1 /* Tags file maker to go with GNU Emacs
2 Copyright (C) 1984, 87, 88, 89, 93, 94, 95
3 Free Software Foundation, Inc. and Ken Arnold
5 This file is not considered part of GNU Emacs.
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2 of the License, or
10 (at your option) any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software Foundation,
19 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
23 * Ctags originally by Ken Arnold.
24 * Fortran added by Jim Kleckner.
25 * Ed Pelegri-Llopart added C typedefs.
26 * Gnu Emacs TAGS format and modifications by RMS?
27 * Sam Kendall added C++.
28 * Francesco Potorti` reorganised C and C++ based on work by Joe Wells.
29 * Regexp tags by Tom Tromey.
31 * Francesco Potorti` (F.Potorti@cnuce.cnr.it) is the current maintainer.
34 char pot_etags_version
[] = "@(#) pot revision number is 11.59";
45 # include <sys/param.h>
52 # define MAXPATHLEN _MAX_PATH
57 /* On some systems, Emacs defines static as nothing for the sake
58 of unexec. We don't want that here since we don't use unexec. */
68 #include <sys/types.h>
71 #if !defined (S_ISREG) && defined (S_IFREG)
72 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
79 #endif /* ETAGS_REGEXPS */
81 /* Define CTAGS to make the program "ctags" compatible with the usual one.
82 Let it undefined to make the program "etags", which makes emacs-style
83 tag tables and tags typedefs, #defines and struct/union/enum by default. */
91 /* Exit codes for success and failure. */
101 #define C_PLPL 0x00001 /* C++ */
102 #define C_STAR 0x00003 /* C* */
103 #define YACC 0x10000 /* yacc file */
105 #define streq(s,t) ((DEBUG &&!(s)&&!(t)&&(abort(),1)) || !strcmp(s,t))
106 #define strneq(s,t,n) ((DEBUG &&!(s)&&!(t)&&(abort(),1)) || !strncmp(s,t,n))
108 #define lowcase(c) tolower ((unsigned char)c)
110 #define iswhite(arg) (_wht[arg]) /* T if char is white */
111 #define begtoken(arg) (_btk[arg]) /* T if char can start token */
112 #define intoken(arg) (_itk[arg]) /* T if char can be in token */
113 #define endtoken(arg) (_etk[arg]) /* T if char ends tokens */
116 # define absolutefn(fn) (fn[0] == '/' \
117 || (isalpha (fn[0]) && fn[1] == ':' && fn[2] == '/'))
119 # define absolutefn(fn) (fn[0] == '/')
124 * xnew -- allocate storage
126 * SYNOPSIS: Type *xnew (int n, Type);
128 #define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type)))
133 { /* sorting structure */
134 char *name
; /* function or type name */
135 char *file
; /* file name */
136 logical is_func
; /* use pattern or line no */
137 logical been_warned
; /* set if noticed dup */
138 int lno
; /* line number tag is on */
139 long cno
; /* character number line starts on */
140 char *pat
; /* search pattern */
141 struct nd_st
*left
, *right
; /* left and right sons */
144 extern char *getenv ();
147 char *savenstr (), *savestr ();
148 char *etags_strchr (), *etags_strrchr ();
149 char *etags_getcwd ();
150 char *relative_filename (), *absolute_filename (), *absolute_dirname ();
151 long *xmalloc (), *xrealloc ();
153 typedef void Lang_function ();
154 #if FALSE /* many compilers barf on this */
155 Lang_function Asm_labels
;
156 Lang_function default_C_entries
;
157 Lang_function C_entries
;
158 Lang_function Cplusplus_entries
;
159 Lang_function Cstar_entries
;
160 Lang_function Erlang_functions
;
161 Lang_function Fortran_functions
;
162 Lang_function Yacc_entries
;
163 Lang_function Lisp_functions
;
164 Lang_function Pascal_functions
;
165 Lang_function Perl_functions
;
166 Lang_function Prolog_functions
;
167 Lang_function Scheme_functions
;
168 Lang_function TeX_functions
;
169 Lang_function just_read_file
;
170 #else /* so let's write it this way */
173 void default_C_entries ();
174 void plain_C_entries ();
175 void Cplusplus_entries ();
176 void Cstar_entries ();
177 void Erlang_functions ();
178 void Fortran_functions ();
179 void Yacc_entries ();
180 void Lisp_functions ();
181 void Pascal_functions ();
182 void Perl_functions ();
183 void Prolog_functions ();
184 void Scheme_functions ();
185 void TeX_functions ();
186 void just_read_file ();
189 Lang_function
*get_language_from_name ();
190 Lang_function
*get_language_from_interpreter ();
191 Lang_function
*get_language_from_suffix ();
192 int total_size_of_entries ();
194 long readline_internal ();
200 void suggest_asking_for_help ();
201 void fatal (), pfatal ();
202 void find_entries ();
208 void process_file ();
213 char searchar
= '/'; /* use /.../ searches */
215 int lineno
; /* line number of current line */
216 long charno
; /* current character number */
218 long linecharno
; /* charno of start of line; not used by C,
219 but by every other language. */
221 char *curfile
; /* current input file name */
222 char *tagfile
; /* output file */
223 char *progname
; /* name this program was invoked with */
224 char *cwd
; /* current working directory */
225 char *tagfiledir
; /* directory of tagfile */
227 FILE *tagf
; /* ioptr for tags file */
228 NODE
*head
; /* the head of the binary tree of tags */
231 * A `struct linebuffer' is a structure which holds a line of text.
232 * `readline' reads a line from a stream into a linebuffer and works
233 * regardless of the length of the line.
235 #define GROW_LINEBUFFER(buf,toksize) \
236 while (buf.size < toksize) \
237 buf.buffer = (char *) xrealloc (buf.buffer, buf.size *= 2)
244 struct linebuffer lb
; /* the current line */
245 struct linebuffer token_name
; /* used by C_entries as a temporary area */
249 struct linebuffer lb
; /* used by C_entries instead of lb */
252 /* boolean "functions" (see init) */
253 logical _wht
[0177], _etk
[0177], _itk
[0177], _btk
[0177];
256 *white
= " \f\t\n\013",
257 /* token ending chars */
258 *endtk
= " \t\n\013\"'#()[]{}=-+%*/&|^~!<>;,.:?",
259 /* token starting chars */
260 *begtk
= "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~@",
261 /* valid in-token chars */
262 *intk
= "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789";
264 logical append_to_tagfile
; /* -a: append to tags */
265 /* The following three default to TRUE for etags, but to FALSE for ctags. */
266 logical typedefs
; /* -t: create tags for typedefs */
267 logical typedefs_and_cplusplus
; /* -T: create tags for typedefs, level */
268 /* 0 struct/enum/union decls, and C++ */
269 /* member functions. */
270 logical constantypedefs
; /* -d: create tags for C #define and enum */
271 /* constants. Enum consts not implemented. */
272 /* -D: opposite of -d. Default under ctags. */
273 logical update
; /* -u: update tags */
274 logical vgrind_style
; /* -v: create vgrind style index output */
275 logical no_warnings
; /* -w: suppress warnings */
276 logical cxref_style
; /* -x: create cxref style output */
277 logical cplusplus
; /* .[hc] means C++, not C */
278 logical noindentypedefs
; /* -I: ignore indentation in C */
280 struct option longopts
[] =
282 { "append", no_argument
, NULL
, 'a' },
283 { "backward-search", no_argument
, NULL
, 'B' },
284 { "c++", no_argument
, NULL
, 'C' },
285 { "cxref", no_argument
, NULL
, 'x' },
286 { "defines", no_argument
, NULL
, 'd' },
287 { "help", no_argument
, NULL
, 'h' },
288 { "help", no_argument
, NULL
, 'H' },
289 { "ignore-indentation", no_argument
, NULL
, 'I' },
290 { "include", required_argument
, NULL
, 'i' },
291 { "language", required_argument
, NULL
, 'l' },
292 { "no-defines", no_argument
, NULL
, 'D' },
293 { "no-regex", no_argument
, NULL
, 'R' },
294 { "no-warn", no_argument
, NULL
, 'w' },
295 { "output", required_argument
, NULL
, 'o' },
296 { "regex", required_argument
, NULL
, 'r' },
297 { "typedefs", no_argument
, NULL
, 't' },
298 { "typedefs-and-c++", no_argument
, NULL
, 'T' },
299 { "update", no_argument
, NULL
, 'u' },
300 { "version", no_argument
, NULL
, 'V' },
301 { "vgrind", no_argument
, NULL
, 'v' },
306 /* Structure defining a regular expression. Elements are
307 the compiled pattern, and the name string. */
310 struct re_pattern_buffer
*pattern
;
311 struct re_registers regs
;
313 logical error_signaled
;
316 /* Number of regexps found. */
317 int num_patterns
= 0;
319 /* Array of all regexps. */
320 struct pattern
*patterns
= NULL
;
321 #endif /* ETAGS_REGEXPS */
327 /* Non-NULL if language fixed. */
328 Lang_function
*lang_func
= NULL
;
331 char *Asm_suffixes
[] = { "a", /* Unix assembler */
332 "asm", /* Microcontroller assembly */
333 "def", /* BSO/Tasking definition includes */
334 "inc", /* Microcontroller include files */
335 "ins", /* Microcontroller include files */
336 "s", "sa", /* Unix assembler */
337 "src", /* BSO/Tasking C compiler output */
341 /* Note that .c and .h can be considered C++, if the --c++ flag was
342 given. That is why default_C_entries is called here. */
343 char *default_C_suffixes
[] =
346 /* .M is for Objective C++ files. */
347 char *Cplusplus_suffixes
[] =
348 { "C", "H", "c++", "cc", "cpp", "cxx", "h++", "hh", "hpp", "hxx", "M", NULL
};
350 char *Cstar_suffixes
[] =
351 { "cs", "hs", NULL
};
353 char *Erlang_suffixes
[] =
354 { "erl", "hrl", NULL
};
356 char *Fortran_suffixes
[] =
357 { "F", "f", "f90", "for", NULL
};
359 char *Lisp_suffixes
[] =
360 { "cl", "clisp", "el", "l", "lisp", "lsp", "ml", NULL
};
362 char *Pascal_suffixes
[] =
363 { "p", "pas", NULL
};
365 char *Perl_suffixes
[] =
366 { "pl", "pm", NULL
};
367 char *Perl_interpreters
[] =
368 { "perl", "@PERL@", NULL
};
370 char *plain_C_suffixes
[] =
371 { "pc", /* Pro*C file */
372 "m", /* Objective C file */
373 "lm", /* Objective lex file */
376 char *Prolog_suffixes
[] =
379 /* Can't do the `SCM' or `scm' prefix with a version number. */
380 char *Scheme_suffixes
[] =
381 { "SCM", "SM", "oak", "sch", "scheme", "scm", "sm", "t", NULL
};
383 char *TeX_suffixes
[] =
384 { "TeX", "bib", "clo", "cls", "ltx", "sty", "tex", NULL
};
386 char *Yacc_suffixes
[] =
387 { "y", "ym", NULL
}; /* .ym is Objective yacc file */
389 /* Table of language names and corresponding functions, file suffixes
390 and interpreter names.
