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 11.63";
46 # include <sys/param.h>
54 # define MAXPATHLEN _MAX_PATH
59 /* On some systems, Emacs defines static as nothing for the sake
60 of unexec. We don't want that here since we don't use unexec. */
70 #include <sys/types.h>
73 #if !defined (S_ISREG) && defined (S_IFREG)
74 # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
81 #endif /* ETAGS_REGEXPS */
83 /* Define CTAGS to make the program "ctags" compatible with the usual one.
84 Let it undefined to make the program "etags", which makes emacs-style
85 tag tables and tags typedefs, #defines and struct/union/enum by default. */
93 /* Exit codes for success and failure. */
103 #define C_PLPL 0x00001 /* C++ */
104 #define C_STAR 0x00003 /* C* */
105 #define YACC 0x10000 /* yacc file */
107 #define streq(s,t) ((DEBUG &&!(s)&&!(t)&&(abort(),1)) || !strcmp(s,t))
108 #define strneq(s,t,n) ((DEBUG &&!(s)&&!(t)&&(abort(),1)) || !strncmp(s,t,n))
110 #define lowcase(c) tolower ((unsigned char)c)
112 #define iswhite(arg) (_wht[arg]) /* T if char is white */
113 #define begtoken(arg) (_btk[arg]) /* T if char can start token */
114 #define intoken(arg) (_itk[arg]) /* T if char can be in token */
115 #define endtoken(arg) (_etk[arg]) /* T if char ends tokens */
118 # define absolutefn(fn) (fn[0] == '/' \
119 || (fn[1] == ':' && fn[2] == '/'))
121 # define absolutefn(fn) (fn[0] == '/')
126 * xnew -- allocate storage
128 * SYNOPSIS: Type *xnew (int n, Type);
130 #define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type)))
135 { /* sorting structure */
136 char *name
; /* function or type name */
137 char *file
; /* file name */
138 logical is_func
; /* use pattern or line no */
139 logical been_warned
; /* set if noticed dup */
140 int lno
; /* line number tag is on */
141 long cno
; /* character number line starts on */
142 char *pat
; /* search pattern */
143 struct nd_st
*left
, *right
; /* left and right sons */
146 extern char *getenv ();
149 char *savenstr (), *savestr ();
150 char *etags_strchr (), *etags_strrchr ();
151 char *etags_getcwd ();
152 char *relative_filename (), *absolute_filename (), *absolute_dirname ();
153 long *xmalloc (), *xrealloc ();
155 typedef void Lang_function ();
156 #if FALSE /* many compilers barf on this */
157 Lang_function Asm_labels
;
158 Lang_function default_C_entries
;
159 Lang_function C_entries
;
160 Lang_function Cplusplus_entries
;
161 Lang_function Cstar_entries
;
162 Lang_function Erlang_functions
;
163 Lang_function Fortran_functions
;
164 Lang_function Yacc_entries
;
165 Lang_function Lisp_functions
;
166 Lang_function Pascal_functions
;
167 Lang_function Perl_functions
;
168 Lang_function Prolog_functions
;
169 Lang_function Scheme_functions
;
170 Lang_function TeX_functions
;
171 Lang_function just_read_file
;
172 #else /* so let's write it this way */
175 void default_C_entries ();
176 void plain_C_entries ();
177 void Cplusplus_entries ();
178 void Cstar_entries ();
179 void Erlang_functions ();
180 void Fortran_functions ();
181 void Yacc_entries ();
182 void Lisp_functions ();
183 void Pascal_functions ();
184 void Perl_functions ();
185 void Prolog_functions ();
186 void Scheme_functions ();
187 void TeX_functions ();
188 void just_read_file ();
191 Lang_function
*get_language_from_name ();
192 Lang_function
*get_language_from_interpreter ();
193 Lang_function
*get_language_from_suffix ();
194 int total_size_of_entries ();
196 long readline_internal ();
202 void suggest_asking_for_help ();
203 void fatal (), pfatal ();
204 void find_entries ();
210 void process_file ();
215 char searchar
= '/'; /* use /.../ searches */
217 int lineno
; /* line number of current line */
218 long charno
; /* current character number */
219 long linecharno
; /* charno of start of line */
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] != '/')
860 cwd
= concat (cwd
, "/", "");
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
, "-"))
880 /* Switch redirected `stdout' to binary mode (setting `_fmode'
881 doesn't take effect until after `stdout' is already open). */
882 if (!isatty (fileno (stdout
)))
883 setmode (fileno (stdout
), O_BINARY
);
887 tagf
= fopen (tagfile
, append_to_tagfile
? "a" : "w");
893 * Loop through files finding functions.
895 for (i
= 0; i
< current_arg
; ++i
)
897 switch (argbuffer
[i
].arg_type
)
900 lang_func
= argbuffer
[i
].function
;
904 add_regex (argbuffer
[i
].what
);
909 while ((this_file
= gfnames (argbuffer
[i
].what
, &got_err
)) != NULL
)
913 error ("Can't find file %s\n", this_file
);
918 this_file
= massage_name (this_file
);
921 this_file
= argbuffer
[i
].what
;
923 /* Input file named "-" means read file names from stdin
925 if (streq (this_file
, "-"))
926 while (readline_internal (&filename_lb
, stdin
) > 0)
927 process_file (filename_lb
.buffer
);
929 process_file (this_file
);
939 while (nincluded_files
-- > 0)
940 fprintf (tagf
, "\f\n%s,include\n", *included_files
++);
946 /* If CTAGS, we are here. process_file did not write the tags yet,
947 because we want them ordered. Let's do it now. */
950 tagf
= fopen (tagfile
, append_to_tagfile
? "a" : "w");
960 for (i
= 0; i
< current_arg
; ++i
)
962 if (argbuffer
[i
].arg_type
!= at_filename
)
965 "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS",
966 tagfile
, argbuffer
[i
].what
, tagfile
);
967 if (system (cmd
) != GOOD
)
968 fatal ("failed to execute shell command");
970 append_to_tagfile
= TRUE
;
973 tagf
= fopen (tagfile
, append_to_tagfile
? "a" : "w");
982 sprintf (cmd
, "sort %s -o %s", tagfile
, tagfile
);
990 * Return a Lang_function given the name.
993 get_language_from_name (name
)
996 struct lang_entry
*lang
;
999 for (lang
= lang_names
; lang
->name
!= NULL
; lang
++)
1001 if (streq (name
, lang
->name
))
1002 return lang
->function
;
1005 fprintf (stderr
, "%s: language \"%s\" not recognized.\n",
1007 suggest_asking_for_help ();
1009 /* This point should never be reached. The function should either
1010 return a function pointer or never return. Note that a NULL
1011 pointer cannot be considered as an error, as it means that the
1012 language has not been explicitely imposed by the user ("auto"). */
1013 return NULL
; /* avoid warnings from compiler */
1018 * Return a Lang_function given the interpreter name.
1021 get_language_from_interpreter (interpreter
)
1024 struct lang_entry
*lang
;
1027 if (interpreter
== NULL
)
1029 for (lang
= lang_names
; lang
->name
!= NULL
; lang
++)
1030 if (lang
->interpreters
!= NULL
)
1031 for (iname
= lang
->interpreters
; *iname
!= NULL
; iname
++)
1032 if (streq (*iname
, interpreter
))
1033 return lang
->function
;
1041 * Return a Lang_function given the file suffix.
