pos)
5367 size_t origpos;
5369 origpos = pos;
5371 if (ISLOWER (s[pos]) || (s[pos] == '_'))
5373 /* The atom is unquoted. */
5374 pos++;
5375 while (ISALNUM (s[pos]) || (s[pos] == '_'))
5377 pos++;
5379 return pos - origpos;
5381 else if (s[pos] == '\'')
5383 pos++;
5385 for (;;)
5387 if (s[pos] == '\'')
5389 pos++;
5390 if (s[pos] != '\'')
5391 break;
5392 pos++; /* A double quote */
5394 else if (s[pos] == '\0')
5395 /* Multiline quoted atoms are ignored. */
5396 return 0;
5397 else if (s[pos] == '\\')
5399 if (s[pos+1] == '\0')
5400 return 0;
5401 pos += 2;
5403 else
5404 pos++;
5406 return pos - origpos;
5408 else
5409 return 0;
5414 * Support for Erlang
5416 * Generates tags for functions, defines, and records.
5417 * Assumes that Erlang functions start at column 0.
5418 * Original code by Anders Lindgren (1996)
5420 static int erlang_func (char *, char *);
5421 static void erlang_attribute (char *);
5422 static int erlang_atom (char *);
5424 static void
5425 Erlang_functions (FILE *inf)
5427 char *cp, *last;
5428 int len;
5429 int allocated;
5431 allocated = 0;
5432 len = 0;
5433 last = NULL;
5435 LOOP_ON_INPUT_LINES (inf, lb, cp)
5437 if (cp[0] == '\0') /* Empty line */
5438 continue;
5439 else if (iswhite (cp[0])) /* Not function nor attribute */
5440 continue;
5441 else if (cp[0] == '%') /* comment */
5442 continue;
5443 else if (cp[0] == '"') /* Sometimes, strings start in column one */
5444 continue;
5445 else if (cp[0] == '-') /* attribute, e.g. "-define" */
5447 erlang_attribute (cp);
5448 if (last != NULL)
5450 free (last);
5451 last = NULL;
5454 else if ((len = erlang_func (cp, last)) > 0)
5457 * Function. Store the function name so that we only
5458 * generates a tag for the first clause.
5460 if (last == NULL)
5461 last = xnew (len + 1, char);
5462 else if (len + 1 > allocated)
5463 xrnew (last, len + 1, char);
5464 allocated = len + 1;
5465 strncpy (last, cp, len);
5466 last[len] = '\0';
5469 free (last);
5474 * A function definition is added if it matches:
5475 * <beginning of line><Erlang Atom><whitespace>(
5477 * It is added to the tags database if it doesn't match the
5478 * name of the previous clause header.
5480 * Return the size of the name of the function, or 0 if no function
5481 * was found.
5483 static int
5484 erlang_func (char *s, char *last)
5486 /* Name of last clause. */
5488 int pos;
5489 int len;
5491 pos = erlang_atom (s);
5492 if (pos < 1)
5493 return 0;
5495 len = pos;
5496 pos = skip_spaces (s + pos) - s;
5498 /* Save only the first clause. */
5499 if (s[pos++] == '('
5500 && (last == NULL
5501 || len != (int)strlen (last)
5502 || !strneq (s, last, len)))
5504 make_tag (s, len, TRUE, s, pos, lineno, linecharno);
5505 return len;
5508 return 0;
5513 * Handle attributes. Currently, tags are generated for defines
5514 * and records.
5516 * They are on the form:
5517 * -define(foo, bar).
5518 * -define(Foo(M, N), M+N).
5519 * -record(graph, {vtab = notable, cyclic = true}).
5521 static void
5522 erlang_attribute (char *s)
5524 char *cp = s;
5526 if ((LOOKING_AT (cp, "-define") || LOOKING_AT (cp, "-record"))
5527 && *cp++ == '(')
5529 int len = erlang_atom (skip_spaces (cp));
5530 if (len > 0)
5531 make_tag (cp, len, TRUE, s, cp + len - s, lineno, linecharno);
5533 return;
5538 * Consume an Erlang atom (or variable).
5539 * Return the number of bytes consumed, or -1 if there was an error.
5541 static int
5542 erlang_atom (char *s)
5544 int pos = 0;
5546 if (ISALPHA (s[pos]) || s[pos] == '_')
5548 /* The atom is unquoted. */
5550 pos++;
5551 while (ISALNUM (s[pos]) || s[pos] == '_');
5553 else if (s[pos] == '\'')
5555 for (pos++; s[pos] != '\''; pos++)
5556 if (s[pos] == '\0' /* multiline quoted atoms are ignored */
5557 || (s[pos] == '\\' && s[++pos] == '\0'))
5558 return 0;
5559 pos++;
5562 return pos;
5566 static char *scan_separators (char *);
5567 static void add_regex (char *, language *);
5568 static char *substitute (char *, char *, struct re_registers *);
5571 * Take a string like "/blah/" and turn it into "blah", verifying
5572 * that the first and last characters are the same, and handling
5573 * quoted separator characters. Actually, stops on the occurrence of
5574 * an unquoted separator. Also process \t, \n, etc. and turn into
5575 * appropriate characters. Works in place. Null terminates name string.