391 It is ok for a given function to be listed under more than one
392 name. I just didn't. */
396 Lang_function
*function
;
401 struct lang_entry lang_names
[] =
403 { "asm", Asm_labels
, Asm_suffixes
, NULL
},
404 { "c", default_C_entries
, default_C_suffixes
, NULL
},
405 { "c++", Cplusplus_entries
, Cplusplus_suffixes
, NULL
},
406 { "c*", Cstar_entries
, Cstar_suffixes
, NULL
},
407 { "erlang", Erlang_functions
, Erlang_suffixes
, NULL
},
408 { "fortran", Fortran_functions
, Fortran_suffixes
, NULL
},
409 { "lisp", Lisp_functions
, Lisp_suffixes
, NULL
},
410 { "pascal", Pascal_functions
, Pascal_suffixes
, NULL
},
411 { "perl", Perl_functions
, Perl_suffixes
, Perl_interpreters
},
412 { "proc", plain_C_entries
, plain_C_suffixes
, NULL
},
413 { "prolog", Prolog_functions
, Prolog_suffixes
, NULL
},
414 { "scheme", Scheme_functions
, Scheme_suffixes
, NULL
},
415 { "tex", TeX_functions
, TeX_suffixes
, NULL
},
416 { "yacc", Yacc_entries
, Yacc_suffixes
, NULL
},
417 { "auto", NULL
}, /* default guessing scheme */
418 { "none", just_read_file
}, /* regexp matching only */
419 { NULL
, NULL
} /* end of list */
424 print_language_names ()
426 struct lang_entry
*lang
;
429 puts ("\nThese are the currently supported languages, along with the\n\
430 default file name suffixes:");
431 for (lang
= lang_names
; lang
->name
!= NULL
; lang
++)
433 printf ("\t%s\t", lang
->name
);
434 if (lang
->suffixes
!= NULL
)
435 for (ext
= lang
->suffixes
; *ext
!= NULL
; ext
++)
436 printf (" .%s", *ext
);
439 puts ("Where `auto' means use default language for files based on file\n\
440 name suffix, and `none' means only do regexp processing on files.\n\
441 If no language is specified and no matching suffix is found,\n\
442 the first line of the file is read for a sharp-bang (#!) sequence\n\
443 followed by the name of an interpreter. If no such sequence is found,\n\
444 Fortran is tried first; if no tags are found, C is tried next.");
448 # define VERSION "19"
453 printf ("%s for Emacs version %s\n", (CTAGS
) ? "ctags" : "etags", VERSION
);
461 printf ("These are the options accepted by %s. You may use unambiguous\n\
462 abbreviations for the long option names. A - as file name means read\n\
463 names from stdin.", progname
);
465 printf (" Absolute names are stored in the output file as they\n\
466 are. Relative ones are stored relative to the output file's directory.");
469 puts ("-a, --append\n\
470 Append tag entries to existing tags file.");
473 puts ("-B, --backward-search\n\
474 Write the search commands for the tag entries using '?', the\n\
475 backward-search command instead of '/', the forward-search command.");
478 Treat files whose name suffix defaults to C language as C++ files.");
481 puts ("-d, --defines\n\
482 Create tag entries for constant C #defines, too.");
484 puts ("-D, --no-defines\n\
485 Don't create tag entries for constant C #defines. This makes\n\
486 the tags file smaller.");
490 puts ("-i FILE, --include=FILE\n\
491 Include a note in tag file indicating that, when searching for\n\
492 a tag, one should also consult the tags file FILE after\n\
493 checking the current file.");
494 puts ("-l LANG, --language=LANG\n\
495 Force the following files to be considered as written in the\n\
496 named language up to the next --language=LANG option.");
500 puts ("-r /REGEXP/, --regex=/REGEXP/\n\
501 Make a tag for each line matching pattern REGEXP in the\n\
502 following files. REGEXP is anchored (as if preceded by ^).\n\
503 The form /REGEXP/NAME/ creates a named tag. For example Tcl\n\
504 named tags can be created with:\n\
505 --regex=/proc[ \\t]+\\([^ \\t]+\\)/\\1/.");
506 puts ("-R, --no-regex\n\
507 Don't create tags from regexps for the following files.");
508 #endif /* ETAGS_REGEXPS */
509 puts ("-o FILE, --output=FILE\n\
510 Write the tags to FILE.");
511 puts ("-I, --ignore-indentation\n\
512 Don't rely on indentation quite as much as normal. Currently,\n\
513 this means not to assume that a closing brace in the first\n\
514 column is the final brace of a function or structure\n\
515 definition in C and C++.");
519 puts ("-t, --typedefs\n\
520 Generate tag entries for C typedefs.");
521 puts ("-T, --typedefs-and-c++\n\
522 Generate tag entries for C typedefs, C struct/enum/union tags,\n\
523 and C++ member functions.");
524 puts ("-u, --update\n\
525 Update the tag entries for the given files, leaving tag\n\
526 entries for other files in place. Currently, this is\n\
527 implemented by deleting the existing entries for the given\n\
528 files and then rewriting the new entries at the end of the\n\
529 tags file. It is often faster to simply rebuild the entire\n\
530 tag file than to use this.");
531 puts ("-v, --vgrind\n\
532 Generates an index of items intended for human consumption,\n\
533 similar to the output of vgrind. The index is sorted, and\n\
534 gives the page number of each item.");
535 puts ("-w, --no-warn\n\
536 Suppress warning messages about entries defined in multiple\n\
538 puts ("-x, --cxref\n\
539 Like --vgrind, but in the style of cxref, rather than vgrind.\n\
540 The output uses line numbers instead of page numbers, but\n\
541 beyond that the differences are cosmetic; try both to see\n\
545 puts ("-V, --version\n\
546 Print the version of the program.\n\
548 Print this help message.");
550 print_language_names ();
563 /* This structure helps us allow mixing of --lang and filenames. */
566 enum argument_type arg_type
;
568 Lang_function
*function
;
571 #ifdef VMS /* VMS specific functions */
575 /* This is a BUG! ANY arbitrary limit is a BUG!
576 Won't someone please fix this? */
577 #define MAX_FILE_SPEC_LEN 255
580 char body
[MAX_FILE_SPEC_LEN
+ 1];
584 v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names
585 returning in each successive call the next filename matching the input
586 spec. The function expects that each in_spec passed
587 to it will be processed to completion; in particular, up to and
588 including the call following that in which the last matching name
589 is returned, the function ignores the value of in_spec, and will
590 only start processing a new spec with the following call.
591 If an error occurs, on return out_spec contains the value
592 of in_spec when the error occurred.
594 With each successive filename returned in out_spec, the
595 function's return value is one. When there are no more matching
596 names the function returns zero. If on the first call no file
597 matches in_spec, or there is any other error, -1 is returned.
602 #define OUTSIZE MAX_FILE_SPEC_LEN
608 static long context
= 0;
609 static struct dsc$descriptor_s o
;
610 static struct dsc$descriptor_s i
;
611 static logical pass1
= TRUE
;
618 o
.dsc$a_pointer
= (char *) out
;
619 o
.dsc$w_length
= (short)OUTSIZE
;
620 i
.dsc$a_pointer
= in
;
621 i
.dsc$w_length
= (short)strlen(in
);
622 i
.dsc$b_dtype
= DSC$K_DTYPE_T
;
623 i
.dsc$b_class
= DSC$K_CLASS_S
;
624 o
.dsc$b_dtype
= DSC$K_DTYPE_VT
;
625 o
.dsc$b_class
= DSC$K_CLASS_VS
;
627 if ((status
= lib$
find_file(&i
, &o
, &context
, 0, 0)) == RMS$_NORMAL
)
629 out
->body
[out
->curlen
] = EOS
;
632 else if (status
== RMS$_NMF
)
636 strcpy(out
->body
, in
);
639 lib$
find_file_end(&context
);
645 v1.01 nmm 19-Aug-85 gfnames - return in successive calls the
646 name of each file specified by the provided arg expanding wildcards.
649 gfnames (arg
, p_error
)
653 static vspec filename
= {MAX_FILE_SPEC_LEN
, "\0"};
655 switch (fn_exp (&filename
, arg
))
659 return filename
.body
;
665 return filename
.body
;
669 #ifndef OLD /* Newer versions of VMS do provide `system'. */
673 fprintf (stderr
, "system() function not implemented under VMS\n");
677 #define VERSION_DELIM ';'
678 char *massage_name (s
)
684 if (*s
== VERSION_DELIM
)
702 unsigned int nincluded_files
= 0;
703 char **included_files
= xnew (argc
, char *);
706 int current_arg
= 0, file_count
= 0;
707 struct linebuffer filename_lb
;
713 _fmode
= O_BINARY
; /* all of files are treated as binary files */
718 /* Allocate enough no matter what happens. Overkill, but each one
720 argbuffer
= xnew (argc
, argument
);
723 /* Set syntax for regular expression routines. */
724 re_set_syntax (RE_SYNTAX_EMACS
);
725 #endif /* ETAGS_REGEXPS */
728 * If etags, always find typedefs and structure tags. Why not?
729 * Also default is to find macro constants.
732 typedefs
= typedefs_and_cplusplus
= constantypedefs
= TRUE
;
736 int opt
= getopt_long (argc
, argv
,
737 "-aCdDf:Il:o:r:RStTi:BuvxwVhH", longopts
, 0);
745 /* If getopt returns 0, then it has already processed a
746 long-named option. We should do nothing. */
750 /* This means that a filename has been seen. Record it. */
751 argbuffer
[current_arg
].arg_type
= at_filename
;
752 argbuffer
[current_arg
].what
= optarg
;
757 /* Common options. */
759 append_to_tagfile
= TRUE
;
765 constantypedefs
= TRUE
;
768 constantypedefs
= FALSE
;
770 case 'f': /* for compatibility with old makefiles */
774 fprintf (stderr
, "%s: -%c option may only be given once.\n",
776 suggest_asking_for_help ();
781 case 'S': /* for backward compatibility */
782 noindentypedefs
= TRUE
;
785 argbuffer
[current_arg
].function
= get_language_from_name (optarg
);
786 argbuffer
[current_arg
].arg_type
= at_language
;
791 argbuffer
[current_arg
].arg_type
= at_regexp
;
792 argbuffer
[current_arg
].what
= optarg
;
796 argbuffer
[current_arg
].arg_type
= at_regexp
;
797 argbuffer
[current_arg
].what
= NULL
;
800 #endif /* ETAGS_REGEXPS */
812 typedefs
= typedefs_and_cplusplus
= TRUE
;
817 included_files
[nincluded_files
++] = optarg
;
838 suggest_asking_for_help ();
842 for (; optind
< argc
; ++optind
)
844 argbuffer
[current_arg
].arg_type
= at_filename
;
845 argbuffer
[current_arg
].what
= argv
[optind
];
850 if (nincluded_files
== 0 && file_count
== 0)
852 fprintf (stderr
, "%s: No input files specified.\n", progname
);
853 suggest_asking_for_help ();
857 tagfile
= CTAGS
? "tags" : "TAGS";
858 cwd
= etags_getcwd (); /* the current working directory */
859 if (cwd
[strlen(cwd
)-1] != '/')
861 if (streq (tagfile
, "-"))
864 tagfiledir
= absolute_dirname (tagfile
, cwd
);
866 init (); /* set up boolean "functions" */
869 initbuffer (&token_name
);
870 initbuffer (&lbs
[0].lb
);
871 initbuffer (&lbs
[1].lb
);
872 initbuffer (&filename_lb
);
876 if (streq (tagfile
, "-"))
879 tagf
= fopen (tagfile
, append_to_tagfile
? "a" : "w");
885 * Loop through files finding functions.