1044 get_language_from_suffix (suffix
)
1047 struct lang_entry
*lang
;
1052 for (lang
= lang_names
; lang
->name
!= NULL
; lang
++)
1053 if (lang
->suffixes
!= NULL
)
1054 for (ext
= lang
->suffixes
; *ext
!= NULL
; ext
++)
1055 if (streq (*ext
, suffix
))
1056 return lang
->function
;
1063 * This routine is called on each file argument.
1069 struct stat stat_buf
;
1074 for (p
= file
; *p
!= '\0'; p
++)
1079 if (stat (file
, &stat_buf
) == 0 && !S_ISREG (stat_buf
.st_mode
))
1081 fprintf (stderr
, "Skipping %s: it is not a regular file.\n", file
);
1084 if (streq (file
, tagfile
) && !streq (tagfile
, "-"))
1086 fprintf (stderr
, "Skipping inclusion of %s in self.\n", file
);
1089 inf
= fopen (file
, "r");
1096 find_entries (file
, inf
);
1102 if (absolutefn (file
))
1104 /* file is an absolute filename. Canonicalise it. */
1105 filename
= absolute_filename (file
, cwd
);
1109 /* file is a filename relative to cwd. Make it relative
1110 to the directory of the tags file. */
1111 filename
= relative_filename (file
, tagfiledir
);
1113 fprintf (tagf
, "\f\n%s,%d\n", filename
, total_size_of_entries (head
));
1122 * This routine sets up the boolean pseudo-functions which work
1123 * by setting boolean flags dependent upon the corresponding character
1124 * Every char which is NOT in that string is not a white char. Therefore,
1125 * all of the array "_wht" is set to FALSE, and then the elements
1126 * subscripted by the chars in "white" are set to TRUE. Thus "_wht"
1127 * of a char is TRUE if it is the string "white", else FALSE.
1135 for (i
= 0; i
< 0177; i
++)
1136 _wht
[i
] = _etk
[i
] = _itk
[i
] = _btk
[i
] = FALSE
;
1137 for (sp
= white
; *sp
; sp
++)
1139 for (sp
= endtk
; *sp
; sp
++)
1141 for (sp
= intk
; *sp
; sp
++)
1143 for (sp
= begtk
; *sp
; sp
++)
1145 _wht
[0] = _wht
['\n'];
1146 _etk
[0] = _etk
['\n'];
1147 _btk
[0] = _btk
['\n'];
1148 _itk
[0] = _itk
['\n'];
1152 * This routine opens the specified file and calls the function
1153 * which finds the function and type definitions.
1156 find_entries (file
, inf
)
1161 Lang_function
*function
;
1162 NODE
*old_last_node
;
1163 extern NODE
*last_node
;
1166 /* Memory leakage here: the memory block pointed by curfile is never
1167 released. The amount of memory leaked here is the sum of the
1168 lengths of the input file names. */
1169 curfile
= savestr (file
);
1171 /* If user specified a language, use it. */
1172 function
= lang_func
;
1173 if (function
!= NULL
)
1180 cp
= etags_strrchr (file
, '.');
1184 function
= get_language_from_suffix (cp
);
1185 if (function
!= NULL
)
1193 /* Look for sharp-bang as the first two characters. */
1194 if (readline_internal (&lb
, inf
) > 2
1195 && lb
.buffer
[0] == '#'
1196 && lb
.buffer
[1] == '!')
1200 /* Set lp to point at the first char after the last slash in the
1201 line or, if no slashes, at the first nonblank. Then set cp to
1202 the first successive blank and terminate the string. */
1203 lp
= etags_strrchr (lb
.buffer
+2, '/');
1207 for (lp
= lb
.buffer
+2; *lp
!= '\0' && isspace (*lp
); lp
++)
1209 for (cp
= lp
; *cp
!= '\0' && !isspace (*cp
); cp
++)
1213 if (strlen (lp
) > 0)
1215 function
= get_language_from_interpreter (lp
);
1216 if (function
!= NULL
)
1227 old_last_node
= last_node
;
1228 Fortran_functions (inf
);
1230 /* No Fortran entries found. Try C. */
1231 if (old_last_node
== last_node
)
1234 default_C_entries (inf
);
1242 pfnote (name
, is_func
, linestart
, linelen
, lno
, cno
)
1243 char *name
; /* tag name, or NULL if unnamed */
1244 logical is_func
; /* tag is a function */
1245 char *linestart
; /* start of the line where tag is */
1246 int linelen
; /* length of the line where tag is */
1247 int lno
; /* line number */
1248 long cno
; /* character number */
1252 if (CTAGS
&& name
== NULL
)
1255 np
= xnew (1, NODE
);
1257 /* If ctags mode, change name "main" to M<thisfilename>. */
1258 if (CTAGS
&& !cxref_style
&& streq (name
, "main"))
1260 register char *fp
= etags_strrchr (curfile
, '/');
1261 np
->name
= concat ("M", fp
== 0 ? curfile
: fp
+ 1, "");
1262 fp
= etags_strrchr (np
->name
, '.');
1263 if (fp
&& fp
[1] != '\0' && fp
[2] == '\0')
1268 np
->been_warned
= FALSE
;
1270 np
->is_func
= is_func
;
1272 /* Our char numbers are 0-base, because of C language tradition?
1273 ctags compatibility? old versions compatibility? I don't know.
1274 Anyway, since emacs's are 1-base we expect etags.el to take care
1275 of the difference. If we wanted to have 1-based numbers, we would
1276 uncomment the +1 below. */
1277 np
->cno
= cno
/* + 1 */ ;
1278 np
->left
= np
->right
= NULL
;
1279 if (CTAGS
&& !cxref_style
)
1281 if (strlen (linestart
) < 50)
1282 np
->pat
= concat (linestart
, "$", "");
1284 np
->pat
= savenstr (linestart
, 50);
1287 np
->pat
= savenstr (linestart
, linelen
);
1289 add_node (np
, &head
);
1294 * recurse on left children, iterate on right children.
1298 register NODE
*node
;
1302 register NODE
*node_right
= node
->right
;
1303 free_tree (node
->left
);
1304 if (node
->name
!= NULL
)
1307 free ((char *) node
);
1314 * Adds a node to the tree of nodes. In etags mode, we don't keep
1315 * it sorted; we just keep a linear list. In ctags mode, maintain
1316 * an ordered tree, with no attempt at balancing.
1318 * add_node is the only function allowed to add nodes, so it can
1321 NODE
*last_node
= NULL
;
1323 add_node (node
, cur_node_p
)
1324 NODE
*node
, **cur_node_p
;
1327 register NODE
*cur_node
= *cur_node_p
;
1329 if (cur_node
== NULL
)
1339 if (last_node
== NULL
)
1340 fatal ("internal error in add_node", 0);
1341 last_node
->right
= node
;
1347 dif
= strcmp (node
->name
, cur_node
->name
);
1350 * If this tag name matches an existing one, then
1351 * do not add the node, but maybe print a warning.