5576 * Returns pointer to terminating separator, or NULL for
5577 * unterminated regexps.
5579 static char *
5580 scan_separators (char *name)
5582 char sep = name[0];
5583 char *copyto = name;
5584 bool quoted = FALSE;
5586 for (++name; *name != '\0'; ++name)
5588 if (quoted)
5590 switch (*name)
5592 case 'a': *copyto++ = '\007'; break; /* BEL (bell) */
5593 case 'b': *copyto++ = '\b'; break; /* BS (back space) */
5594 case 'd': *copyto++ = 0177; break; /* DEL (delete) */
5595 case 'e': *copyto++ = 033; break; /* ESC (delete) */
5596 case 'f': *copyto++ = '\f'; break; /* FF (form feed) */
5597 case 'n': *copyto++ = '\n'; break; /* NL (new line) */
5598 case 'r': *copyto++ = '\r'; break; /* CR (carriage return) */
5599 case 't': *copyto++ = '\t'; break; /* TAB (horizontal tab) */
5600 case 'v': *copyto++ = '\v'; break; /* VT (vertical tab) */
5601 default:
5602 if (*name == sep)
5603 *copyto++ = sep;
5604 else
5606 /* Something else is quoted, so preserve the quote. */
5607 *copyto++ = '\\';
5608 *copyto++ = *name;
5610 break;
5612 quoted = FALSE;
5614 else if (*name == '\\')
5615 quoted = TRUE;
5616 else if (*name == sep)
5617 break;
5618 else
5619 *copyto++ = *name;
5621 if (*name != sep)
5622 name = NULL; /* signal unterminated regexp */
5624 /* Terminate copied string. */
5625 *copyto = '\0';
5626 return name;
5629 /* Look at the argument of --regex or --no-regex and do the right
5630 thing. Same for each line of a regexp file. */
5631 static void
5632 analyse_regex (char *regex_arg)
5634 if (regex_arg == NULL)
5636 free_regexps (); /* --no-regex: remove existing regexps */
5637 return;
5640 /* A real --regexp option or a line in a regexp file. */
5641 switch (regex_arg[0])
5643 /* Comments in regexp file or null arg to --regex. */
5644 case '\0':
5645 case ' ':
5646 case '\t':
5647 break;
5649 /* Read a regex file. This is recursive and may result in a
5650 loop, which will stop when the file descriptors are exhausted. */
5651 case '@':
5653 FILE *regexfp;
5654 linebuffer regexbuf;
5655 char *regexfile = regex_arg + 1;
5657 /* regexfile is a file containing regexps, one per line. */
5658 regexfp = fopen (regexfile, "r");
5659 if (regexfp == NULL)
5661 pfatal (regexfile);
5662 return;
5664 linebuffer_init (&regexbuf);
5665 while (readline_internal (&regexbuf, regexfp) > 0)
5666 analyse_regex (regexbuf.buffer);
5667 free (regexbuf.buffer);
5668 fclose (regexfp);
5670 break;
5672 /* Regexp to be used for a specific language only. */
5673 case '{':
5675 language *lang;
5676 char *lang_name = regex_arg + 1;
5677 char *cp;
5679 for (cp = lang_name; *cp != '}'; cp++)
5680 if (*cp == '\0')
5682 error ("unterminated language name in regex: %s", regex_arg);
5683 return;
5685 *cp++ = '\0';
5686 lang = get_language_from_langname (lang_name);
5687 if (lang == NULL)
5688 return;
5689 add_regex (cp, lang);
5691 break;
5693 /* Regexp to be used for any language. */
5694 default:
5695 add_regex (regex_arg, NULL);
5696 break;
5700 /* Separate the regexp pattern, compile it,
5701 and care for optional name and modifiers. */
5702 static void
5703 add_regex (char *regexp_pattern, language *lang)
5705 static struct re_pattern_buffer zeropattern;
5706 char sep, *pat, *name, *modifiers;
5707 char empty[] = "";
5708 const char *err;
5709 struct re_pattern_buffer *patbuf;
5710 regexp *rp;
5711 bool
5712 force_explicit_name = TRUE, /* do not use implicit tag names */
5713 ignore_case = FALSE, /* case is significant */
5714 multi_line = FALSE, /* matches are done one line at a time */
5715 single_line = FALSE; /* dot does not match newline */
5718 if (strlen (regexp_pattern) < 3)
5720 error ("null regexp", (char *)NULL);
5721 return;
5723 sep = regexp_pattern[0];
5724 name = scan_separators (regexp_pattern);