887 for (i
= 0; i
< current_arg
; ++i
)
889 switch (argbuffer
[i
].arg_type
)
892 lang_func
= argbuffer
[i
].function
;
896 add_regex (argbuffer
[i
].what
);
901 while ((this_file
= gfnames (argbuffer
[i
].what
, &got_err
)) != NULL
)
905 error ("Can't find file %s\n", this_file
);
910 this_file
= massage_name (this_file
);
913 this_file
= argbuffer
[i
].what
;
915 /* Input file named "-" means read file names from stdin
917 if (streq (this_file
, "-"))
918 while (readline_internal (&filename_lb
, stdin
) > 0)
919 process_file (filename_lb
.buffer
);
921 process_file (this_file
);
931 while (nincluded_files
-- > 0)
932 fprintf (tagf
, "\f\n%s,include\n", *included_files
++);
938 /* If CTAGS, we are here. process_file did not write the tags yet,
939 because we want them ordered. Let's do it now. */
942 tagf
= fopen (tagfile
, append_to_tagfile
? "a" : "w");
952 for (i
= 0; i
< current_arg
; ++i
)
954 if (argbuffer
[i
].arg_type
!= at_filename
)
957 "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS",
958 tagfile
, argbuffer
[i
].what
, tagfile
);
959 if (system (cmd
) != GOOD
)
960 fatal ("failed to execute shell command");
962 append_to_tagfile
= TRUE
;
965 tagf
= fopen (tagfile
, append_to_tagfile
? "a" : "w");
974 sprintf (cmd
, "sort %s -o %s", tagfile
, tagfile
);
982 * Return a Lang_function given the name.
985 get_language_from_name (name
)
988 struct lang_entry
*lang
;
991 for (lang
= lang_names
; lang
->name
!= NULL
; lang
++)
993 if (streq (name
, lang
->name
))
994 return lang
->function
;
997 fprintf (stderr
, "%s: language \"%s\" not recognized.\n",
999 suggest_asking_for_help ();
1001 /* This point should never be reached. The function should either
1002 return a function pointer or never return. Note that a NULL
1003 pointer cannot be considered as an error, as it means that the
1004 language has not been explicitely imposed by the user ("auto"). */
1005 return NULL
; /* avoid warnings from compiler */
1010 * Return a Lang_function given the interpreter name.
1013 get_language_from_interpreter (interpreter
)
1016 struct lang_entry
*lang
;
1019 if (interpreter
== NULL
)
1021 for (lang
= lang_names
; lang
->name
!= NULL
; lang
++)
1022 if (lang
->interpreters
!= NULL
)
1023 for (iname
= lang
->interpreters
; *iname
!= NULL
; iname
++)
1024 if (streq (*iname
, interpreter
))
1025 return lang
->function
;
1033 * Return a Lang_function given the file suffix.
1036 get_language_from_suffix (suffix
)
1039 struct lang_entry
*lang
;
1044 for (lang
= lang_names
; lang
->name
!= NULL
; lang
++)
1045 if (lang
->suffixes
!= NULL
)
1046 for (ext
= lang
->suffixes
; *ext
!= NULL
; ext
++)
1047 if (streq (*ext
, suffix
))
1048 return lang
->function
;
1055 * This routine is called on each file argument.
1061 struct stat stat_buf
;
1064 if (stat (file
, &stat_buf
) == 0 && !S_ISREG (stat_buf
.st_mode
))
1066 fprintf (stderr
, "Skipping %s: it is not a regular file.\n", file
);
1069 if (streq (file
, tagfile
) && !streq (tagfile
, "-"))
1071 fprintf (stderr
, "Skipping inclusion of %s in self.\n", file
);
1074 inf
= fopen (file
, "r");
1081 find_entries (file
, inf
);
1087 if (absolutefn (file
))
1089 /* file is an absolute filename. Canonicalise it. */
1090 filename
= absolute_filename (file
, cwd
);
1094 /* file is a filename relative to cwd. Make it relative
1095 to the directory of the tags file. */
1096 filename
= relative_filename (file
, tagfiledir
);
1098 fprintf (tagf
, "\f\n%s,%d\n", filename
, total_size_of_entries (head
));
1107 * This routine sets up the boolean pseudo-functions which work
1108 * by setting boolean flags dependent upon the corresponding character
1109 * Every char which is NOT in that string is not a white char. Therefore,
1110 * all of the array "_wht" is set to FALSE, and then the elements
1111 * subscripted by the chars in "white" are set to TRUE. Thus "_wht"
1112 * of a char is TRUE if it is the string "white", else FALSE.
1120 for (i
= 0; i
< 0177; i
++)
1121 _wht
[i
] = _etk
[i
] = _itk
[i
] = _btk
[i
] = FALSE
;
1122 for (sp
= white
; *sp
; sp
++)
1124 for (sp
= endtk
; *sp
; sp
++)
1126 for (sp
= intk
; *sp
; sp
++)
1128 for (sp
= begtk
; *sp
; sp
++)
1130 _wht
[0] = _wht
['\n'];
1131 _etk
[0] = _etk
['\n'];
1132 _btk
[0] = _btk
['\n'];
1133 _itk
[0] = _itk
['\n'];
1137 * This routine opens the specified file and calls the function
1138 * which finds the function and type definitions.
1141 find_entries (file
, inf
)
1146 Lang_function
*function
;
1147 NODE
*old_last_node
;
1148 extern NODE
*last_node
;
1151 /* Memory leakage here: the memory block pointed by curfile is never
1152 released. The amount of memory leaked here is the sum of the
1153 lengths of the input file names. */
1154 curfile
= savestr (file
);
1156 /* If user specified a language, use it. */
1157 function
= lang_func
;
1158 if (function
!= NULL
)
1165 cp
= etags_strrchr (file
, '.');
1169 function
= get_language_from_suffix (cp
);
1170 if (function
!= NULL
)
1178 /* Look for sharp-bang as the first two characters. */
1179 if (readline_internal (&lb
, inf
) > 2
1180 && lb
.buffer
[0] == '#'
1181 && lb
.buffer
[1] == '!')
1185 /* Set lp to point at the first char after the last slash in the
1186 line or, if no slashes, at the first nonblank. Then set cp to
1187 the first successive blank and terminate the string. */
1188 lp
= etags_strrchr (lb
.buffer
+2, '/');
1192 for (lp
= lb
.buffer
+2; *lp
!= '\0' && isspace (*lp
); lp
++)
1194 for (cp
= lp
; *cp
!= '\0' && !isspace (*cp
); cp
++)
1198 if (strlen (lp
) > 0)
1200 function
= get_language_from_interpreter (lp
);
1201 if (function
!= NULL
)
1212 old_last_node
= last_node
;
1213 Fortran_functions (inf
);
1215 /* No Fortran entries found. Try C. */
1216 if (old_last_node
== last_node
)
1219 default_C_entries (inf
);
1227 pfnote (name
, is_func
, linestart
, linelen
, lno
, cno
)
1228 char *name
; /* tag name, or NULL if unnamed */
1229 logical is_func
; /* tag is a function */
1230 char *linestart
; /* start of the line where tag is */
1231 int linelen
; /* length of the line where tag is */
1232 int lno
; /* line number */
1233 long cno
; /* character number */
1237 if (CTAGS
&& name
== NULL
)
1240 np
= xnew (1, NODE
);
1242 /* If ctags mode, change name "main" to M<thisfilename>. */
1243 if (CTAGS
&& !cxref_style
&& streq (name
, "main"))
1245 register char *fp
= etags_strrchr (curfile
, '/');
1246 np
->name
= concat ("M", fp
== 0 ? curfile
: fp
+ 1, "");
1247 fp
= etags_strrchr (np
->name
, '.');
1248 if (fp
&& fp
[1] != '\0' && fp
[2] == '\0')
1253 np
->been_warned
= FALSE
;
1255 np
->is_func
= is_func
;
1257 /* Our char numbers are 0-base, because of C language tradition?
1258 ctags compatibility? old versions compatibility? I don't know.
1259 Anyway, since emacs's are 1-base we expect etags.el to take care
1260 of the difference. If we wanted to have 1-based numbers, we would
1261 uncomment the +1 below. */
1262 np
->cno
= cno
/* + 1 */ ;
1263 np
->left
= np
->right
= NULL
;
1264 if (CTAGS
&& !cxref_style
)
1266 if (strlen (linestart
) < 50)
1267 np
->pat
= concat (linestart
, "$", "");
1269 np
->pat
= savenstr (linestart
, 50);
1272 np
->pat
= savenstr (linestart
, linelen
);
1274 add_node (np
, &head
);
1279 * recurse on left children, iterate on right children.
1283 register NODE
*node
;
1287 register NODE
*node_right
= node
->right
;
1288 free_tree (node
->left
);
1289 if (node
->name
!= NULL
)
1292 free ((char *) node
);
1299 * Adds a node to the tree of nodes. In etags mode, we don't keep
1300 * it sorted; we just keep a linear list. In ctags mode, maintain
1301 * an ordered tree, with no attempt at balancing.
1303 * add_node is the only function allowed to add nodes, so it can
1306 NODE
*last_node
= NULL
;
1308 add_node (node
, cur_node_p
)
1309 NODE
*node
, **cur_node_p
;
1312 register NODE
*cur_node
= *cur_node_p
;
1314 if (cur_node
== NULL
)
1324 if (last_node
== NULL
)
1325 fatal ("internal error in add_node", 0);
1326 last_node
->right
= node
;
1332 dif
= strcmp (node
->name
, cur_node
->name
);
1335 * If this tag name matches an existing one, then
1336 * do not add the node, but maybe print a warning.
1340 if (streq (node
->file
, cur_node
->file
))
1344 fprintf (stderr
, "Duplicate entry in file %s, line %d: %s\n",
1345 node
->file
, lineno
, node
->name
);
1346 fprintf (stderr
, "Second entry ignored\n");
1349 else if (!cur_node
->been_warned
&& !no_warnings
)
1353 "Duplicate entry in files %s and %s: %s (Warning only)\n",
1354 node
->file
, cur_node
->file
, node
->name
);
1355 cur_node
->been_warned
= TRUE
;
1360 /* Actually add the node */
1361 add_node (node
, dif
< 0 ? &cur_node
->left
: &cur_node
->right
);
1367 register NODE
*node
;
1374 /* Output subentries that precede this one */
1375 put_entries (node
->left
);
1377 /* Output this entry */
1381 if (node
->name
!= NULL
)
1382 fprintf (tagf
, "%s\177%s\001%d,%d\n",
1383 node
->pat
, node
->name
, node
->lno
, node
->cno
);
1385 fprintf (tagf
, "%s\177%d,%d\n",
1386 node
->pat
, node
->lno
, node
->cno
);
1390 if (node
->name
== NULL
)
1391 error ("internal error: NULL name in ctags mode.", 0);
1396 fprintf (stdout
, "%s %s %d\n",
1397 node
->name
, node
->file
, (node
->lno
+ 63) / 64);
1399 fprintf (stdout
, "%-16s %3d %-16s %s\n",
1400 node
->name
, node
->lno
, node
->file
, node
->pat
);
1404 fprintf (tagf
, "%s\t%s\t", node
->name
, node
->file
);
1408 putc (searchar
, tagf
);
1411 for (sp
= node
->pat
; *sp
; sp
++)
1413 if (*sp
== '\\' || *sp
== searchar
)
1417 putc (searchar
, tagf
);
1420 { /* a typedef; text pattern inadequate */
1421 fprintf (tagf
, "%d", node
->lno
);
1427 /* Output subentries that follow this one */
1428 put_entries (node
->right
);
1431 /* Length of a number's decimal representation. */
1439 for (; num
; num
/= 10)
1445 * Return total number of characters that put_entries will output for
1446 * the nodes in the subtree of the specified node. Works only if
1447 * we are not ctags, but called only in that case. This count
1448 * is irrelevant with the new tags.el, but is still supplied for
1449 * backward compatibility.