1355 if (streq (node
->file
, cur_node
->file
))
1359 fprintf (stderr
, "Duplicate entry in file %s, line %d: %s\n",
1360 node
->file
, lineno
, node
->name
);
1361 fprintf (stderr
, "Second entry ignored\n");
1364 else if (!cur_node
->been_warned
&& !no_warnings
)
1368 "Duplicate entry in files %s and %s: %s (Warning only)\n",
1369 node
->file
, cur_node
->file
, node
->name
);
1370 cur_node
->been_warned
= TRUE
;
1375 /* Actually add the node */
1376 add_node (node
, dif
< 0 ? &cur_node
->left
: &cur_node
->right
);
1382 register NODE
*node
;
1389 /* Output subentries that precede this one */
1390 put_entries (node
->left
);
1392 /* Output this entry */
1396 if (node
->name
!= NULL
)
1397 fprintf (tagf
, "%s\177%s\001%d,%d\n",
1398 node
->pat
, node
->name
, node
->lno
, node
->cno
);
1400 fprintf (tagf
, "%s\177%d,%d\n",
1401 node
->pat
, node
->lno
, node
->cno
);
1405 if (node
->name
== NULL
)
1406 error ("internal error: NULL name in ctags mode.", 0);
1411 fprintf (stdout
, "%s %s %d\n",
1412 node
->name
, node
->file
, (node
->lno
+ 63) / 64);
1414 fprintf (stdout
, "%-16s %3d %-16s %s\n",
1415 node
->name
, node
->lno
, node
->file
, node
->pat
);
1419 fprintf (tagf
, "%s\t%s\t", node
->name
, node
->file
);
1423 putc (searchar
, tagf
);
1426 for (sp
= node
->pat
; *sp
; sp
++)
1428 if (*sp
== '\\' || *sp
== searchar
)
1432 putc (searchar
, tagf
);
1435 { /* a typedef; text pattern inadequate */
1436 fprintf (tagf
, "%d", node
->lno
);
1442 /* Output subentries that follow this one */
1443 put_entries (node
->right
);
1446 /* Length of a number's decimal representation. */
1454 for (; num
; num
/= 10)
1460 * Return total number of characters that put_entries will output for
1461 * the nodes in the subtree of the specified node. Works only if
1462 * we are not ctags, but called only in that case. This count
1463 * is irrelevant with the new tags.el, but is still supplied for
1464 * backward compatibility.
1467 total_size_of_entries (node
)
1468 register NODE
*node
;
1476 for (; node
; node
= node
->right
)
1478 /* Count left subentries. */
1479 total
+= total_size_of_entries (node
->left
);
1481 /* Count this entry */
1482 total
+= strlen (node
->pat
) + 1;
1483 total
+= number_len ((long) node
->lno
) + 1 + number_len (node
->cno
) + 1;
1484 if (node
->name
!= NULL
)
1485 total
+= 1 + strlen (node
->name
); /* \001name */
1492 * The C symbol tables.
1496 st_none
, st_C_objprot
, st_C_objimpl
, st_C_objend
, st_C_gnumacro
,
1497 st_C_struct
, st_C_enum
, st_C_define
, st_C_typedef
, st_C_typespec
1500 /* Feed stuff between (but not including) %[ and %] lines to:
1501 gperf -c -k1,3 -o -p -r -t
1503 struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
1505 @interface, 0, st_C_objprot
1506 @protocol, 0, st_C_objprot
1507 @implementation,0, st_C_objimpl
1508 @end, 0, st_C_objend
1509 class, C_PLPL, st_C_struct
1510 domain, C_STAR, st_C_struct
1511 union, 0, st_C_struct
1512 struct, 0, st_C_struct
1514 typedef, 0, st_C_typedef
1515 define, 0, st_C_define
1516 long, 0, st_C_typespec
1517 short, 0, st_C_typespec
1518 int, 0, st_C_typespec
1519 char, 0, st_C_typespec
1520 float, 0, st_C_typespec
1521 double, 0, st_C_typespec
1522 signed, 0, st_C_typespec
1523 unsigned, 0, st_C_typespec
1524 auto, 0, st_C_typespec
1525 void, 0, st_C_typespec
1526 extern, 0, st_C_typespec
1527 static, 0, st_C_typespec
1528 const, 0, st_C_typespec
1529 volatile, 0, st_C_typespec
1530 # DEFUN used in emacs, the next three used in glibc (SYSCALL only for mach).
1531 DEFUN, 0, st_C_gnumacro
1532 SYSCALL, 0, st_C_gnumacro
1533 ENTRY, 0, st_C_gnumacro
1534 PSEUDO, 0, st_C_gnumacro
1535 # These are defined inside C functions, so currently they are not met.
1536 # EXFUN used in glibc, DEFVAR_* in emacs.
1537 #EXFUN, 0, st_C_gnumacro
1538 #DEFVAR_, 0, st_C_gnumacro
1540 and replace lines between %< and %> with its output. */
1542 /* C code produced by gperf version 1.8.1 (K&R C version) */
1543 /* Command-line: gperf -c -k1,3 -o -p -r -t */
1546 struct C_stab_entry
{ char *name
; int c_ext
; enum sym_type type
; };
1548 #define MIN_WORD_LENGTH 3
1549 #define MAX_WORD_LENGTH 15
1550 #define MIN_HASH_VALUE 7
1551 #define MAX_HASH_VALUE 63
1554 57 is the maximum key range
1562 static unsigned char hash_table
[] =
1564 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1565 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1566 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1567 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1568 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1569 63, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1570 63, 63, 63, 63, 17, 63, 63, 63, 4, 14,
1571 4, 63, 63, 63, 63, 63, 63, 63, 63, 63,
1572 8, 63, 63, 0, 23, 63, 63, 63, 63, 63,
1573 63, 63, 63, 63, 63, 63, 63, 28, 63, 28,
1574 10, 31, 27, 18, 63, 6, 63, 63, 26, 1,
1575 11, 2, 29, 63, 29, 16, 26, 13, 15, 63,
1576 63, 63, 63, 63, 63, 63, 63, 63,
1578 return len
+ hash_table
[str
[2]] + hash_table
[str
[0]];
1581 struct C_stab_entry
*
1582 in_word_set (str
, len
)
1587 static struct C_stab_entry wordlist
[] =
1589 {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
1590 {"SYSCALL", 0, st_C_gnumacro
},
1591 {"",}, {"",}, {"",}, {"",}, {"",},
1592 {"DEFUN", 0, st_C_gnumacro
},
1593 {"",}, {"",}, {"",},
1594 {"domain", C_STAR
, st_C_struct
},
1595 {"",}, {"",}, {"",}, {"",}, {"",},
1596 {"short", 0, st_C_typespec
},
1597 {"union", 0, st_C_struct
},
1598 {"void", 0, st_C_typespec
},
1600 {"PSEUDO", 0, st_C_gnumacro
},
1601 {"double", 0, st_C_typespec
},
1603 {"@end", 0, st_C_objend
},
1604 {"@implementation", 0, st_C_objimpl
},
1605 {"float", 0, st_C_typespec
},
1606 {"int", 0, st_C_typespec
},
1608 {"unsigned", 0, st_C_typespec
},
1609 {"@interface", 0, st_C_objprot
},
1611 {"signed", 0, st_C_typespec
},
1612 {"long", 0, st_C_typespec
},
1613 {"ENTRY", 0, st_C_gnumacro
},
1614 {"define", 0, st_C_define
},
1615 {"const", 0, st_C_typespec
},
1616 {"",}, {"",}, {"",},
1617 {"enum", 0, st_C_enum
},
1618 {"volatile", 0, st_C_typespec
},
1619 {"static", 0, st_C_typespec
},
1620 {"struct", 0, st_C_struct
},
1621 {"",}, {"",}, {"",},
1622 {"@protocol", 0, st_C_objprot
},
1624 {"auto", 0, st_C_typespec
},
1626 {"char", 0, st_C_typespec
},
1627 {"class", C_PLPL
, st_C_struct
},
1628 {"typedef", 0, st_C_typedef
},
1629 {"extern", 0, st_C_typespec
},
1632 if (len
<= MAX_WORD_LENGTH
&& len
>= MIN_WORD_LENGTH
)
1634 register int key
= hash (str
, len
);
1636 if (key
<= MAX_HASH_VALUE
&& key
>= MIN_HASH_VALUE
)
1638 register char *s
= wordlist
[key
].name
;
1640 if (*s
== *str
&& !strncmp (str
+ 1, s
+ 1, len
- 1))
1641 return &wordlist
[key
];
1649 C_symtype(str
, len
, c_ext
)
1654 register struct C_stab_entry
*se
= in_word_set(str
, len
);
1656 if (se
== NULL
|| (se
->c_ext
&& !(c_ext
& se
->c_ext
)))
1662 * C functions are recognized using a simple finite automaton.