5725 if (name == NULL)
5727 error ("%s: unterminated regexp", regexp_pattern);
5728 return;
5730 if (name[1] == sep)
5732 error ("null name for regexp \"%s\"", regexp_pattern);
5733 return;
5735 modifiers = scan_separators (name);
5736 if (modifiers == NULL) /* no terminating separator --> no name */
5738 modifiers = name;
5739 name = empty;
5741 else
5742 modifiers += 1; /* skip separator */
5744 /* Parse regex modifiers. */
5745 for (; modifiers[0] != '\0'; modifiers++)
5746 switch (modifiers[0])
5748 case 'N':
5749 if (modifiers == name)
5750 error ("forcing explicit tag name but no name, ignoring", NULL);
5751 force_explicit_name = TRUE;
5752 break;
5753 case 'i':
5754 ignore_case = TRUE;
5755 break;
5756 case 's':
5757 single_line = TRUE;
5758 /* FALLTHRU */
5759 case 'm':
5760 multi_line = TRUE;
5761 need_filebuf = TRUE;
5762 break;
5763 default:
5765 char wrongmod [2];
5766 wrongmod[0] = modifiers[0];
5767 wrongmod[1] = '\0';
5768 error ("invalid regexp modifier `%s', ignoring", wrongmod);
5770 break;
5773 patbuf = xnew (1, struct re_pattern_buffer);
5774 *patbuf = zeropattern;
5775 if (ignore_case)
5777 static char lc_trans[CHARS];
5778 int i;
5779 for (i = 0; i < CHARS; i++)
5780 lc_trans[i] = lowcase (i);
5781 patbuf->translate = lc_trans; /* translation table to fold case */
5784 if (multi_line)
5785 pat = concat ("^", regexp_pattern, ""); /* anchor to beginning of line */
5786 else
5787 pat = regexp_pattern;
5789 if (single_line)
5790 re_set_syntax (RE_SYNTAX_EMACS | RE_DOT_NEWLINE);
5791 else
5792 re_set_syntax (RE_SYNTAX_EMACS);
5794 err = re_compile_pattern (pat, strlen (pat), patbuf);
5795 if (multi_line)
5796 free (pat);
5797 if (err != NULL)
5799 error ("%s while compiling pattern", err);
5800 return;
5803 rp = p_head;
5804 p_head = xnew (1, regexp);
5805 p_head->pattern = savestr (regexp_pattern);
5806 p_head->p_next = rp;
5807 p_head->lang = lang;
5808 p_head->pat = patbuf;
5809 p_head->name = savestr (name);
5810 p_head->error_signaled = FALSE;
5811 p_head->force_explicit_name = force_explicit_name;
5812 p_head->ignore_case = ignore_case;
5813 p_head->multi_line = multi_line;
5817 * Do the substitutions indicated by the regular expression and
5818 * arguments.
5820 static char *
5821 substitute (char *in, char *out, struct re_registers *regs)
5823 char *result, *t;
5824 int size, dig, diglen;
5826 result = NULL;
5827 size = strlen (out);
5829 /* Pass 1: figure out how much to allocate by finding all \N strings. */
5830 if (out[size - 1] == '\\')
5831 fatal ("pattern error in \"%s\"", out);
5832 for (t = etags_strchr (out, '\\');
5833 t != NULL;
5834 t = etags_strchr (t + 2, '\\'))
5835 if (ISDIGIT (t[1]))
5837 dig = t[1] - '0';
5838 diglen = regs->end[dig] - regs->start[dig];
5839 size += diglen - 2;
5841 else
5842 size -= 1;
5844 /* Allocate space and do the substitutions. */
5845 assert (size >= 0);
5846 result = xnew (size + 1, char);
5848 for (t = result; *out != '\0'; out++)
5849 if (*out == '\\' && ISDIGIT (*++out))
5851 dig = *out - '0';
5852 diglen = regs->end[dig] - regs->start[dig];
5853 strncpy (t, in + regs->start[dig], diglen);
5854 t += diglen;
5856 else
5857 *t++ = *out;
5858 *t = '\0';
5860 assert (t <= result + size);
5861 assert (t - result == (int)strlen (result));
5863 return result;
5866 /* Deallocate all regexps. */
5867 static void
5868 free_regexps (void)
5870 regexp *rp;
5871 while (p_head != NULL)
5873 rp = p_head->p_next;
5874 free (p_head->pattern);
5875 free (p_head->name);
5876 free (p_head);
5877 p_head = rp;
5879 return;
5883 * Reads the whole file as a single string from `filebuf' and looks for
5884 * multi-line regular expressions, creating tags on matches.
5885 * readline already dealt with normal regexps.
5887 * Idea by Ben Wing <ben@666.com> (2002).