1452 total_size_of_entries (node
)
1453 register NODE
*node
;
1461 for (; node
; node
= node
->right
)
1463 /* Count left subentries. */
1464 total
+= total_size_of_entries (node
->left
);
1466 /* Count this entry */
1467 total
+= strlen (node
->pat
) + 1;
1468 total
+= number_len ((long) node
->lno
) + 1 + number_len (node
->cno
) + 1;
1469 if (node
->name
!= NULL
)
1470 total
+= 1 + strlen (node
->name
); /* \001name */
1477 * The C symbol tables.
1481 st_none
, st_C_objprot
, st_C_objimpl
, st_C_objend
, st_C_gnumacro
,
1482 st_C_struct
, st_C_enum
, st_C_define
, st_C_typedef
, st_C_typespec
1485 /* Feed stuff between (but not including) %[ and %] lines to:
1486 gperf -c -k1,3 -o -p -r -t
1488 struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
1490 @interface, 0, st_C_objprot
1491 @protocol, 0, st_C_objprot
1492 @implementation,0, st_C_objimpl
1493 @end, 0, st_C_objend
1494 class, C_PLPL, st_C_struct
1495 domain, C_STAR, st_C_struct
1496 union, 0, st_C_struct
1497 struct, 0, st_C_struct
1499 typedef, 0, st_C_typedef
1500 define, 0, st_C_define
1501 long, 0, st_C_typespec
1502 short, 0, st_C_typespec
1503 int, 0, st_C_typespec
1504 char, 0, st_C_typespec
1505 float, 0, st_C_typespec
1506 double, 0, st_C_typespec
1507 signed, 0, st_C_typespec
1508 unsigned, 0, st_C_typespec
1509 auto, 0, st_C_typespec
1510 void, 0, st_C_typespec
1511 extern, 0, st_C_typespec
1512 static, 0, st_C_typespec
1513 const, 0, st_C_typespec
1514 volatile, 0, st_C_typespec
1515 # DEFUN used in emacs, the next three used in glibc (SYSCALL only for mach).
1516 DEFUN, 0, st_C_gnumacro
1517 SYSCALL, 0, st_C_gnumacro
1518 ENTRY, 0, st_C_gnumacro
1519 PSEUDO, 0, st_C_gnumacro
1520 # These are defined inside C functions, so currently they are not met.
1521 # EXFUN used in glibc, DEFVAR_* in emacs.
1522 #EXFUN, 0, st_C_gnumacro
1523 #DEFVAR_, 0, st_C_gnumacro
1525 and replace lines between %< and %> with its output. */
1527 /* C code produced by gperf version 1.8.1 (K&R C version) */
1528 /* Command-line: gperf -c -k1,3 -o -p -r -t */
1531 struct C_stab_entry
{ char *name
; int c_ext
; enum sym_type type
; };
1533 #define MIN_WORD_LENGTH 3
1534 #define MAX_WORD_LENGTH 15
1535 #define MIN_HASH_VALUE 7
1536 #define MAX_HASH_VALUE 63
1539 57 is the maximum key range
1547 static unsigned char hash_table
[] =
1549 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1550 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1551 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1552 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1553 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1554 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1555 63, 63, 63, 63, 17, 63, 63, 63, 4, 14,
1556 4, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1557 8, 63, 63, 0, 23, 63, 63, 63, 63, 63,
1558 63, 63, 63, 63, 63, 63, 63, 28, 63, 28,
1559 10, 31, 27, 18, 63, 6, 63, 63, 26, 1,
1560 11, 2, 29, 63, 29, 16, 26, 13, 15, 63,
1561 63, 63, 63, 63, 63, 63, 63, 63,
1563 return len
+ hash_table
[str
[2]] + hash_table
[str
[0]];
1566 struct C_stab_entry
*
1567 in_word_set (str
, len
)
1572 static struct C_stab_entry wordlist
[] =
1574 {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
1575 {"SYSCALL", 0, st_C_gnumacro
},
1576 {"",}, {"",}, {"",}, {"",}, {"",},
1577 {"DEFUN", 0, st_C_gnumacro
},
1578 {"",}, {"",}, {"",},
1579 {"domain", C_STAR
, st_C_struct
},
1580 {"",}, {"",}, {"",}, {"",}, {"",},
1581 {"short", 0, st_C_typespec
},
1582 {"union", 0, st_C_struct
},
1583 {"void", 0, st_C_typespec
},
1585 {"PSEUDO", 0, st_C_gnumacro
},
1586 {"double", 0, st_C_typespec
},
1588 {"@end", 0, st_C_objend
},
1589 {"@implementation", 0, st_C_objimpl
},
1590 {"float", 0, st_C_typespec
},
1591 {"int", 0, st_C_typespec
},
1593 {"unsigned", 0, st_C_typespec
},
1594 {"@interface", 0, st_C_objprot
},
1596 {"signed", 0, st_C_typespec
},
1597 {"long", 0, st_C_typespec
},
1598 {"ENTRY", 0, st_C_gnumacro
},
1599 {"define", 0, st_C_define
},
1600 {"const", 0, st_C_typespec
},
1601 {"",}, {"",}, {"",},
1602 {"enum", 0, st_C_enum
},
1603 {"volatile", 0, st_C_typespec
},
1604 {"static", 0, st_C_typespec
},
1605 {"struct", 0, st_C_struct
},
1606 {"",}, {"",}, {"",},
1607 {"@protocol", 0, st_C_objprot
},
1609 {"auto", 0, st_C_typespec
},
1611 {"char", 0, st_C_typespec
},
1612 {"class", C_PLPL
, st_C_struct
},
1613 {"typedef", 0, st_C_typedef
},
1614 {"extern", 0, st_C_typespec
},
1617 if (len
<= MAX_WORD_LENGTH
&& len
>= MIN_WORD_LENGTH
)
1619 register int key
= hash (str
, len
);
1621 if (key
<= MAX_HASH_VALUE
&& key
>= MIN_HASH_VALUE
)
1623 register char *s
= wordlist
[key
].name
;
1625 if (*s
== *str
&& !strncmp (str
+ 1, s
+ 1, len
- 1))
1626 return &wordlist
[key
];
1634 C_symtype(str
, len
, c_ext
)
1639 register struct C_stab_entry
*se
= in_word_set(str
, len
);
1641 if (se
== NULL
|| (se
->c_ext
&& !(c_ext
& se
->c_ext
)))
1647 * C functions are recognized using a simple finite automaton.
1648 * funcdef is its state variable.
1652 fnone
, /* nothing seen */
1653 ftagseen
, /* function-like tag seen */
1654 fstartlist
, /* just after open parenthesis */
1655 finlist
, /* in parameter list */
1656 flistseen
, /* after parameter list */
1657 fignore
/* before open brace */
1662 * typedefs are recognized using a simple finite automaton.
1663 * typdef is its state variable.
1667 tnone
, /* nothing seen */
1668 ttypedseen
, /* typedef keyword seen */
1669 tinbody
, /* inside typedef body */
1670 tend
, /* just before typedef tag */
1671 tignore
/* junk after typedef tag */
1676 * struct-like structures (enum, struct and union) are recognized
1677 * using another simple finite automaton. `structdef' is its state
1682 snone
, /* nothing seen yet */
1683 skeyseen
, /* struct-like keyword seen */
1684 stagseen
, /* struct-like tag seen */
1685 scolonseen
, /* colon seen after struct-like tag */
1686 sinbody
/* in struct body: recognize member func defs*/
1690 * When structdef is stagseen, scolonseen, or sinbody, structtag is the
1691 * struct tag, and structtype is the type of the preceding struct-like
1694 char *structtag
= "<uninited>";
1695 enum sym_type structtype
;
1698 * When objdef is different from onone, objtag is the name of the class.
1700 char *objtag
= "<uninited>";
1703 * Yet another little state machine to deal with preprocessor lines.
1707 dnone
, /* nothing seen */
1708 dsharpseen
, /* '#' seen as first char on line */
1709 ddefineseen
, /* '#' and 'define' seen */
1710 dignorerest
/* ignore rest of line */
1714 * State machine for Objective C protocols and implementations.
1718 onone
, /* nothing seen */
1719 oprotocol
, /* @interface or @protocol seen */
1720 oimplementation
, /* @implementations seen */
1721 otagseen
, /* class name seen */
1722 oparenseen
, /* parenthesis before category seen */
1723 ocatseen
, /* category name seen */
1724 oinbody
, /* in @implementation body */
1725 omethodsign
, /* in @implementation body, after +/- */
1726 omethodtag
, /* after method name */
1727 omethodcolon
, /* after method colon */
1728 omethodparm
, /* after method parameter */
1729 oignore
/* wait for @end */
1733 * Set this to TRUE, and the next token considered is called a function.
1734 * Used only for GNU emacs's function-defining macros.
1736 logical next_token_is_func
;
1739 * TRUE in the rules part of a yacc file, FALSE outside (parse as C).
1744 * methodlen is the length of the method name stored in token_name.
1750 * checks to see if the current token is at the start of a
1751 * function, or corresponds to a typedef, or is a struct/union/enum
1754 * *IS_FUNC gets TRUE iff the token is a function or macro with args.
1755 * C_EXT is which language we are looking at.
1757 * In the future we will need some way to adjust where the end of
1758 * the token is; for instance, implementing the C++ keyword
1759 * `operator' properly will adjust the end of the token to be after
1760 * whatever follows `operator'.
1768 * next_token_is_func IN OUT
1772 consider_token (str
, len
, c
, c_ext
, cblev
, parlev
, is_func
)
1773 register char *str
; /* IN: token pointer */
1774 register int len
; /* IN: token length */
1775 register char c
; /* IN: first char after the token */
1776 int c_ext
; /* IN: C extensions mask */
1777 int cblev
; /* IN: curly brace level */
1778 int parlev
; /* IN: parenthesis level */
1779 logical
*is_func
; /* OUT: function found */
1781 enum sym_type toktype
= C_symtype (str
, len
, c_ext
);
1784 * Advance the definedef state machine.
1789 /* We're not on a preprocessor line. */
1792 if (toktype
== st_C_define
)
1794 definedef
= ddefineseen
;
1798 definedef
= dignorerest
;
1803 * Make a tag for any macro, unless it is a constant
1804 * and constantypedefs is FALSE.
1806 definedef
= dignorerest
;
1807 *is_func
= (c
== '(');
1808 if (!*is_func
&& !constantypedefs
)
1815 error ("internal error: definedef value.", 0);
1824 if (toktype
== st_C_typedef
)
1827 typdef
= ttypedseen
;
1843 /* Do not return here, so the structdef stuff has a chance. */
1857 * This structdef business is currently only invoked when cblev==0.
1858 * It should be recursively invoked whatever the curly brace level,
1859 * and a stack of states kept, to allow for definitions of structs
1862 * This structdef business is NOT invoked when we are ctags and the
1863 * file is plain C. This is because a struct tag may have the same
1864 * name as another tag, and this loses with ctags.