1663 * funcdef is its state variable.
1667 fnone
, /* nothing seen */
1668 ftagseen
, /* function-like tag seen */
1669 fstartlist
, /* just after open parenthesis */
1670 finlist
, /* in parameter list */
1671 flistseen
, /* after parameter list */
1672 fignore
/* before open brace */
1677 * typedefs are recognized using a simple finite automaton.
1678 * typdef is its state variable.
1682 tnone
, /* nothing seen */
1683 ttypedseen
, /* typedef keyword seen */
1684 tinbody
, /* inside typedef body */
1685 tend
, /* just before typedef tag */
1686 tignore
/* junk after typedef tag */
1691 * struct-like structures (enum, struct and union) are recognized
1692 * using another simple finite automaton. `structdef' is its state
1697 snone
, /* nothing seen yet */
1698 skeyseen
, /* struct-like keyword seen */
1699 stagseen
, /* struct-like tag seen */
1700 scolonseen
, /* colon seen after struct-like tag */
1701 sinbody
/* in struct body: recognize member func defs*/
1705 * When structdef is stagseen, scolonseen, or sinbody, structtag is the
1706 * struct tag, and structtype is the type of the preceding struct-like
1709 char *structtag
= "<uninited>";
1710 enum sym_type structtype
;
1713 * When objdef is different from onone, objtag is the name of the class.
1715 char *objtag
= "<uninited>";
1718 * Yet another little state machine to deal with preprocessor lines.
1722 dnone
, /* nothing seen */
1723 dsharpseen
, /* '#' seen as first char on line */
1724 ddefineseen
, /* '#' and 'define' seen */
1725 dignorerest
/* ignore rest of line */
1729 * State machine for Objective C protocols and implementations.
1733 onone
, /* nothing seen */
1734 oprotocol
, /* @interface or @protocol seen */
1735 oimplementation
, /* @implementations seen */
1736 otagseen
, /* class name seen */
1737 oparenseen
, /* parenthesis before category seen */
1738 ocatseen
, /* category name seen */
1739 oinbody
, /* in @implementation body */
1740 omethodsign
, /* in @implementation body, after +/- */
1741 omethodtag
, /* after method name */
1742 omethodcolon
, /* after method colon */
1743 omethodparm
, /* after method parameter */
1744 oignore
/* wait for @end */
1748 * Set this to TRUE, and the next token considered is called a function.
1749 * Used only for GNU emacs's function-defining macros.
1751 logical next_token_is_func
;
1754 * TRUE in the rules part of a yacc file, FALSE outside (parse as C).
1759 * methodlen is the length of the method name stored in token_name.
1765 * checks to see if the current token is at the start of a
1766 * function, or corresponds to a typedef, or is a struct/union/enum
1769 * *IS_FUNC gets TRUE iff the token is a function or macro with args.
1770 * C_EXT is which language we are looking at.
1772 * In the future we will need some way to adjust where the end of
1773 * the token is; for instance, implementing the C++ keyword
1774 * `operator' properly will adjust the end of the token to be after
1775 * whatever follows `operator'.
1783 * next_token_is_func IN OUT
1787 consider_token (str
, len
, c
, c_ext
, cblev
, parlev
, is_func
)
1788 register char *str
; /* IN: token pointer */
1789 register int len
; /* IN: token length */
1790 register char c
; /* IN: first char after the token */
1791 int c_ext
; /* IN: C extensions mask */
1792 int cblev
; /* IN: curly brace level */
1793 int parlev
; /* IN: parenthesis level */
1794 logical
*is_func
; /* OUT: function found */
1796 enum sym_type toktype
= C_symtype (str
, len
, c_ext
);
1799 * Advance the definedef state machine.
1804 /* We're not on a preprocessor line. */
1807 if (toktype
== st_C_define
)
1809 definedef
= ddefineseen
;
1813 definedef
= dignorerest
;
1818 * Make a tag for any macro, unless it is a constant
1819 * and constantypedefs is FALSE.
1821 definedef
= dignorerest
;
1822 *is_func
= (c
== '(');
1823 if (!*is_func
&& !constantypedefs
)
1830 error ("internal error: definedef value.", 0);
1839 if (toktype
== st_C_typedef
)
1842 typdef
= ttypedseen
;
1858 /* Do not return here, so the structdef stuff has a chance. */
1872 * This structdef business is currently only invoked when cblev==0.
1873 * It should be recursively invoked whatever the curly brace level,
1874 * and a stack of states kept, to allow for definitions of structs
1877 * This structdef business is NOT invoked when we are ctags and the
1878 * file is plain C. This is because a struct tag may have the same
1879 * name as another tag, and this loses with ctags.
1881 * This if statement deals with the typdef state machine as
1882 * follows: if typdef==ttypedseen and token is struct/union/class/enum,
1883 * return FALSE. All the other code here is for the structdef
1890 if (typdef
== ttypedseen
1891 || (typedefs_and_cplusplus
&& cblev
== 0 && structdef
== snone
))
1893 structdef
= skeyseen
;
1894 structtype
= toktype
;
1898 if (structdef
== skeyseen
)
1900 /* Save the tag for struct/union/class, for functions that may be
1902 if (structtype
== st_C_struct
)
1903 structtag
= savenstr (str
, len
);
1905 structtag
= "<enum>";
1906 structdef
= stagseen
;
1910 /* Avoid entering funcdef stuff if typdef is going on. */
1911 if (typdef
!= tnone
)
1917 /* Detect GNU macros. */
1918 if (definedef
== dnone
&& toktype
== st_C_gnumacro
)
1920 next_token_is_func
= TRUE
;
1923 if (next_token_is_func
)
1925 next_token_is_func
= FALSE
;
1932 * Detecting Objective C constructs.