5889 static void
5890 regex_tag_multiline (void)
5892 char *buffer = filebuf.buffer;
5893 regexp *rp;
5894 char *name;
5896 for (rp = p_head; rp != NULL; rp = rp->p_next)
5898 int match = 0;
5900 if (!rp->multi_line)
5901 continue; /* skip normal regexps */
5903 /* Generic initialisations before parsing file from memory. */
5904 lineno = 1; /* reset global line number */
5905 charno = 0; /* reset global char number */
5906 linecharno = 0; /* reset global char number of line start */
5908 /* Only use generic regexps or those for the current language. */
5909 if (rp->lang != NULL && rp->lang != curfdp->lang)
5910 continue;
5912 while (match >= 0 && match < filebuf.len)
5914 match = re_search (rp->pat, buffer, filebuf.len, charno,
5915 filebuf.len - match, &rp->regs);
5916 switch (match)
5918 case -2:
5919 /* Some error. */
5920 if (!rp->error_signaled)
5922 error ("regexp stack overflow while matching \"%s\"",
5923 rp->pattern);
5924 rp->error_signaled = TRUE;
5926 break;
5927 case -1:
5928 /* No match. */
5929 break;
5930 default:
5931 if (match == rp->regs.end[0])
5933 if (!rp->error_signaled)
5935 error ("regexp matches the empty string: \"%s\"",
5936 rp->pattern);
5937 rp->error_signaled = TRUE;
5939 match = -3; /* exit from while loop */
5940 break;
5943 /* Match occurred. Construct a tag. */
5944 while (charno < rp->regs.end[0])
5945 if (buffer[charno++] == '\n')
5946 lineno++, linecharno = charno;
5947 name = rp->name;
5948 if (name[0] == '\0')
5949 name = NULL;
5950 else /* make a named tag */
5951 name = substitute (buffer, rp->name, &rp->regs);
5952 if (rp->force_explicit_name)
5953 /* Force explicit tag name, if a name is there. */
5954 pfnote (name, TRUE, buffer + linecharno,
5955 charno - linecharno + 1, lineno, linecharno);
5956 else
5957 make_tag (name, strlen (name), TRUE, buffer + linecharno,
5958 charno - linecharno + 1, lineno, linecharno);
5959 break;
5966 static bool
5967 nocase_tail (const char *cp)
5969 register int len = 0;
5971 while (*cp != '\0' && lowcase (*cp) == lowcase (dbp[len]))
5972 cp++, len++;
5973 if (*cp == '\0' && !intoken (dbp[len]))
5975 dbp += len;
5976 return TRUE;
5978 return FALSE;
5981 static void
5982 get_tag (register char *bp, char **namepp)
5984 register char *cp = bp;
5986 if (*bp != '\0')
5988 /* Go till you get to white space or a syntactic break */
5989 for (cp = bp + 1; !notinname (*cp); cp++)
5990 continue;
5991 make_tag (bp, cp - bp, TRUE,
5992 lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
5995 if (namepp != NULL)
5996 *namepp = savenstr (bp, cp - bp);
6000 * Read a line of text from `stream' into `lbp', excluding the
6001 * newline or CR-NL, if any. Return the number of characters read from
6002 * `stream', which is the length of the line including the newline.
6004 * On DOS or Windows we do not count the CR character, if any before the
6005 * NL, in the returned length; this mirrors the behavior of Emacs on those
6006 * platforms (for text files, it translates CR-NL to NL as it reads in the
6007 * file).
6009 * If multi-line regular expressions are requested, each line read is
6010 * appended to `filebuf'.
6012 static long
6013 readline_internal (linebuffer *lbp, register FILE *stream)
6015 char *buffer = lbp->buffer;
6016 register char *p = lbp->buffer;
6017 register char *pend;
6018 int chars_deleted;
6020 pend = p + lbp->size; /* Separate to avoid 386/IX compiler bug. */
6022 for (;;)
6024 register int c = getc (stream);
6025 if (p == pend)
6027 /* We're at the end of linebuffer: expand it. */
6028 lbp->size *= 2;
6029 xrnew (buffer, lbp->size, char);
6030 p += buffer - lbp->buffer;
6031 pend = buffer + lbp->size;
6032 lbp->buffer = buffer;
6034 if (c == EOF)
6036 *p = '\0';
6037 chars_deleted = 0;
6038 break;
6040 if (c == '\n')
6042 if (p > buffer && p[-1] == '\r')
6044 p -= 1;
6045 #ifdef DOS_NT
6046 /* Assume CRLF->LF translation will be performed by Emacs
6047 when loading this file, so CRs won't appear in the buffer.