1866 * This if statement deals with the typdef state machine as
1867 * follows: if typdef==ttypedseen and token is struct/union/class/enum,
1868 * return FALSE. All the other code here is for the structdef
1875 if (typdef
== ttypedseen
1876 || (typedefs_and_cplusplus
&& cblev
== 0 && structdef
== snone
))
1878 structdef
= skeyseen
;
1879 structtype
= toktype
;
1883 if (structdef
== skeyseen
)
1885 /* Save the tag for struct/union/class, for functions that may be
1887 if (structtype
== st_C_struct
)
1888 structtag
= savenstr (str
, len
);
1890 structtag
= "<enum>";
1891 structdef
= stagseen
;
1895 /* Avoid entering funcdef stuff if typdef is going on. */
1896 if (typdef
!= tnone
)
1902 /* Detect GNU macros. */
1903 if (definedef
== dnone
&& toktype
== st_C_gnumacro
)
1905 next_token_is_func
= TRUE
;
1908 if (next_token_is_func
)
1910 next_token_is_func
= FALSE
;
1917 * Detecting Objective C constructs.
1928 objdef
= oimplementation
;
1932 case oimplementation
:
1933 /* Save the class tag for functions that may be defined inside. */
1934 objtag
= savenstr (str
, len
);
1938 /* Save the class tag for categories. */
1939 objtag
= savenstr (str
, len
);
1952 objdef
= omethodtag
;
1954 GROW_LINEBUFFER (token_name
, methodlen
+1);
1955 strncpy (token_name
.buffer
, str
, len
);
1956 token_name
.buffer
[methodlen
] = '\0';
1962 objdef
= omethodparm
;
1967 objdef
= omethodtag
;
1969 GROW_LINEBUFFER (token_name
, methodlen
+1);
1970 strncat (token_name
.buffer
, str
, len
);
1975 if (toktype
== st_C_objend
)
1977 /* Memory leakage here: the string pointed by objtag is
1978 never released, because many tests would be needed to
1979 avoid breaking on incorrect input code. The amount of
1980 memory leaked here is the sum of the lengths of the
1992 if (funcdef
!= finlist
&& funcdef
!= fignore
)
1993 funcdef
= fnone
; /* should be useless */
1996 if (funcdef
== fnone
)
2009 * This routine finds functions, typedefs, #define's and
2010 * struct/union/enum definitions in C syntax and adds them
2024 #define current_lb_is_new (newndx == curndx)
2025 #define switch_line_buffers() (curndx = 1 - curndx)
2027 #define curlb (lbs[curndx].lb)
2028 #define othlb (lbs[1-curndx].lb)
2029 #define newlb (lbs[newndx].lb)
2030 #define curlinepos (lbs[curndx].linepos)
2031 #define othlinepos (lbs[1-curndx].linepos)
2032 #define newlinepos (lbs[newndx].linepos)
2034 #define CNL_SAVE_DEFINEDEF \
2036 curlinepos = charno; \
2038 charno += readline (&curlb, inf); \
2039 lp = curlb.buffer; \
2046 CNL_SAVE_DEFINEDEF; \
2047 if (savetok.valid) \
2050 savetok.valid = FALSE; \
2052 definedef = dnone; \
2055 /* Ideally this macro should never be called wihen tok.valid is FALSE,
2056 but this would mean that the state machines always guess right. */
2057 #define make_tag(isfun) do \
2059 char *name = NULL; \
2060 if (CTAGS || tok.named) \
2061 name = savestr (token_name.buffer); \
2062 pfnote (name, isfun, tok.buffer, tok.linelen, tok.lineno, tok.linepos); \
2063 tok.valid = FALSE; \
2067 C_entries (c_ext
, inf
)
2068 int c_ext
; /* extension of C */
2069 FILE *inf
; /* input file */
2071 register char c
; /* latest char read; '\0' for end of line */
2072 register char *lp
; /* pointer one beyond the character `c' */
2073 int curndx
, newndx
; /* indices for current and new lb */
2074 TOKEN tok
; /* latest token read */
2075 register int tokoff
; /* offset in line of start of current token */
2076 register int toklen
; /* length of current token */
2077 int cblev
; /* current curly brace level */
2078 int parlev
; /* current parenthesis level */
2079 logical incomm
, inquote
, inchar
, quotednl
, midtoken
;
2081 TOKEN savetok
; /* token saved during preprocessor handling */
2084 curndx
= newndx
= 0;
2090 funcdef
= fnone
; typdef
= tnone
; structdef
= snone
;
2091 definedef
= dnone
; objdef
= onone
;
2092 next_token_is_func
= yacc_rules
= FALSE
;
2093 midtoken
= inquote
= inchar
= incomm
= quotednl
= FALSE
;
2094 tok
.valid
= savetok
.valid
= FALSE
;
2097 cplpl
= c_ext
& C_PLPL
;
2104 /* If we're at the end of the line, the next character is a
2105 '\0'; don't skip it, because it's the thing that tells us
2106 to read the next line. */
2127 /* Newlines inside comments do not end macro definitions in
2142 /* Newlines inside strings do not end macro definitions
2143 in traditional cpp, even though compilers don't
2144 usually accept them. */
2155 /* Hmmm, something went wrong. */
2169 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2174 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2184 else if (/* cplpl && */ *lp
== '/')
2192 if ((c_ext
& YACC
) && *lp
== '%')
2194 /* entering or exiting rules section in yacc file */
2196 definedef
= dnone
; funcdef
= fnone
;
2197 typdef
= tnone
; structdef
= snone
;
2198 next_token_is_func
= FALSE
;
2199 midtoken
= inquote
= inchar
= incomm
= quotednl
= FALSE
;
2201 yacc_rules
= !yacc_rules
;
2207 if (definedef
== dnone
)
2210 logical cpptoken
= TRUE
;
2212 /* Look back on this line. If all blanks, or nonblanks
2213 followed by an end of comment, this is a preprocessor
2215 for (cp
= newlb
.buffer
; cp
< lp
-1; cp
++)
2218 if (*cp
== '*' && *(cp
+1) == '/')
2227 definedef
= dsharpseen
;
2228 } /* if (definedef == dnone) */
2234 /* Consider token only if some complicated conditions are satisfied. */
2235 if ((definedef
!= dnone
2236 || (cblev
== 0 && structdef
!= scolonseen
)
2237 || (cblev
== 1 && cplpl
&& structdef
== sinbody
))
2238 && typdef
!= tignore
2239 && definedef
!= dignorerest
2240 && funcdef
!= finlist
)
2246 if (c
== ':' && cplpl
&& *lp
== ':' && begtoken(*(lp
+ 1)))
2249 * This handles :: in the middle, but not at the
2250 * beginning of an identifier.
2257 logical is_func
= FALSE
;
2260 || consider_token (newlb
.buffer
+ tokoff
, toklen
, c
,
2261 c_ext
, cblev
, parlev
, &is_func
))
2263 if (structdef
== sinbody
2264 && definedef
== dnone
2266 /* function defined in C++ class body */
2268 GROW_LINEBUFFER (token_name
,
2269 strlen(structtag
)+2+toklen
+1);
2270 strcpy (token_name
.buffer
, structtag
);
2271 strcat (token_name
.buffer
, "::");
2272 strncat (token_name
.buffer
,
2273 newlb
.buffer
+tokoff
, toklen
);
2276 else if (objdef
== ocatseen
)
2277 /* Objective C category */
2279 GROW_LINEBUFFER (token_name
,
2280 strlen(objtag
)+2+toklen
+1);
2281 strcpy (token_name
.buffer
, objtag
);
2282 strcat (token_name
.buffer
, "(");
2283 strncat (token_name
.buffer
,
2284 newlb
.buffer
+tokoff
, toklen
);
2285 strcat (token_name
.buffer
, ")");
2288 else if (objdef
== omethodtag
2289 || objdef
== omethodparm
)
2290 /* Objective C method */
2296 GROW_LINEBUFFER (token_name
, toklen
+1);
2297 strncpy (token_name
.buffer
,
2298 newlb
.buffer
+tokoff
, toklen
);
2299 token_name
.buffer
[toklen
] = '\0';
2300 if (structdef
== stagseen
2303 && definedef
== dignorerest
)) /* macro */
2308 tok
.lineno
= lineno
;
2309 tok
.linelen
= tokoff
+ toklen
+ 1;
2310 tok
.buffer
= newlb
.buffer
;
2311 tok
.linepos
= newlinepos
;
2314 if (definedef
== dnone
2315 && (funcdef
== ftagseen
2316 || structdef
== stagseen
2318 || objdef
!= onone
))
2320 if (current_lb_is_new
)
2321 switch_line_buffers ();
2328 } /* if (endtoken (c)) */
2329 else if (intoken (c
))
2334 } /* if (midtoken) */
2335 else if (begtoken (c
))
2353 if (structdef
== stagseen
)
2359 if (!yacc_rules
|| lp
== newlb
.buffer
+ 1)
2361 tokoff
= lp
- 1 - newlb
.buffer
;
2366 } /* if (begtoken) */
2367 } /* if must look at token */
2370 /* Detect end of line, colon, comma, semicolon and various braces
2371 after having handled a token.*/
2375 if (definedef
!= dnone
)
2385 objdef
= omethodcolon
;
2387 GROW_LINEBUFFER (token_name
, methodlen
+1);
2388 strcat (token_name
.buffer
, ":");
2391 if (structdef
== stagseen
)
2392 structdef
= scolonseen
;
2409 if (definedef
!= dnone
)
2420 if (funcdef
!= fignore
)
2423 /* The following instruction invalidates the token.
2424 Probably the token should be invalidated in all
2425 other cases where some state machine is reset. */
2428 if (structdef
== stagseen
)
2432 if (definedef
!= dnone
)
2442 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2444 if (structdef
== stagseen
)
2448 if (definedef
!= dnone
)
2450 if (cblev
== 0 && typdef
== tend
)
2456 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2458 if (structdef
== stagseen
)
2462 if (definedef
!= dnone
)
2464 if (objdef
== otagseen
&& parlev
== 0)
2465 objdef
= oparenseen
;
2473 /* Make sure that the next char is not a '*'.