1943 objdef
= oimplementation
;
1947 case oimplementation
:
1948 /* Save the class tag for functions that may be defined inside. */
1949 objtag
= savenstr (str
, len
);
1953 /* Save the class tag for categories. */
1954 objtag
= savenstr (str
, len
);
1967 objdef
= omethodtag
;
1969 GROW_LINEBUFFER (token_name
, methodlen
+1);
1970 strncpy (token_name
.buffer
, str
, len
);
1971 token_name
.buffer
[methodlen
] = '\0';
1977 objdef
= omethodparm
;
1982 objdef
= omethodtag
;
1984 GROW_LINEBUFFER (token_name
, methodlen
+1);
1985 strncat (token_name
.buffer
, str
, len
);
1990 if (toktype
== st_C_objend
)
1992 /* Memory leakage here: the string pointed by objtag is
1993 never released, because many tests would be needed to
1994 avoid breaking on incorrect input code. The amount of
1995 memory leaked here is the sum of the lengths of the
2007 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2008 funcdef
= fnone
; /* should be useless */
2011 if (funcdef
== fnone
)
2024 * This routine finds functions, typedefs, #define's and
2025 * struct/union/enum definitions in C syntax and adds them
2039 #define current_lb_is_new (newndx == curndx)
2040 #define switch_line_buffers() (curndx = 1 - curndx)
2042 #define curlb (lbs[curndx].lb)
2043 #define othlb (lbs[1-curndx].lb)
2044 #define newlb (lbs[newndx].lb)
2045 #define curlinepos (lbs[curndx].linepos)
2046 #define othlinepos (lbs[1-curndx].linepos)
2047 #define newlinepos (lbs[newndx].linepos)
2049 #define CNL_SAVE_DEFINEDEF \
2051 curlinepos = charno; \
2053 linecharno = charno; \
2054 charno += readline (&curlb, inf); \
2055 lp = curlb.buffer; \
2062 CNL_SAVE_DEFINEDEF; \
2063 if (savetok.valid) \
2066 savetok.valid = FALSE; \
2068 definedef = dnone; \
2071 /* Ideally this macro should never be called wihen tok.valid is FALSE,
2072 but this would mean that the state machines always guess right. */
2073 #define make_tag(isfun) do \
2075 char *name = NULL; \
2076 if (CTAGS || tok.named) \
2077 name = savestr (token_name.buffer); \
2078 pfnote (name, isfun, tok.buffer, tok.linelen, tok.lineno, tok.linepos); \
2079 tok.valid = FALSE; \
2083 C_entries (c_ext
, inf
)
2084 int c_ext
; /* extension of C */
2085 FILE *inf
; /* input file */
2087 register char c
; /* latest char read; '\0' for end of line */
2088 register char *lp
; /* pointer one beyond the character `c' */
2089 int curndx
, newndx
; /* indices for current and new lb */
2090 TOKEN tok
; /* latest token read */
2091 register int tokoff
; /* offset in line of start of current token */
2092 register int toklen
; /* length of current token */
2093 int cblev
; /* current curly brace level */
2094 int parlev
; /* current parenthesis level */
2095 logical incomm
, inquote
, inchar
, quotednl
, midtoken
;
2097 TOKEN savetok
; /* token saved during preprocessor handling */
2100 curndx
= newndx
= 0;
2106 funcdef
= fnone
; typdef
= tnone
; structdef
= snone
;
2107 definedef
= dnone
; objdef
= onone
;
2108 next_token_is_func
= yacc_rules
= FALSE
;
2109 midtoken
= inquote
= inchar
= incomm
= quotednl
= FALSE
;
2110 tok
.valid
= savetok
.valid
= FALSE
;
2113 cplpl
= c_ext
& C_PLPL
;
2120 /* If we're at the end of the line, the next character is a
2121 '\0'; don't skip it, because it's the thing that tells us
2122 to read the next line. */
2143 /* Newlines inside comments do not end macro definitions in
2158 /* Newlines inside strings do not end macro definitions
2159 in traditional cpp, even though compilers don't
2160 usually accept them. */
2171 /* Hmmm, something went wrong. */
2185 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2190 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2200 else if (/* cplpl && */ *lp
== '/')
2208 if ((c_ext
& YACC
) && *lp
== '%')
2210 /* entering or exiting rules section in yacc file */
2212 definedef
= dnone
; funcdef
= fnone
;
2213 typdef
= tnone
; structdef
= snone
;
2214 next_token_is_func
= FALSE
;
2215 midtoken
= inquote
= inchar
= incomm
= quotednl
= FALSE
;
2217 yacc_rules
= !yacc_rules
;
2223 if (definedef
== dnone
)
2226 logical cpptoken
= TRUE
;
2228 /* Look back on this line. If all blanks, or nonblanks
2229 followed by an end of comment, this is a preprocessor
2231 for (cp
= newlb
.buffer
; cp
< lp
-1; cp
++)
2234 if (*cp
== '*' && *(cp
+1) == '/')
2243 definedef
= dsharpseen
;
2244 } /* if (definedef == dnone) */
2250 /* Consider token only if some complicated conditions are satisfied. */
2251 if ((definedef
!= dnone
2252 || (cblev
== 0 && structdef
!= scolonseen
)
2253 || (cblev
== 1 && cplpl
&& structdef
== sinbody
))
2254 && typdef
!= tignore
2255 && definedef
!= dignorerest
2256 && funcdef
!= finlist
)
2262 if (c
== ':' && cplpl
&& *lp
== ':' && begtoken(*(lp
+ 1)))
2265 * This handles :: in the middle, but not at the
2266 * beginning of an identifier.
2273 logical is_func
= FALSE
;
2276 || consider_token (newlb
.buffer
+ tokoff
, toklen
, c
,
2277 c_ext
, cblev
, parlev
, &is_func
))
2279 if (structdef
== sinbody
2280 && definedef
== dnone
2282 /* function defined in C++ class body */
2284 GROW_LINEBUFFER (token_name
,
2285 strlen(structtag
)+2+toklen
+1);
2286 strcpy (token_name
.buffer
, structtag
);
2287 strcat (token_name
.buffer
, "::");
2288 strncat (token_name
.buffer
,
2289 newlb
.buffer
+tokoff
, toklen
);
2292 else if (objdef
== ocatseen
)
2293 /* Objective C category */
2295 GROW_LINEBUFFER (token_name
,
2296 strlen(objtag
)+2+toklen
+1);
2297 strcpy (token_name
.buffer
, objtag
);
2298 strcat (token_name
.buffer
, "(");
2299 strncat (token_name
.buffer
,
2300 newlb
.buffer
+tokoff
, toklen
);
2301 strcat (token_name
.buffer
, ")");
2304 else if (objdef
== omethodtag
2305 || objdef
== omethodparm
)
2306 /* Objective C method */
2312 GROW_LINEBUFFER (token_name
, toklen
+1);
2313 strncpy (token_name
.buffer
,
2314 newlb
.buffer
+tokoff
, toklen
);
2315 token_name
.buffer
[toklen
] = '\0';
2316 if (structdef
== stagseen
2319 && definedef
== dignorerest
)) /* macro */
2324 tok
.lineno
= lineno
;
2325 tok
.linelen
= tokoff
+ toklen
+ 1;
2326 tok
.buffer
= newlb
.buffer
;
2327 tok
.linepos
= newlinepos
;
2330 if (definedef
== dnone
2331 && (funcdef
== ftagseen
2332 || structdef
== stagseen
2334 || objdef
!= onone
))
2336 if (current_lb_is_new
)
2337 switch_line_buffers ();
2344 } /* if (endtoken (c)) */
2345 else if (intoken (c
))
2350 } /* if (midtoken) */
2351 else if (begtoken (c
))
2369 if (structdef
== stagseen
)
2375 if (!yacc_rules
|| lp
== newlb
.buffer
+ 1)
2377 tokoff
= lp
- 1 - newlb
.buffer
;
2382 } /* if (begtoken) */
2383 } /* if must look at token */
2386 /* Detect end of line, colon, comma, semicolon and various braces
2387 after having handled a token.*/
2391 if (definedef
!= dnone
)
2401 objdef
= omethodcolon
;
2403 GROW_LINEBUFFER (token_name
, methodlen
+1);
2404 strcat (token_name
.buffer
, ":");
2407 if (structdef
== stagseen
)
2408 structdef
= scolonseen
;
2425 if (definedef
!= dnone
)
2436 if (funcdef
!= fignore
)
2439 /* The following instruction invalidates the token.
2440 Probably the token should be invalidated in all
2441 other cases where some state machine is reset. */
2444 if (structdef
== stagseen
)
2448 if (definedef
!= dnone
)
2458 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2460 if (structdef
== stagseen
)
2464 if (definedef
!= dnone
)
2466 if (cblev
== 0 && typdef
== tend
)
2472 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2474 if (structdef
== stagseen
)
2478 if (definedef
!= dnone
)
2480 if (objdef
== otagseen
&& parlev
== 0)
2481 objdef
= oparenseen
;
2489 /* Make sure that the next char is not a '*'.