6048 It would be cleaner to compensate within Emacs;
6049 however, Emacs does not know how many CRs were deleted
6050 before any given point in the file. */
6051 chars_deleted = 1;
6052 #else
6053 chars_deleted = 2;
6054 #endif
6056 else
6058 chars_deleted = 1;
6060 *p = '\0';
6061 break;
6063 *p++ = c;
6065 lbp->len = p - buffer;
6067 if (need_filebuf /* we need filebuf for multi-line regexps */
6068 && chars_deleted > 0) /* not at EOF */
6070 while (filebuf.size <= filebuf.len + lbp->len + 1) /* +1 for \n */
6072 /* Expand filebuf. */
6073 filebuf.size *= 2;
6074 xrnew (filebuf.buffer, filebuf.size, char);
6076 strncpy (filebuf.buffer + filebuf.len, lbp->buffer, lbp->len);
6077 filebuf.len += lbp->len;
6078 filebuf.buffer[filebuf.len++] = '\n';
6079 filebuf.buffer[filebuf.len] = '\0';
6082 return lbp->len + chars_deleted;
6086 * Like readline_internal, above, but in addition try to match the
6087 * input line against relevant regular expressions and manage #line
6088 * directives.
6090 static void
6091 readline (linebuffer *lbp, FILE *stream)
6093 long result;
6095 linecharno = charno; /* update global char number of line start */
6096 result = readline_internal (lbp, stream); /* read line */
6097 lineno += 1; /* increment global line number */
6098 charno += result; /* increment global char number */
6100 /* Honor #line directives. */
6101 if (!no_line_directive)
6103 static bool discard_until_line_directive;
6105 /* Check whether this is a #line directive. */
6106 if (result > 12 && strneq (lbp->buffer, "#line ", 6))
6108 unsigned int lno;
6109 int start = 0;
6111 if (sscanf (lbp->buffer, "#line %u \"%n", &lno, &start) >= 1
6112 && start > 0) /* double quote character found */
6114 char *endp = lbp->buffer + start;
6116 while ((endp = etags_strchr (endp, '"')) != NULL
6117 && endp[-1] == '\\')
6118 endp++;
6119 if (endp != NULL)
6120 /* Ok, this is a real #line directive. Let's deal with it. */
6122 char *taggedabsname; /* absolute name of original file */
6123 char *taggedfname; /* name of original file as given */
6124 char *name; /* temp var */
6126 discard_until_line_directive = FALSE; /* found it */
6127 name = lbp->buffer + start;
6128 *endp = '\0';
6129 canonicalize_filename (name);
6130 taggedabsname = absolute_filename (name, tagfiledir);
6131 if (filename_is_absolute (name)
6132 || filename_is_absolute (curfdp->infname))
6133 taggedfname = savestr (taggedabsname);
6134 else
6135 taggedfname = relative_filename (taggedabsname,tagfiledir);
6137 if (streq (curfdp->taggedfname, taggedfname))
6138 /* The #line directive is only a line number change. We
6139 deal with this afterwards. */
6140 free (taggedfname);
6141 else
6142 /* The tags following this #line directive should be
6143 attributed to taggedfname. In order to do this, set
6144 curfdp accordingly. */
6146 fdesc *fdp; /* file description pointer */
6148 /* Go look for a file description already set up for the
6149 file indicated in the #line directive. If there is
6150 one, use it from now until the next #line
6151 directive. */
6152 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
6153 if (streq (fdp->infname, curfdp->infname)
6154 && streq (fdp->taggedfname, taggedfname))
6155 /* If we remove the second test above (after the &&)
6156 then all entries pertaining to the same file are
6157 coalesced in the tags file. If we use it, then
6158 entries pertaining to the same file but generated
6159 from different files (via #line directives) will
6160 go into separate sections in the tags file. These
6161 alternatives look equivalent. The first one
6162 destroys some apparently useless information. */
6164 curfdp = fdp;
6165 free (taggedfname);
6166 break;
6168 /* Else, if we already tagged the real file, skip all
6169 input lines until the next #line directive. */
6170 if (fdp == NULL) /* not found */
6171 for (fdp = fdhead; fdp != NULL; fdp = fdp->next)
6172 if (streq (fdp->infabsname, taggedabsname))
6174 discard_until_line_directive = TRUE;
6175 free (taggedfname);
6176 break;
6178 /* Else create a new file description and use that from
6179 now on, until the next #line directive. */
6180 if (fdp == NULL) /* not found */
6182 fdp = fdhead;
6183 fdhead = xnew (1, fdesc);
6184 *fdhead = *curfdp; /* copy curr. file description */
6185 fdhead->next = fdp;
6186 fdhead->infname = savestr (curfdp->infname);
6187 fdhead->infabsname = savestr (curfdp->infabsname);
6188 fdhead->infabsdir = savestr (curfdp->infabsdir);
6189 fdhead->taggedfname = taggedfname;
6190 fdhead->usecharno = FALSE;
6191 fdhead->prop = NULL;
6192 fdhead->written = FALSE;
6193 curfdp = fdhead;
6196 free (taggedabsname);
6197 lineno = lno - 1;
6198 readline (lbp, stream);
6199 return;
6200 } /* if a real #line directive */
6201 } /* if #line is followed by a number */
6202 } /* if line begins with "#line " */
6204 /* If we are here, no #line directive was found. */
6205 if (discard_until_line_directive)
6207 if (result > 0)
6209 /* Do a tail recursion on ourselves, thus discarding the contents
6210 of the line buffer. */
6211 readline (lbp, stream);
6212 return;
6214 /* End of file. */
6215 discard_until_line_directive = FALSE;
6216 return;
6218 } /* if #line directives should be considered */
6221 int match;
6222 regexp *rp;
6223 char *name;
6225 /* Match against relevant regexps. */
6226 if (lbp->len > 0)
6227 for (rp = p_head; rp != NULL; rp = rp->p_next)
6229 /* Only use generic regexps or those for the current language.