2474 This handles constructs like:
2475 typedef void OperatorFun (int fun); */
2482 } /* switch (typdef) */
2485 funcdef
= fstartlist
;
2494 if (definedef
!= dnone
)
2496 if (objdef
== ocatseen
&& parlev
== 1)
2507 funcdef
= flistseen
;
2510 if (cblev
== 0 && typdef
== tend
)
2516 else if (parlev
< 0) /* can happen due to ill-conceived #if's. */
2520 if (definedef
!= dnone
)
2522 if (typdef
== ttypedseen
)
2526 case skeyseen
: /* unnamed struct */
2527 structtag
= "_anonymous_";
2528 structdef
= sinbody
;
2531 case scolonseen
: /* named struct */
2532 structdef
= sinbody
;
2557 /* Neutralize `extern "C" {' grot and look inside structs. */
2558 if (cblev
== 0 && structdef
== snone
&& typdef
== tnone
)
2565 if (definedef
!= dnone
)
2567 if (funcdef
== fstartlist
)
2568 funcdef
= fnone
; /* avoid tagging `foo' in `foo (*bar()) ()' */
2571 if (definedef
!= dnone
)
2573 if (!noindentypedefs
&& lp
== newlb
.buffer
+ 1)
2575 cblev
= 0; /* reset curly brace level if first column */
2576 parlev
= 0; /* also reset paren level, just in case... */
2582 if (typdef
== tinbody
)
2584 /* Memory leakage here: the string pointed by structtag is
2585 never released, because I fear to miss something and
2586 break things while freeing the area. The amount of
2587 memory leaked here is the sum of the lengths of the
2589 if (structdef == sinbody)
2590 free (structtag); */
2593 structtag
= "<error>";
2598 if (objdef
== oinbody
&& cblev
== 0)
2600 objdef
= omethodsign
;
2604 case '=': case '#': case '~': case '&': case '%': case '/':
2605 case '|': case '^': case '!': case '<': case '>': case '.': case '?':
2606 if (definedef
!= dnone
)
2608 /* These surely cannot follow a function tag. */
2609 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2613 if (objdef
== otagseen
)
2618 /* If a macro spans multiple lines don't reset its state. */
2626 } /* while not eof */
2630 * Process either a C++ file or a C file depending on the setting
2634 default_C_entries (inf
)
2637 C_entries (cplusplus
? C_PLPL
: 0, inf
);
2640 /* Always do plain ANSI C. */
2642 plain_C_entries (inf
)
2648 /* Always do C++. */
2650 Cplusplus_entries (inf
)
2653 C_entries (C_PLPL
, inf
);
2661 C_entries (C_STAR
, inf
);
2664 /* Always do Yacc. */
2669 C_entries (YACC
, inf
);
2672 /* Fortran parsing */
2680 register int len
= 0;
2682 while (*cp
&& lowcase(*cp
) == lowcase(dbp
[len
]))
2684 if (*cp
== '\0' && !intoken(dbp
[len
]))
2695 while (isspace (*dbp
))
2700 while (isspace (*dbp
))
2702 if (strneq (dbp
, "(*)", 3))
2707 if (!isdigit (*dbp
))
2709 --dbp
; /* force failure */
2714 while (isdigit (*dbp
));
2723 while (isspace (*dbp
))
2728 linecharno
= charno
;
2729 charno
+= readline (&lb
, inf
);
2734 while (isspace (*dbp
))
2743 && (isalpha (*cp
) || isdigit (*cp
) || (*cp
== '_') || (*cp
== '$')));
2746 pfnote ((CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
, TRUE
,
2747 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
2751 Fortran_functions (inf
)
2760 linecharno
= charno
;
2761 charno
+= readline (&lb
, inf
);
2764 dbp
++; /* Ratfor escape to fortran */
2765 while (isspace (*dbp
))
2769 switch (lowcase (*dbp
))
2772 if (tail ("integer"))
2780 if (tail ("logical"))
2784 if (tail ("complex") || tail ("character"))
2788 if (tail ("double"))
2790 while (isspace (*dbp
))
2794 if (tail ("precision"))
2800 while (isspace (*dbp
))
2804 switch (lowcase (*dbp
))
2807 if (tail ("function"))
2811 if (tail ("subroutine"))
2819 if (tail ("program"))
2824 if (tail ("procedure"))
2832 * Bob Weiner, Motorola Inc., 4/3/94
2833 * Unix and microcontroller assembly tag handling
2834 * look for '^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]'
2848 linecharno
= charno
;
2849 charno
+= readline (&lb
, inf
);
2852 /* If first char is alphabetic or one of [_.$], test for colon
2853 following identifier. */
2854 if (isalpha (*cp
) || *cp
== '_' || *cp
== '.' || *cp
== '$')
2856 /* Read past label. */
2858 while (isalnum (*cp
) || *cp
== '_' || *cp
== '.' || *cp
== '$')
2860 if (*cp
== ':' || isspace (*cp
))
2862 /* Found end of label, so copy it and add it to the table. */
2863 pfnote ((CTAGS
) ? savenstr(lb
.buffer
, cp
-lb
.buffer
) : NULL
, TRUE
,
2864 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
2871 * Perl support by Bart Robinson <lomew@cs.utah.edu>
2872 * Perl sub names: look for /^sub[ \t\n]+[^ \t\n{]+/
2875 Perl_functions (inf
)
2886 linecharno
= charno
;
2887 charno
+= readline (&lb
, inf
);
2890 if (*cp
++ == 's' && *cp
++ == 'u' && *cp
++ == 'b' && isspace(*cp
++))
2892 while (*cp
&& isspace(*cp
))
2894 while (*cp
&& ! isspace(*cp
) && *cp
!= '{')
2896 pfnote ((CTAGS
) ? savenstr (lb
.buffer
, cp
-lb
.buffer
) : NULL
, TRUE
,
2897 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
2902 /* Added by Mosur Mohan, 4/22/88 */
2903 /* Pascal parsing */
2905 #define GET_NEW_LINE \
2907 linecharno = charno; lineno++; \
2908 charno += 1 + readline (&lb, inf); \
2913 * Locates tags for procedures & functions. Doesn't do any type- or
2914 * var-definitions. It does look for the keyword "extern" or
2915 * "forward" immediately following the procedure statement; if found,
2916 * the tag is skipped.
2919 Pascal_functions (inf
)
2922 struct linebuffer tline
; /* mostly copied from C_entries */
2924 int save_lineno
, save_len
;
2925 char c
, *cp
, *namebuf
;
2927 logical
/* each of these flags is TRUE iff: */
2928 incomment
, /* point is inside a comment */
2929 inquote
, /* point is inside '..' string */
2930 get_tagname
, /* point is after PROCEDURE/FUNCTION
2931 keyword, so next item = potential tag */
2932 found_tag
, /* point is after a potential tag */
2933 inparms
, /* point is within parameter-list */
2934 verify_tag
; /* point has passed the parm-list, so the
2935 next token will determine whether this
2936 is a FORWARD/EXTERN to be ignored, or
2937 whether it is a real tag */
2944 initbuffer (&tline
);
2946 incomment
= inquote
= FALSE
;
2947 found_tag
= FALSE
; /* have a proc name; check if extern */
2948 get_tagname
= FALSE
; /* have found "procedure" keyword */
2949 inparms
= FALSE
; /* found '(' after "proc" */
2950 verify_tag
= FALSE
; /* check if "extern" is ahead */
2952 /* long main loop to get next char */
2956 if (c
== '\0') /* if end of line */
2961 if (!((found_tag
&& verify_tag
) ||
2963 c
= *dbp
++; /* only if don't need *dbp pointing
2964 to the beginning of the name of
2965 the procedure or function */
2969 if (c
== '}') /* within { } comments */
2971 else if (c
== '*' && *dbp
== ')') /* within (* *) comments */
2988 inquote
= TRUE
; /* found first quote */
2990 case '{': /* found open { comment */
2994 if (*dbp
== '*') /* found open (* comment */
2999 else if (found_tag
) /* found '(' after tag, i.e., parm-list */
3002 case ')': /* end of parms list */
3007 if (found_tag
&& !inparms
) /* end of proc or fn stmt */
3014 if (found_tag
&& verify_tag
&& (*dbp
!= ' '))
3016 /* check if this is an "extern" declaration */
3019 if (lowcase (*dbp
== 'e'))
3021 if (tail ("extern")) /* superfluous, really! */
3027 else if (lowcase (*dbp
) == 'f')
3029 if (tail ("forward")) /* check for forward reference */
3035 if (found_tag
&& verify_tag
) /* not external proc, so make tag */
3039 pfnote (namebuf
, TRUE
,
3040 tline
.buffer
, save_len
, save_lineno
, save_lcno
);
3044 if (get_tagname
) /* grab name of proc or fn */
3049 /* save all values for later tagging */
3050 GROW_LINEBUFFER (tline
, strlen (lb
.buffer
) + 1);
3051 strcpy (tline
.buffer
, lb
.buffer
);
3052 save_lineno
= lineno
;
3053 save_lcno
= linecharno
;
3055 /* grab block name */
3056 for (cp
= dbp
+ 1; *cp
&& (!endtoken (*cp
)); cp
++)
3058 namebuf
= (CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
;
3059 dbp
= cp
; /* set dbp to e-o-token */
3060 save_len
= dbp
- lb
.buffer
+ 1;
3061 get_tagname
= FALSE
;
3065 /* and proceed to check for "extern" */
3067 else if (!incomment
&& !inquote
&& !found_tag
)
3069 /* check for proc/fn keywords */
3070 switch (lowcase (c
))
3073 if (tail ("rocedure")) /* c = 'p', dbp has advanced */
3077 if (tail ("unction"))
3082 } /* while not eof */
3084 free (tline
.buffer
);
3088 * lisp tag functions
3089 * look for (def or (DEF, quote or QUOTE
3093 register char *strp
;
3095 return ((strp
[1] == 'd' || strp
[1] == 'D')
3096 && (strp
[2] == 'e' || strp
[2] == 'E')
3097 && (strp
[3] == 'f' || strp
[3] == 'F'));
3102 register char *strp
;
3104 return ((*(++strp
) == 'q' || *strp
== 'Q')
3105 && (*(++strp
) == 'u' || *strp
== 'U')
3106 && (*(++strp
) == 'o' || *strp
== 'O')
3107 && (*(++strp
) == 't' || *strp
== 'T')
3108 && (*(++strp
) == 'e' || *strp
== 'E')
3109 && isspace(*(++strp
)));
3117 if (*dbp
== '\'') /* Skip prefix quote */
3119 else if (*dbp
== '(' && L_isquote (dbp
)) /* Skip "(quote " */
3122 while (isspace(*dbp
))
3125 for (cp
= dbp
/*+1*/;
3126 *cp
&& *cp
!= '(' && *cp
!= ' ' && *cp
!= ')';
3132 pfnote ((CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
, TRUE
,
3133 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
3137 Lisp_functions (inf
)
3146 linecharno
= charno
;
3147 charno
+= readline (&lb
, inf
);
3153 while (!isspace (*dbp
))
3155 while (isspace (*dbp
))
3161 /* Check for (foo::defmumble name-defined ... */
3164 while (*dbp
&& !isspace (*dbp
)
3165 && *dbp
!= ':' && *dbp
!= '(' && *dbp
!= ')');
3170 while (*dbp
== ':');
3172 if (L_isdef (dbp
- 1))
3174 while (!isspace (*dbp
))
3176 while (isspace (*dbp
))
3187 * Scheme tag functions
3188 * look for (def... xyzzy
3189 * look for (def... (xyzzy
3190 * look for (def ... ((...(xyzzy ....
3191 * look for (set! xyzzy
3197 Scheme_functions (inf
)
3206 linecharno
= charno
;
3207 charno
+= readline (&lb
, inf
);
3209 if (dbp
[0] == '(' &&
3210 (dbp
[1] == 'D' || dbp
[1] == 'd') &&
3211 (dbp
[2] == 'E' || dbp
[2] == 'e') &&
3212 (dbp
[3] == 'F' || dbp
[3] == 'f'))
3214 while (!isspace (*dbp
))
3216 /* Skip over open parens and white space */
3217 while (*dbp
&& (isspace (*dbp
) || *dbp
== '('))
3221 if (dbp
[0] == '(' &&
3222 (dbp
[1] == 'S' || dbp
[1] == 's') &&
3223 (dbp
[2] == 'E' || dbp
[2] == 'e') &&
3224 (dbp
[3] == 'T' || dbp
[3] == 't') &&
3225 (dbp
[4] == '!' || dbp
[4] == '!') &&
3228 while (!isspace (*dbp
))
3230 /* Skip over white space */
3231 while (isspace (*dbp
))
3245 /* Go till you get to white space or a syntactic break */
3247 *cp
&& *cp
!= '(' && *cp
!= ')' && !isspace (*cp
);
3250 pfnote ((CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
, TRUE
,
3251 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
3254 /* Find tags in TeX and LaTeX input files. */
3256 /* TEX_toktab is a table of TeX control sequences that define tags.