2490 This handles constructs like:
2491 typedef void OperatorFun (int fun); */
2498 } /* switch (typdef) */
2501 funcdef
= fstartlist
;
2510 if (definedef
!= dnone
)
2512 if (objdef
== ocatseen
&& parlev
== 1)
2523 funcdef
= flistseen
;
2526 if (cblev
== 0 && typdef
== tend
)
2532 else if (parlev
< 0) /* can happen due to ill-conceived #if's. */
2536 if (definedef
!= dnone
)
2538 if (typdef
== ttypedseen
)
2542 case skeyseen
: /* unnamed struct */
2543 structtag
= "_anonymous_";
2544 structdef
= sinbody
;
2547 case scolonseen
: /* named struct */
2548 structdef
= sinbody
;
2573 /* Neutralize `extern "C" {' grot and look inside structs. */
2574 if (cblev
== 0 && structdef
== snone
&& typdef
== tnone
)
2581 if (definedef
!= dnone
)
2583 if (funcdef
== fstartlist
)
2584 funcdef
= fnone
; /* avoid tagging `foo' in `foo (*bar()) ()' */
2587 if (definedef
!= dnone
)
2589 if (!noindentypedefs
&& lp
== newlb
.buffer
+ 1)
2591 cblev
= 0; /* reset curly brace level if first column */
2592 parlev
= 0; /* also reset paren level, just in case... */
2598 if (typdef
== tinbody
)
2600 /* Memory leakage here: the string pointed by structtag is
2601 never released, because I fear to miss something and
2602 break things while freeing the area. The amount of
2603 memory leaked here is the sum of the lengths of the
2605 if (structdef == sinbody)
2606 free (structtag); */
2609 structtag
= "<error>";
2614 if (objdef
== oinbody
&& cblev
== 0)
2616 objdef
= omethodsign
;
2620 case '=': case '#': case '~': case '&': case '%': case '/':
2621 case '|': case '^': case '!': case '<': case '>': case '.': case '?':
2622 if (definedef
!= dnone
)
2624 /* These surely cannot follow a function tag. */
2625 if (funcdef
!= finlist
&& funcdef
!= fignore
)
2629 if (objdef
== otagseen
)
2634 /* If a macro spans multiple lines don't reset its state. */
2642 } /* while not eof */
2646 * Process either a C++ file or a C file depending on the setting
2650 default_C_entries (inf
)
2653 C_entries (cplusplus
? C_PLPL
: 0, inf
);
2656 /* Always do plain ANSI C. */
2658 plain_C_entries (inf
)
2664 /* Always do C++. */
2666 Cplusplus_entries (inf
)
2669 C_entries (C_PLPL
, inf
);
2677 C_entries (C_STAR
, inf
);
2680 /* Always do Yacc. */
2685 C_entries (YACC
, inf
);
2688 /* Fortran parsing */
2696 register int len
= 0;
2698 while (*cp
&& lowcase(*cp
) == lowcase(dbp
[len
]))
2700 if (*cp
== '\0' && !intoken(dbp
[len
]))
2711 while (isspace (*dbp
))
2716 while (isspace (*dbp
))
2718 if (strneq (dbp
, "(*)", 3))
2723 if (!isdigit (*dbp
))
2725 --dbp
; /* force failure */
2730 while (isdigit (*dbp
));
2739 while (isspace (*dbp
))
2744 linecharno
= charno
;
2745 charno
+= readline (&lb
, inf
);
2750 while (isspace (*dbp
))
2759 && (isalpha (*cp
) || isdigit (*cp
) || (*cp
== '_') || (*cp
== '$')));
2762 pfnote ((CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
, TRUE
,
2763 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
2767 Fortran_functions (inf
)
2776 linecharno
= charno
;
2777 charno
+= readline (&lb
, inf
);
2780 dbp
++; /* Ratfor escape to fortran */
2781 while (isspace (*dbp
))
2785 switch (lowcase (*dbp
))
2788 if (tail ("integer"))
2796 if (tail ("logical"))
2800 if (tail ("complex") || tail ("character"))
2804 if (tail ("double"))
2806 while (isspace (*dbp
))
2810 if (tail ("precision"))
2816 while (isspace (*dbp
))
2820 switch (lowcase (*dbp
))
2823 if (tail ("function"))
2827 if (tail ("subroutine"))
2835 if (tail ("program"))
2840 if (tail ("procedure"))
2848 * Bob Weiner, Motorola Inc., 4/3/94
2849 * Unix and microcontroller assembly tag handling
2850 * look for '^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]'
2864 linecharno
= charno
;
2865 charno
+= readline (&lb
, inf
);
2868 /* If first char is alphabetic or one of [_.$], test for colon
2869 following identifier. */
2870 if (isalpha (*cp
) || *cp
== '_' || *cp
== '.' || *cp
== '$')
2872 /* Read past label. */
2874 while (isalnum (*cp
) || *cp
== '_' || *cp
== '.' || *cp
== '$')
2876 if (*cp
== ':' || isspace (*cp
))
2878 /* Found end of label, so copy it and add it to the table. */
2879 pfnote ((CTAGS
) ? savenstr(lb
.buffer
, cp
-lb
.buffer
) : NULL
, TRUE
,
2880 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
2887 * Perl support by Bart Robinson <lomew@cs.utah.edu>
2888 * Perl sub names: look for /^sub[ \t\n]+[^ \t\n{]+/
2891 Perl_functions (inf
)
2902 linecharno
= charno
;
2903 charno
+= readline (&lb
, inf
);
2906 if (*cp
++ == 's' && *cp
++ == 'u' && *cp
++ == 'b' && isspace(*cp
++))
2908 while (*cp
&& isspace(*cp
))
2910 while (*cp
&& ! isspace(*cp
) && *cp
!= '{')
2912 pfnote ((CTAGS
) ? savenstr (lb
.buffer
, cp
-lb
.buffer
) : NULL
, TRUE
,
2913 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
2918 /* Added by Mosur Mohan, 4/22/88 */
2919 /* Pascal parsing */
2922 * Locates tags for procedures & functions. Doesn't do any type- or
2923 * var-definitions. It does look for the keyword "extern" or
2924 * "forward" immediately following the procedure statement; if found,
2925 * the tag is skipped.