6230 Also do not use multiline regexps, which is the job of
6231 regex_tag_multiline. */
6232 if ((rp->lang != NULL && rp->lang != fdhead->lang)
6233 || rp->multi_line)
6234 continue;
6236 match = re_match (rp->pat, lbp->buffer, lbp->len, 0, &rp->regs);
6237 switch (match)
6239 case -2:
6240 /* Some error. */
6241 if (!rp->error_signaled)
6243 error ("regexp stack overflow while matching \"%s\"",
6244 rp->pattern);
6245 rp->error_signaled = TRUE;
6247 break;
6248 case -1:
6249 /* No match. */
6250 break;
6251 case 0:
6252 /* Empty string matched. */
6253 if (!rp->error_signaled)
6255 error ("regexp matches the empty string: \"%s\"", rp->pattern);
6256 rp->error_signaled = TRUE;
6258 break;
6259 default:
6260 /* Match occurred. Construct a tag. */
6261 name = rp->name;
6262 if (name[0] == '\0')
6263 name = NULL;
6264 else /* make a named tag */
6265 name = substitute (lbp->buffer, rp->name, &rp->regs);
6266 if (rp->force_explicit_name)
6267 /* Force explicit tag name, if a name is there. */
6268 pfnote (name, TRUE, lbp->buffer, match, lineno, linecharno);
6269 else
6270 make_tag (name, strlen (name), TRUE,
6271 lbp->buffer, match, lineno, linecharno);
6272 break;
6280 * Return a pointer to a space of size strlen(cp)+1 allocated
6281 * with xnew where the string CP has been copied.
6283 static char *
6284 savestr (const char *cp)
6286 return savenstr (cp, strlen (cp));
6290 * Return a pointer to a space of size LEN+1 allocated with xnew where
6291 * the string CP has been copied for at most the first LEN characters.
6293 static char *
6294 savenstr (const char *cp, int len)
6296 register char *dp;
6298 dp = xnew (len + 1, char);
6299 strncpy (dp, cp, len);
6300 dp[len] = '\0';
6301 return dp;
6305 * Return the ptr in sp at which the character c last
6306 * appears; NULL if not found
6308 * Identical to POSIX strrchr, included for portability.
6310 static char *
6311 etags_strrchr (register const char *sp, register int c)
6313 register const char *r;
6315 r = NULL;
6318 if (*sp == c)
6319 r = sp;
6320 } while (*sp++);
6321 return (char *)r;
6325 * Return the ptr in sp at which the character c first
6326 * appears; NULL if not found
6328 * Identical to POSIX strchr, included for portability.
6330 static char *
6331 etags_strchr (register const char *sp, register int c)
6335 if (*sp == c)
6336 return (char *)sp;
6337 } while (*sp++);
6338 return NULL;
6342 * Compare two strings, ignoring case for alphabetic characters.
6344 * Same as BSD's strcasecmp, included for portability.
6346 static int
6347 etags_strcasecmp (register const char *s1, register const char *s2)
6349 while (*s1 != '\0'
6350 && (ISALPHA (*s1) && ISALPHA (*s2)
6351 ? lowcase (*s1) == lowcase (*s2)
6352 : *s1 == *s2))
6353 s1++, s2++;
6355 return (ISALPHA (*s1) && ISALPHA (*s2)
6356 ? lowcase (*s1) - lowcase (*s2)
6357 : *s1 - *s2);
6361 * Compare two strings, ignoring case for alphabetic characters.
6362 * Stop after a given number of characters
6364 * Same as BSD's strncasecmp, included for portability.