3257 Each TEX_tabent records one such control sequence.
3258 CONVERT THIS TO USE THE Stab TYPE!! */
3265 struct TEX_tabent
*TEX_toktab
= NULL
; /* Table with tag tokens */
3267 /* Default set of control sequences to put into TEX_toktab.
3268 The value of environment var TEXTAGS is prepended to this. */
3270 char *TEX_defenv
= "\
3271 :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem\
3272 :part:appendix:entry:index";
3275 struct TEX_tabent
*TEX_decode_env ();
3277 #if TeX_named_tokens
3281 char TEX_esc
= '\\';
3282 char TEX_opgrp
= '{';
3283 char TEX_clgrp
= '}';
3286 * TeX/LaTeX scanning loop.
3297 /* Select either \ or ! as escape character. */
3300 /* Initialize token table once from environment. */
3302 TEX_toktab
= TEX_decode_env ("TEXTAGS", TEX_defenv
);
3305 { /* Scan each line in file */
3307 linecharno
= charno
;
3308 charno
+= readline (&lb
, inf
);
3311 while (dbp
= etags_strchr (dbp
, TEX_esc
)) /* Look at each esc in line */
3317 linecharno
+= dbp
- lasthit
;
3319 i
= TEX_Token (lasthit
);
3323 lb
.buffer
, strlen (lb
.buffer
), lineno
, linecharno
);
3324 #if TeX_named_tokens
3325 TEX_getit (lasthit
, TEX_toktab
[i
].len
);
3327 break; /* We only save a line once */
3333 #define TEX_LESC '\\'
3334 #define TEX_SESC '!'
3337 /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
3338 chars accordingly. */
3345 while ((c
= getc (inf
)) != EOF
)
3347 /* Skip to next line if we hit the TeX comment char. */
3351 else if (c
== TEX_LESC
|| c
== TEX_SESC
)
3370 /* Read environment and prepend it to the default string.
3371 Build token table. */
3373 TEX_decode_env (evarname
, defenv
)
3377 register char *env
, *p
;
3379 struct TEX_tabent
*tab
;
3382 /* Append default string to environment. */
3383 env
= getenv (evarname
);
3387 env
= concat (env
, defenv
, "");
3389 /* Allocate a token table */
3390 for (size
= 1, p
= env
; p
;)
3391 if ((p
= etags_strchr (p
, ':')) && *(++p
))
3393 /* Add 1 to leave room for null terminator. */
3394 tab
= xnew (size
+ 1, struct TEX_tabent
);
3396 /* Unpack environment string into token table. Be careful about */
3397 /* zero-length strings (leading ':', "::" and trailing ':') */
3400 p
= etags_strchr (env
, ':');
3401 if (!p
) /* End of environment string. */
3402 p
= env
+ strlen (env
);
3404 { /* Only non-zero strings. */
3405 tab
[i
].name
= savenstr (env
, p
- env
);
3406 tab
[i
].len
= strlen (tab
[i
].name
);
3413 tab
[i
].name
= NULL
; /* Mark end of table. */
3421 #if TeX_named_tokens
3422 /* Record a tag defined by a TeX command of length LEN and starting at NAME.
3423 The name being defined actually starts at (NAME + LEN + 1).
3424 But we seem to include the TeX command in the tag name. */
3426 TEX_getit (name
, len
)
3430 char *p
= name
+ len
;
3435 /* Let tag name extend to next group close (or end of line) */
3436 while (*p
&& *p
!= TEX_clgrp
)
3438 pfnote (savenstr (name
, p
-name
), TRUE
,
3439 lb
.buffer
, strlen (lb
.buffer
), lineno
, linecharno
);
3443 /* If the text at CP matches one of the tag-defining TeX command names,
3444 return the pointer to the first occurrence of that command in TEX_toktab.
3445 Otherwise return -1.
3446 Keep the capital `T' in `Token' for dumb truncating compilers
3447 (this distinguishes it from `TEX_toktab' */
3454 for (i
= 0; TEX_toktab
[i
].len
> 0; i
++)
3455 if (strneq (TEX_toktab
[i
].name
, cp
, TEX_toktab
[i
].len
))
3461 * Prolog support (rewritten) by Anders Lindgren, Mar. 96
3463 * Assumes that the predicate starts at column 0.
3464 * Only the first clause of a predicate is added.
3467 Prolog_functions (inf
)
3471 void prolog_skip_comment ();
3488 linecharno
+= charno
;
3489 charno
= readline (&lb
, inf
);
3491 if (dbp
[0] == '\0') /* Empty line */
3493 else if (isspace (dbp
[0])) /* Not a predicate */
3495 else if (dbp
[0] == '/' && dbp
[1] == '*') /* comment. */
3496 prolog_skip_comment (&lb
, inf
, &lineno
, &linecharno
);
3497 else if (len
= prolog_pred (dbp
, last
))
3499 /* Predicate. Store the function name so that we only
3500 * generates a tag for the first clause. */
3502 last
= xnew(len
+ 1, char);
3503 else if (len
+ 1 > allocated
)
3504 last
= (char *) xrealloc(last
, len
+ 1);
3505 allocated
= len
+ 1;
3506 strncpy (last
, dbp
, len
);
3514 prolog_skip_comment (plb
, inf
)
3515 struct linebuffer
*plb
;
3522 for (cp
= plb
->buffer
; *cp
!= '\0'; cp
++)
3523 if (cp
[0] == '*' && cp
[1] == '/')
3526 linecharno
+= readline (plb
, inf
);
3532 * A predicate definition is added if it matches:
3533 * <beginning of line><Prolog Atom><whitespace>(
3535 * It is added to the tags database if it doesn't match the
3536 * name of the previous clause header.
3538 * Return the size of the name of the predicate, or 0 if no header
3542 prolog_pred (s
, last
)
3544 char *last
; /* Name of last clause. */
3552 pos
= prolog_atom(s
, 0);
3557 pos
+= prolog_white(s
, pos
);
3559 if ((s
[pos
] == '(') || (s
[pos
] == '.'))
3564 /* Save only the first clause. */
3565 if ((last
== NULL
) ||
3566 (len
!= strlen(last
)) ||
3567 (strncmp(s
, last
, len
) != 0))
3569 pfnote ((CTAGS
) ? savenstr (s
, len
) : NULL
, TRUE
,
3570 s
, pos
, lineno
, linecharno
);
3578 * Consume a Prolog atom.
3579 * Return the number of bytes consumed, or -1 if there was an error.
3581 * A prolog atom, in this context, could be one of:
3582 * - An alphanumeric sequence, starting with a lower case letter.
3583 * - A quoted arbitrary string. Single quotes can escape themselves.
3584 * Backslash quotes everything.
3587 prolog_atom (s
, pos
)
3595 if (islower(s
[pos
]) || (s
[pos
] == '_'))
3597 /* The atom is unquoted. */
3599 while (isalnum(s
[pos
]) || (s
[pos
] == '_'))
3603 return pos
- origpos
;
3605 else if (s
[pos
] == '\'')
3616 pos
++; /* A double quote */
3618 else if (s
[pos
] == '\0')
3619 /* Multiline quoted atoms are ignored. */
3621 else if (s
[pos
] == '\\')
3623 if (s
[pos
+1] == '\0')
3630 return pos
- origpos
;
3636 /* Consume whitespace. Return the number of bytes eaten. */
3638 prolog_white (s
, pos
)
3646 while (isspace(s
[pos
]))
3649 return pos
- origpos
;
3653 * Support for Erlang -- Anders Lindgren, Feb 1996.
3655 * Generates tags for functions, defines, and records.
3657 * Assumes that Erlang functions start at column 0.
3660 Erlang_functions (inf
)
3664 void erlang_attribute ();
3681 linecharno
+= charno
;
3682 charno
= readline (&lb
, inf
);
3684 if (dbp
[0] == '\0') /* Empty line */
3686 else if (isspace (dbp
[0])) /* Not function nor attribute */
3688 else if (dbp
[0] == '%') /* comment */
3690 else if (dbp
[0] == '"') /* Sometimes, strings start in column one */
3692 else if (dbp
[0] == '-') /* attribute, e.g. "-define" */
3694 erlang_attribute(dbp
);
3697 else if (len
= erlang_func (dbp
, last
))
3700 * Function. Store the function name so that we only
3701 * generates a tag for the first clause.
3704 last
= xnew(len
+ 1, char);
3705 else if (len
+ 1 > allocated
)
3706 last
= (char *) xrealloc(last
, len
+ 1);
3707 allocated
= len
+ 1;
3708 strncpy (last
, dbp
, len
);
3716 * A function definition is added if it matches:
3717 * <beginning of line><Erlang Atom><whitespace>(
3719 * It is added to the tags database if it doesn't match the
3720 * name of the previous clause header.
3722 * Return the size of the name of the function, or 0 if no function
3726 erlang_func (s
, last
)
3728 char *last
; /* Name of last clause. */
3731 int erlang_white ();
3736 pos
= erlang_atom(s
, 0);
3741 pos
+= erlang_white(s
, pos
);
3743 if (s
[pos
++] == '(')
3745 /* Save only the first clause. */
3746 if ((last
== NULL
) ||
3747 (len
!= strlen(last
)) ||
3748 (strncmp(s
, last
, len
) != 0))
3750 pfnote ((CTAGS
) ? savenstr (s
, len
) : NULL
, TRUE
,
3751 s
, pos
, lineno
, linecharno
);
3760 * Handle attributes. Currently, tags are generated for defines
3763 * They are on the form:
3764 * -define(foo, bar).
3765 * -define(Foo(M, N), M+N).
3766 * -record(graph, {vtab = notable, cyclic = true}).
3769 erlang_attribute (s
)
3773 int erlang_white ();
3778 if ((strncmp(s
, "-define", 7) == 0) ||
3779 (strncmp(s
, "-record", 7) == 0))
3782 pos
+= erlang_white(s
, pos
);
3784 if (s
[pos
++] == '(')
3786 pos
+= erlang_white(s
, pos
);
3788 if (len
= erlang_atom(s
, pos
))
3790 pfnote ((CTAGS
) ? savenstr (& s
[pos
], len
) : NULL
, TRUE
,
3791 s
, pos
+ len
, lineno
, linecharno
);
3800 * Consume an Erlang atom (or variable).
3801 * Return the number of bytes consumed, or -1 if there was an error.
3804 erlang_atom (s
, pos
)
3812 if (isalpha (s
[pos
]) || s
[pos
] == '_')
3814 /* The atom is unquoted. */
3816 while (isalnum (s
[pos
]) || s
[pos
] == '_')
3818 return pos
- origpos
;
3820 else if (s
[pos
] == '\'')
3831 else if (s
[pos
] == '\0')
3832 /* Multiline quoted atoms are ignored. */
3834 else if (s
[pos
] == '\\')
3836 if (s
[pos
+1] == '\0')
3843 return pos
- origpos
;
3849 /* Consume whitespace. Return the number of bytes eaten */
3851 erlang_white (s
, pos
)
3859 while (isspace (s
[pos
]))
3862 return pos
- origpos
;
3865 #ifdef ETAGS_REGEXPS
3866 /* Take a string like "/blah/" and turn it into "blah", making sure
3867 that the first and last characters are the same, and handling
3868 quoted separator characters. Actually, stops on the occurrence of
3869 an unquoted separator. Also turns "\t" into a Tab character.