2928 Pascal_functions (inf
)
2931 struct linebuffer tline
; /* mostly copied from C_entries */
2933 int save_lineno
, save_len
;
2934 char c
, *cp
, *namebuf
;
2936 logical
/* each of these flags is TRUE iff: */
2937 incomment
, /* point is inside a comment */
2938 inquote
, /* point is inside '..' string */
2939 get_tagname
, /* point is after PROCEDURE/FUNCTION
2940 keyword, so next item = potential tag */
2941 found_tag
, /* point is after a potential tag */
2942 inparms
, /* point is within parameter-list */
2943 verify_tag
; /* point has passed the parm-list, so the
2944 next token will determine whether this
2945 is a FORWARD/EXTERN to be ignored, or
2946 whether it is a real tag */
2953 initbuffer (&tline
);
2955 incomment
= inquote
= FALSE
;
2956 found_tag
= FALSE
; /* have a proc name; check if extern */
2957 get_tagname
= FALSE
; /* have found "procedure" keyword */
2958 inparms
= FALSE
; /* found '(' after "proc" */
2959 verify_tag
= FALSE
; /* check if "extern" is ahead */
2961 /* long main loop to get next char */
2965 if (c
== '\0') /* if end of line */
2968 linecharno
= charno
;
2969 charno
+= readline (&lb
, inf
);
2973 if (!((found_tag
&& verify_tag
) ||
2975 c
= *dbp
++; /* only if don't need *dbp pointing
2976 to the beginning of the name of
2977 the procedure or function */
2981 if (c
== '}') /* within { } comments */
2983 else if (c
== '*' && *dbp
== ')') /* within (* *) comments */
3000 inquote
= TRUE
; /* found first quote */
3002 case '{': /* found open { comment */
3006 if (*dbp
== '*') /* found open (* comment */
3011 else if (found_tag
) /* found '(' after tag, i.e., parm-list */
3014 case ')': /* end of parms list */
3019 if (found_tag
&& !inparms
) /* end of proc or fn stmt */
3026 if (found_tag
&& verify_tag
&& (*dbp
!= ' '))
3028 /* check if this is an "extern" declaration */
3031 if (lowcase (*dbp
== 'e'))
3033 if (tail ("extern")) /* superfluous, really! */
3039 else if (lowcase (*dbp
) == 'f')
3041 if (tail ("forward")) /* check for forward reference */
3047 if (found_tag
&& verify_tag
) /* not external proc, so make tag */
3051 pfnote (namebuf
, TRUE
,
3052 tline
.buffer
, save_len
, save_lineno
, save_lcno
);
3056 if (get_tagname
) /* grab name of proc or fn */
3061 /* save all values for later tagging */
3062 GROW_LINEBUFFER (tline
, strlen (lb
.buffer
) + 1);
3063 strcpy (tline
.buffer
, lb
.buffer
);
3064 save_lineno
= lineno
;
3065 save_lcno
= linecharno
;
3067 /* grab block name */
3068 for (cp
= dbp
+ 1; *cp
&& (!endtoken (*cp
)); cp
++)
3070 namebuf
= (CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
;
3071 dbp
= cp
; /* set dbp to e-o-token */
3072 save_len
= dbp
- lb
.buffer
+ 1;
3073 get_tagname
= FALSE
;
3077 /* and proceed to check for "extern" */
3079 else if (!incomment
&& !inquote
&& !found_tag
)
3081 /* check for proc/fn keywords */
3082 switch (lowcase (c
))
3085 if (tail ("rocedure")) /* c = 'p', dbp has advanced */
3089 if (tail ("unction"))
3094 } /* while not eof */
3096 free (tline
.buffer
);
3100 * lisp tag functions
3101 * look for (def or (DEF, quote or QUOTE
3105 register char *strp
;
3107 return ((strp
[1] == 'd' || strp
[1] == 'D')
3108 && (strp
[2] == 'e' || strp
[2] == 'E')
3109 && (strp
[3] == 'f' || strp
[3] == 'F'));
3114 register char *strp
;
3116 return ((*(++strp
) == 'q' || *strp
== 'Q')
3117 && (*(++strp
) == 'u' || *strp
== 'U')
3118 && (*(++strp
) == 'o' || *strp
== 'O')
3119 && (*(++strp
) == 't' || *strp
== 'T')
3120 && (*(++strp
) == 'e' || *strp
== 'E')
3121 && isspace(*(++strp
)));
3129 if (*dbp
== '\'') /* Skip prefix quote */
3131 else if (*dbp
== '(' && L_isquote (dbp
)) /* Skip "(quote " */
3134 while (isspace(*dbp
))
3137 for (cp
= dbp
/*+1*/;
3138 *cp
&& *cp
!= '(' && *cp
!= ' ' && *cp
!= ')';
3144 pfnote ((CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
, TRUE
,
3145 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
3149 Lisp_functions (inf
)
3158 linecharno
= charno
;
3159 charno
+= readline (&lb
, inf
);
3165 while (!isspace (*dbp
))
3167 while (isspace (*dbp
))
3173 /* Check for (foo::defmumble name-defined ... */
3176 while (*dbp
&& !isspace (*dbp
)
3177 && *dbp
!= ':' && *dbp
!= '(' && *dbp
!= ')');
3182 while (*dbp
== ':');
3184 if (L_isdef (dbp
- 1))
3186 while (!isspace (*dbp
))
3188 while (isspace (*dbp
))
3199 * Scheme tag functions
3200 * look for (def... xyzzy
3201 * look for (def... (xyzzy
3202 * look for (def ... ((...(xyzzy ....
3203 * look for (set! xyzzy
3209 Scheme_functions (inf
)
3218 linecharno
= charno
;
3219 charno
+= readline (&lb
, inf
);
3221 if (dbp
[0] == '(' &&
3222 (dbp
[1] == 'D' || dbp
[1] == 'd') &&
3223 (dbp
[2] == 'E' || dbp
[2] == 'e') &&
3224 (dbp
[3] == 'F' || dbp
[3] == 'f'))
3226 while (!isspace (*dbp
))
3228 /* Skip over open parens and white space */
3229 while (*dbp
&& (isspace (*dbp
) || *dbp
== '('))
3233 if (dbp
[0] == '(' &&
3234 (dbp
[1] == 'S' || dbp
[1] == 's') &&
3235 (dbp
[2] == 'E' || dbp
[2] == 'e') &&
3236 (dbp
[3] == 'T' || dbp
[3] == 't') &&
3237 (dbp
[4] == '!' || dbp
[4] == '!') &&
3240 while (!isspace (*dbp
))
3242 /* Skip over white space */
3243 while (isspace (*dbp
))
3257 /* Go till you get to white space or a syntactic break */
3259 *cp
&& *cp
!= '(' && *cp
!= ')' && !isspace (*cp
);
3262 pfnote ((CTAGS
) ? savenstr (dbp
, cp
-dbp
) : NULL
, TRUE
,
3263 lb
.buffer
, cp
- lb
.buffer
+ 1, lineno
, linecharno
);
3266 /* Find tags in TeX and LaTeX input files. */
3268 /* TEX_toktab is a table of TeX control sequences that define tags.
3269 Each TEX_tabent records one such control sequence.
3270 CONVERT THIS TO USE THE Stab TYPE!! */
3277 struct TEX_tabent
*TEX_toktab
= NULL
; /* Table with tag tokens */
3279 /* Default set of control sequences to put into TEX_toktab.
3280 The value of environment var TEXTAGS is prepended to this. */
3282 char *TEX_defenv
= "\
3283 :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem\
3284 :part:appendix:entry:index";
3287 struct TEX_tabent
*TEX_decode_env ();
3289 #if TeX_named_tokens
3293 char TEX_esc
= '\\';
3294 char TEX_opgrp
= '{';
3295 char TEX_clgrp
= '}';
3298 * TeX/LaTeX scanning loop.
3309 /* Select either \ or ! as escape character. */
3312 /* Initialize token table once from environment. */
3314 TEX_toktab
= TEX_decode_env ("TEXTAGS", TEX_defenv
);
3317 { /* Scan each line in file */
3319 linecharno
= charno
;
3320 charno
+= readline (&lb
, inf
);
3323 while (dbp
= etags_strchr (dbp
, TEX_esc
)) /* Look at each esc in line */
3329 linecharno
+= dbp
- lasthit
;
3331 i
= TEX_Token (lasthit
);
3335 lb
.buffer
, strlen (lb
.buffer
), lineno
, linecharno
);
3336 #if TeX_named_tokens
3337 TEX_getit (lasthit
, TEX_toktab
[i
].len
);
3339 break; /* We only save a line once */
3345 #define TEX_LESC '\\'
3346 #define TEX_SESC '!'
3349 /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
3350 chars accordingly. */
3357 while ((c
= getc (inf
)) != EOF
)
3359 /* Skip to next line if we hit the TeX comment char. */
3363 else if (c
== TEX_LESC
|| c
== TEX_SESC
)
3382 /* Read environment and prepend it to the default string.