6366 static int
6367 etags_strncasecmp (register const char *s1, register const char *s2, register int n)
6369 while (*s1 != '\0' && n-- > 0
6370 && (ISALPHA (*s1) && ISALPHA (*s2)
6371 ? lowcase (*s1) == lowcase (*s2)
6372 : *s1 == *s2))
6373 s1++, s2++;
6375 if (n < 0)
6376 return 0;
6377 else
6378 return (ISALPHA (*s1) && ISALPHA (*s2)
6379 ? lowcase (*s1) - lowcase (*s2)
6380 : *s1 - *s2);
6383 /* Skip spaces (end of string is not space), return new pointer. */
6384 static char *
6385 skip_spaces (char *cp)
6387 while (iswhite (*cp))
6388 cp++;
6389 return cp;
6392 /* Skip non spaces, except end of string, return new pointer. */
6393 static char *
6394 skip_non_spaces (char *cp)
6396 while (*cp != '\0' && !iswhite (*cp))
6397 cp++;
6398 return cp;
6401 /* Print error message and exit. */
6402 void
6403 fatal (const char *s1, const char *s2)
6405 error (s1, s2);
6406 exit (EXIT_FAILURE);
6409 static void
6410 pfatal (const char *s1)
6412 perror (s1);
6413 exit (EXIT_FAILURE);
6416 static void
6417 suggest_asking_for_help (void)
6419 fprintf (stderr, "\tTry `%s %s' for a complete list of options.\n",
6420 progname, NO_LONG_OPTIONS ? "-h" : "--help");
6421 exit (EXIT_FAILURE);
6424 /* Print error message. `s1' is printf control string, `s2' is arg for it. */
6425 static void
6426 error (const char *s1, const char *s2)
6428 fprintf (stderr, "%s: ", progname);
6429 fprintf (stderr, s1, s2);
6430 fprintf (stderr, "\n");
6433 /* Return a newly-allocated string whose contents
6434 concatenate those of s1, s2, s3. */
6435 static char *
6436 concat (const char *s1, const char *s2, const char *s3)
6438 int len1 = strlen (s1), len2 = strlen (s2), len3 = strlen (s3);
6439 char *result = xnew (len1 + len2 + len3 + 1, char);
6441 strcpy (result, s1);
6442 strcpy (result + len1, s2);
6443 strcpy (result + len1 + len2, s3);
6444 result[len1 + len2 + len3] = '\0';
6446 return result;
6450 /* Does the same work as the system V getcwd, but does not need to
6451 guess the buffer size in advance. */
6452 static char *
6453 etags_getcwd (void)
6455 #ifdef HAVE_GETCWD
6456 int bufsize = 200;
6457 char *path = xnew (bufsize, char);
6459 while (getcwd (path, bufsize) == NULL)
6461 if (errno != ERANGE)
6462 pfatal ("getcwd");
6463 bufsize *= 2;
6464 free (path);
6465 path = xnew (bufsize, char);
6468 canonicalize_filename (path);
6469 return path;
6471 #else /* not HAVE_GETCWD */
6472 #if MSDOS
6474 char *p, path[MAXPATHLEN + 1]; /* Fixed size is safe on MSDOS. */
6476 getwd (path);
6478 for (p = path; *p != '\0'; p++)
6479 if (*p == '\\')
6480 *p = '/';
6481 else
6482 *p = lowcase (*p);
6484 return strdup (path);
6485 #else /* not MSDOS */
6486 linebuffer path;
6487 FILE *pipe;
6489 linebuffer_init (&path);
6490 pipe = (FILE *) popen ("pwd 2>/dev/null", "r");
6491 if (pipe == NULL || readline_internal (&path, pipe) == 0)
6492 pfatal ("pwd");
6493 pclose (pipe);
6495 return path.buffer;
6496 #endif /* not MSDOS */
6497 #endif /* not HAVE_GETCWD */
6500 /* Return a newly allocated string containing the file name of FILE
6501 relative to the absolute directory DIR (which should end with a slash). */
6502 static char *
6503 relative_filename (char *file, char *dir)
6505 char *fp, *dp, *afn, *res;
6506 int i;
6508 /* Find the common root of file and dir (with a trailing slash). */
6509 afn = absolute_filename (file, cwd);
6510 fp = afn;
6511 dp = dir;
6512 while (*fp++ == *dp++)
6513 continue;
6514 fp--, dp--; /* back to the first differing char */
6515 #ifdef DOS_NT
6516 if (fp == afn && afn[0] != '/') /* cannot build a relative name */
6517 return afn;
6518 #endif
6519 do /* look at the equal chars until '/' */
6520 fp--, dp--;
6521 while (*fp != '/');
6523 /* Build a sequence of "../" strings for the resulting relative file name. */
6524 i = 0;
6525 while ((dp = etags_strchr (dp + 1, '/')) != NULL)
6526 i += 1;
6527 res = xnew (3*i + strlen (fp + 1) + 1, char);
6528 res[0] = '\0';
6529 while (i-- > 0)
6530 strcat (res, "../");
6532 /* Add the file name relative to the common root of file and dir. */
6533 strcat (res, fp + 1);
6534 free (afn);
6536 return res;
6539 /* Return a newly allocated string containing the absolute file name
6540 of FILE given DIR (which should end with a slash). */
6541 static char *
6542 absolute_filename (char *file, char *dir)
6544 char *slashp, *cp, *res;
6546 if (filename_is_absolute (file))
6547 res = savestr (file);
6548 #ifdef DOS_NT
6549 /* We don't support non-absolute file names with a drive
6550 letter, like `d:NAME' (it's too much hassle). */
6551 else if (file[1] == ':')
6552 fatal ("%s: relative file names with drive letters not supported", file);
6553 #endif
6554 else
6555 res = concat (dir, file, "");
6557 /* Delete the "/dirname/.." and "/." substrings. */
6558 slashp = etags_strchr (res, '/');
6559 while (slashp != NULL && slashp[0] != '\0')
6561 if (slashp[1] == '.')