3870 Returns pointer to terminating separator. Works in place. Null
3871 terminates name string. */
3873 scan_separators (name
)
3877 char *copyto
= name
;
3878 logical quoted
= FALSE
;
3880 for (++name
; *name
!= '\0'; ++name
)
3886 else if (*name
== sep
)
3890 /* Something else is quoted, so preserve the quote. */
3896 else if (*name
== '\\')
3898 else if (*name
== sep
)
3904 /* Terminate copied string. */
3909 /* Turn a name, which is an ed-style (but Emacs syntax) regular
3910 expression, into a real regular expression by compiling it. */
3912 add_regex (regexp_pattern
)
3913 char *regexp_pattern
;
3917 struct re_pattern_buffer
*patbuf
;
3919 if (regexp_pattern
== NULL
)
3921 /* Remove existing regexps. */
3927 if (regexp_pattern
[0] == '\0')
3929 error ("missing regexp", 0);
3932 if (regexp_pattern
[strlen(regexp_pattern
)-1] != regexp_pattern
[0])
3934 error ("%s: unterminated regexp", regexp_pattern
);
3937 name
= scan_separators (regexp_pattern
);
3938 if (regexp_pattern
[0] == '\0')
3940 error ("null regexp", 0);
3943 (void) scan_separators (name
);
3945 patbuf
= xnew (1, struct re_pattern_buffer
);
3946 patbuf
->translate
= NULL
;
3947 patbuf
->fastmap
= NULL
;
3948 patbuf
->buffer
= NULL
;
3949 patbuf
->allocated
= 0;
3951 err
= re_compile_pattern (regexp_pattern
, strlen (regexp_pattern
), patbuf
);
3954 error ("%s while compiling pattern", err
);
3959 if (num_patterns
== 1)
3960 patterns
= xnew (1, struct pattern
);
3962 patterns
= ((struct pattern
*)
3964 (num_patterns
* sizeof (struct pattern
))));
3965 patterns
[num_patterns
- 1].pattern
= patbuf
;
3966 patterns
[num_patterns
- 1].name_pattern
= savestr (name
);
3967 patterns
[num_patterns
- 1].error_signaled
= FALSE
;
3971 * Do the substitutions indicated by the regular expression and
3975 substitute (in
, out
, regs
)
3977 struct re_registers
*regs
;
3979 char *result
= NULL
, *t
;
3982 /* Pass 1: figure out how much size to allocate. */
3983 for (t
= out
; *t
; ++t
)
3990 fprintf (stderr
, "%s: pattern substitution ends prematurely\n",
3997 size
+= regs
->end
[dig
] - regs
->start
[dig
];
4002 /* Allocate space and do the substitutions. */
4003 result
= xnew (size
+ 1, char);
4012 /* Using "dig2" satisfies my debugger. Bleah. */
4013 int dig2
= *out
- '0';
4014 strncpy (result
+ size
, in
+ regs
->start
[dig2
],
4015 regs
->end
[dig2
] - regs
->start
[dig2
]);
4016 size
+= regs
->end
[dig2
] - regs
->start
[dig2
];
4019 result
[size
++] = *out
;
4022 result
[size
++] = *out
;
4024 result
[size
] = '\0';
4029 #endif /* ETAGS_REGEXPS */
4030 /* Initialize a linebuffer for use */
4032 initbuffer (linebuffer
)
4033 struct linebuffer
*linebuffer
;
4035 linebuffer
->size
= 200;
4036 linebuffer
->buffer
= xnew (200, char);
4040 * Read a line of text from `stream' into `linebuffer'.
4041 * Return the number of characters read from `stream',
4042 * which is the length of the line including the newline, if any.
4045 readline_internal (linebuffer
, stream
)
4046 struct linebuffer
*linebuffer
;
4047 register FILE *stream
;
4049 char *buffer
= linebuffer
->buffer
;
4050 register char *p
= linebuffer
->buffer
;
4051 register char *pend
;
4054 pend
= p
+ linebuffer
->size
; /* Separate to avoid 386/IX compiler bug. */
4058 register int c
= getc (stream
);
4061 linebuffer
->size
*= 2;
4062 buffer
= (char *) xrealloc (buffer
, linebuffer
->size
);
4063 p
+= buffer
- linebuffer
->buffer
;
4064 pend
= buffer
+ linebuffer
->size
;
4065 linebuffer
->buffer
= buffer
;
4075 if (p
> buffer
&& p
[-1] == '\r')
4090 return p
- buffer
+ chars_deleted
;
4094 * Like readline_internal, above, but try to match the input
4095 * line against any existing regular expressions.
4098 readline (linebuffer
, stream
)
4099 struct linebuffer
*linebuffer
;
4102 /* Read new line. */
4103 long result
= readline_internal (linebuffer
, stream
);
4104 #ifdef ETAGS_REGEXPS
4107 /* Match against all listed patterns. */
4108 for (i
= 0; i
< num_patterns
; ++i
)
4110 int match
= re_match (patterns
[i
].pattern
, linebuffer
->buffer
,
4111 (int)result
, 0, &patterns
[i
].regs
);
4116 if (!patterns
[i
].error_signaled
)
4118 error ("error while matching pattern %d", i
);
4119 patterns
[i
].error_signaled
= TRUE
;
4126 /* Match occurred. Construct a tag. */
4127 if (patterns
[i
].name_pattern
[0] != '\0')
4129 /* Make a named tag. */
4130 char *name
= substitute (linebuffer
->buffer
,
4131 patterns
[i
].name_pattern
,
4135 linebuffer
->buffer
, match
, lineno
, linecharno
);
4139 /* Make an unnamed tag. */
4141 linebuffer
->buffer
, match
, lineno
, linecharno
);
4146 #endif /* ETAGS_REGEXPS */
4152 * Read a file, but do no processing. This is used to do regexp
4153 * matching on files that have no language defined.
4156 just_read_file (inf
)
4165 linecharno
= charno
;
4166 charno
+= readline (&lb
, inf
) + 1;
4172 * Return a pointer to a space of size strlen(cp)+1 allocated
4173 * with xnew where the string CP has been copied.
4179 return savenstr (cp
, strlen (cp
));
4183 * Return a pointer to a space of size LEN+1 allocated with xnew where
4184 * the string CP has been copied for at most the first LEN characters.
4193 dp
= xnew (len
+ 1, char);
4194 strncpy (dp
, cp
, len
);
4200 * Return the ptr in sp at which the character c last
4201 * appears; NULL if not found
4203 * Identical to System V strrchr, included for portability.
4206 etags_strrchr (sp
, c
)
4207 register char *sp
, c
;
4222 * Return the ptr in sp at which the character c first
4223 * appears; NULL if not found
4225 * Identical to System V strchr, included for portability.
4228 etags_strchr (sp
, c
)
4229 register char *sp
, c
;
4239 /* Print error message and exit. */
4257 suggest_asking_for_help ()
4259 fprintf (stderr
, "\tTry `%s --help' for a complete list of options.\n",
4264 /* Print error message. `s1' is printf control string, `s2' is arg for it. */
4269 fprintf (stderr
, "%s: ", progname
);
4270 fprintf (stderr
, s1
, s2
);
4271 fprintf (stderr
, "\n");
4274 /* Return a newly-allocated string whose contents
4275 concatenate those of s1, s2, s3. */
4280 int len1
= strlen (s1
), len2
= strlen (s2
), len3
= strlen (s3
);
4281 char *result
= xnew (len1
+ len2
+ len3
+ 1, char);
4283 strcpy (result
, s1
);
4284 strcpy (result
+ len1
, s2
);
4285 strcpy (result
+ len1
+ len2
, s3
);
4286 result
[len1
+ len2
+ len3
] = '\0';
4291 /* Does the same work as the system V getcwd, but does not need to
4292 guess the buffer size in advance. */
4297 char *p
, path
[MAXPATHLEN
+ 1]; /* Fixed size is safe on MSDOS. */
4305 *p
++ = lowcase (*p
);
4307 return strdup (path
);
4308 #else /* not DOS_NT */
4311 char *path
= xnew (bufsize
, char);
4313 while (getcwd (path
, bufsize
) == NULL
)
4315 if (errno
!= ERANGE
)
4318 path
= xnew (bufsize
, char);
4322 #else /* not DOS_NT and not HAVE_GETCWD */
4323 struct linebuffer path
;
4327 pipe
= (FILE *) popen ("pwd 2>/dev/null", "r");
4328 if (pipe
== NULL
|| readline_internal (&path
, pipe
) == 0)
4333 #endif /* not HAVE_GETCWD */
4334 #endif /* not DOS_NT */
4337 /* Return a newly allocated string containing the filename
4338 of FILE relative to the absolute directory DIR (which
4339 should end with a slash). */
4341 relative_filename (file
, dir
)
4344 char *fp
, *dp
, *abs
, *res
;
4346 /* Find the common root of file and dir. */
4347 abs
= absolute_filename (file
, cwd
);
4350 while (*fp
++ == *dp
++)
4359 /* Build a sequence of "../" strings for the resulting relative filename. */
4360 for (dp
= etags_strchr (dp
+ 1, '/'), res
= "";
4362 dp
= etags_strchr (dp
+ 1, '/'))
4364 res
= concat (res
, "../", "");
4367 /* Add the filename relative to the common root of file and dir. */
4368 res
= concat (res
, fp
+ 1, "");
4374 /* Return a newly allocated string containing the
4375 absolute filename of FILE given CWD (which should
4376 end with a slash). */
4378 absolute_filename (file
, cwd
)
4381 char *slashp
, *cp
, *res
;
4383 if (absolutefn (file
))
4384 res
= concat (file
, "", "");
4386 res
= concat (cwd
, file
, "");
4388 /* Delete the "/dirname/.." and "/." substrings. */
4389 slashp
= etags_strchr (res
, '/');
4390 while (slashp
!= NULL
&& slashp
[0] != '\0')
4392 if (slashp
[1] == '.')
4394 if (slashp
[2] == '.'
4395 && (slashp
[3] == '/' || slashp
[3] == '\0'))
4400 while (cp
>= res
&& *cp
!= '/');
4403 strcpy (cp
, slashp
+ 3);
4405 else /* else (cp == res) */
4407 if (slashp
[3] != '\0')
4408 strcpy (cp
, slashp
+ 4);
4415 else if (slashp
[2] == '/' || slashp
[2] == '\0')
4417 strcpy (slashp
, slashp
+ 2);
4422 slashp
= etags_strchr (slashp
+ 1, '/');
4428 /* Return a newly allocated string containing the absolute
4429 filename of dir where FILE resides given CWD (which should
4430 end with a slash). */
4432 absolute_dirname (file
, cwd
)
4438 slashp
= etags_strrchr (file
, '/');
4443 res
= absolute_filename (file
, cwd
);
4449 /* Like malloc but get fatal error if memory is exhausted. */
4454 long *result
= (long *) malloc (size
);
4456 fatal ("virtual memory exhausted", 0);
4461 xrealloc (ptr
, size
)
4465 long *result
= (long *) realloc (ptr
, size
);
4467 fatal ("virtual memory exhausted");