6563 if (slashp[2] == '.'
6564 && (slashp[3] == '/' || slashp[3] == '\0'))
6566 cp = slashp;
6568 cp--;
6569 while (cp >= res && !filename_is_absolute (cp));
6570 if (cp < res)
6571 cp = slashp; /* the absolute name begins with "/.." */
6572 #ifdef DOS_NT
6573 /* Under MSDOS and NT we get `d:/NAME' as absolute
6574 file name, so the luser could say `d:/../NAME'.
6575 We silently treat this as `d:/NAME'. */
6576 else if (cp[0] != '/')
6577 cp = slashp;
6578 #endif
6579 memmove (cp, slashp + 3, strlen (slashp + 2));
6580 slashp = cp;
6581 continue;
6583 else if (slashp[2] == '/' || slashp[2] == '\0')
6585 memmove (slashp, slashp + 2, strlen (slashp + 1));
6586 continue;
6590 slashp = etags_strchr (slashp + 1, '/');
6593 if (res[0] == '\0') /* just a safety net: should never happen */
6595 free (res);
6596 return savestr ("/");
6598 else
6599 return res;
6602 /* Return a newly allocated string containing the absolute
6603 file name of dir where FILE resides given DIR (which should
6604 end with a slash). */
6605 static char *
6606 absolute_dirname (char *file, char *dir)
6608 char *slashp, *res;
6609 char save;
6611 slashp = etags_strrchr (file, '/');
6612 if (slashp == NULL)
6613 return savestr (dir);
6614 save = slashp[1];
6615 slashp[1] = '\0';
6616 res = absolute_filename (file, dir);
6617 slashp[1] = save;
6619 return res;
6622 /* Whether the argument string is an absolute file name. The argument
6623 string must have been canonicalized with canonicalize_filename. */
6624 static bool
6625 filename_is_absolute (char *fn)
6627 return (fn[0] == '/'
6628 #ifdef DOS_NT
6629 || (ISALPHA (fn[0]) && fn[1] == ':' && fn[2] == '/')
6630 #endif
6634 /* Downcase DOS drive letter and collapse separators into single slashes.
6635 Works in place. */
6636 static void
6637 canonicalize_filename (register char *fn)
6639 register char* cp;
6640 char sep = '/';
6642 #ifdef DOS_NT
6643 /* Canonicalize drive letter case. */
6644 # define ISUPPER(c) isupper (CHAR (c))
6645 if (fn[0] != '\0' && fn[1] == ':' && ISUPPER (fn[0]))
6646 fn[0] = lowcase (fn[0]);
6648 sep = '\\';
6649 #endif
6651 /* Collapse multiple separators into a single slash. */
6652 for (cp = fn; *cp != '\0'; cp++, fn++)
6653 if (*cp == sep)
6655 *fn = '/';
6656 while (cp[1] == sep)
6657 cp++;
6659 else
6660 *fn = *cp;
6661 *fn = '\0';
6665 /* Initialize a linebuffer for use. */
6666 static void
6667 linebuffer_init (linebuffer *lbp)
6669 lbp->size = (DEBUG) ? 3 : 200;
6670 lbp->buffer = xnew (lbp->size, char);
6671 lbp->buffer[0] = '\0';
6672 lbp->len = 0;
6675 /* Set the minimum size of a string contained in a linebuffer. */
6676 static void
6677 linebuffer_setlen (linebuffer *lbp, int toksize)
6679 while (lbp->size <= toksize)
6681 lbp->size *= 2;
6682 xrnew (lbp->buffer, lbp->size, char);
6684 lbp->len = toksize;
6687 /* Like malloc but get fatal error if memory is exhausted. */
6688 static PTR
6689 xmalloc (size_t size)
6691 PTR result = (PTR) malloc (size);
6692 if (result == NULL)
6693 fatal ("virtual memory exhausted", (char *)NULL);
6694 return result;
6697 static PTR
6698 xrealloc (char *ptr, size_t size)
6700 PTR result = (PTR) realloc (ptr, size);
6701 if (result == NULL)
6702 fatal ("virtual memory exhausted", (char *)NULL);
6703 return result;
6707 * Local Variables:
6708 * indent-tabs-mode: t
6709 * tab-width: 8
6710 * fill-column: 79
6711 * c-font-lock-extra-types: ("FILE" "bool" "language" "linebuffer" "fdesc" "node" "regexp")
6712 * c-file-style: "gnu"
6713 * End:
6716 /* etags.c ends here */