3383 Build token table. */
3385 TEX_decode_env (evarname
, defenv
)
3389 register char *env
, *p
;
3391 struct TEX_tabent
*tab
;
3394 /* Append default string to environment. */
3395 env
= getenv (evarname
);
3399 env
= concat (env
, defenv
, "");
3401 /* Allocate a token table */
3402 for (size
= 1, p
= env
; p
;)
3403 if ((p
= etags_strchr (p
, ':')) && *(++p
))
3405 /* Add 1 to leave room for null terminator. */
3406 tab
= xnew (size
+ 1, struct TEX_tabent
);
3408 /* Unpack environment string into token table. Be careful about */
3409 /* zero-length strings (leading ':', "::" and trailing ':') */
3412 p
= etags_strchr (env
, ':');
3413 if (!p
) /* End of environment string. */
3414 p
= env
+ strlen (env
);
3416 { /* Only non-zero strings. */
3417 tab
[i
].name
= savenstr (env
, p
- env
);
3418 tab
[i
].len
= strlen (tab
[i
].name
);
3425 tab
[i
].name
= NULL
; /* Mark end of table. */
3433 #if TeX_named_tokens
3434 /* Record a tag defined by a TeX command of length LEN and starting at NAME.
3435 The name being defined actually starts at (NAME + LEN + 1).
3436 But we seem to include the TeX command in the tag name. */
3438 TEX_getit (name
, len
)
3442 char *p
= name
+ len
;
3447 /* Let tag name extend to next group close (or end of line) */
3448 while (*p
&& *p
!= TEX_clgrp
)
3450 pfnote (savenstr (name
, p
-name
), TRUE
,
3451 lb
.buffer
, strlen (lb
.buffer
), lineno
, linecharno
);
3455 /* If the text at CP matches one of the tag-defining TeX command names,
3456 return the pointer to the first occurrence of that command in TEX_toktab.
3457 Otherwise return -1.
3458 Keep the capital `T' in `Token' for dumb truncating compilers
3459 (this distinguishes it from `TEX_toktab' */
3466 for (i
= 0; TEX_toktab
[i
].len
> 0; i
++)
3467 if (strneq (TEX_toktab
[i
].name
, cp
, TEX_toktab
[i
].len
))
3473 * Prolog support (rewritten) by Anders Lindgren, Mar. 96
3475 * Assumes that the predicate starts at column 0.
3476 * Only the first clause of a predicate is added.
3479 Prolog_functions (inf
)
3483 void prolog_skip_comment ();
3500 linecharno
+= charno
;
3501 charno
= readline (&lb
, inf
);
3503 if (dbp
[0] == '\0') /* Empty line */
3505 else if (isspace (dbp
[0])) /* Not a predicate */
3507 else if (dbp
[0] == '/' && dbp
[1] == '*') /* comment. */
3508 prolog_skip_comment (&lb
, inf
, &lineno
, &linecharno
);
3509 else if (len
= prolog_pred (dbp
, last
))
3511 /* Predicate. Store the function name so that we only
3512 * generates a tag for the first clause. */
3514 last
= xnew(len
+ 1, char);
3515 else if (len
+ 1 > allocated
)
3516 last
= (char *) xrealloc(last
, len
+ 1);
3517 allocated
= len
+ 1;
3518 strncpy (last
, dbp
, len
);
3526 prolog_skip_comment (plb
, inf
)
3527 struct linebuffer
*plb
;
3534 for (cp
= plb
->buffer
; *cp
!= '\0'; cp
++)
3535 if (cp
[0] == '*' && cp
[1] == '/')
3538 linecharno
+= readline (plb
, inf
);
3544 * A predicate definition is added if it matches:
3545 * <beginning of line><Prolog Atom><whitespace>(
3547 * It is added to the tags database if it doesn't match the
3548 * name of the previous clause header.
3550 * Return the size of the name of the predicate, or 0 if no header
3554 prolog_pred (s
, last
)
3556 char *last
; /* Name of last clause. */
3564 pos
= prolog_atom(s
, 0);
3569 pos
+= prolog_white(s
, pos
);
3571 if ((s
[pos
] == '(') || (s
[pos
] == '.'))
3576 /* Save only the first clause. */
3577 if ((last
== NULL
) ||
3578 (len
!= strlen(last
)) ||
3579 (strncmp(s
, last
, len
) != 0))
3581 pfnote ((CTAGS
) ? savenstr (s
, len
) : NULL
, TRUE
,
3582 s
, pos
, lineno
, linecharno
);
3590 * Consume a Prolog atom.
3591 * Return the number of bytes consumed, or -1 if there was an error.
3593 * A prolog atom, in this context, could be one of:
3594 * - An alphanumeric sequence, starting with a lower case letter.
3595 * - A quoted arbitrary string. Single quotes can escape themselves.
3596 * Backslash quotes everything.
3599 prolog_atom (s
, pos
)
3607 if (islower(s
[pos
]) || (s
[pos
] == '_'))
3609 /* The atom is unquoted. */
3611 while (isalnum(s
[pos
]) || (s
[pos
] == '_'))
3615 return pos
- origpos
;
3617 else if (s
[pos
] == '\'')
3628 pos
++; /* A double quote */
3630 else if (s
[pos
] == '\0')
3631 /* Multiline quoted atoms are ignored. */
3633 else if (s
[pos
] == '\\')
3635 if (s
[pos
+1] == '\0')
3642 return pos
- origpos
;
3648 /* Consume whitespace. Return the number of bytes eaten. */
3650 prolog_white (s
, pos
)
3658 while (isspace(s
[pos
]))
3661 return pos
- origpos
;
3665 * Support for Erlang -- Anders Lindgren, Feb 1996.
3667 * Generates tags for functions, defines, and records.
3669 * Assumes that Erlang functions start at column 0.
3672 Erlang_functions (inf
)
3676 void erlang_attribute ();
3693 linecharno
+= charno
;
3694 charno
= readline (&lb
, inf
);
3696 if (dbp
[0] == '\0') /* Empty line */
3698 else if (isspace (dbp
[0])) /* Not function nor attribute */
3700 else if (dbp
[0] == '%') /* comment */
3702 else if (dbp
[0] == '"') /* Sometimes, strings start in column one */
3704 else if (dbp
[0] == '-') /* attribute, e.g. "-define" */
3706 erlang_attribute(dbp
);
3709 else if (len
= erlang_func (dbp
, last
))
3712 * Function. Store the function name so that we only
3713 * generates a tag for the first clause.
3716 last
= xnew(len
+ 1, char);
3717 else if (len
+ 1 > allocated
)
3718 last
= (char *) xrealloc(last
, len
+ 1);
3719 allocated
= len
+ 1;
3720 strncpy (last
, dbp
, len
);
3728 * A function definition is added if it matches:
3729 * <beginning of line><Erlang Atom><whitespace>(
3731 * It is added to the tags database if it doesn't match the
3732 * name of the previous clause header.
3734 * Return the size of the name of the function, or 0 if no function
3738 erlang_func (s
, last
)
3740 char *last
; /* Name of last clause. */
3743 int erlang_white ();
3748 pos
= erlang_atom(s
, 0);
3753 pos
+= erlang_white(s
, pos
);
3755 if (s
[pos
++] == '(')
3757 /* Save only the first clause. */
3758 if ((last
== NULL
) ||
3759 (len
!= strlen(last
)) ||
3760 (strncmp(s
, last
, len
) != 0))
3762 pfnote ((CTAGS
) ? savenstr (s
, len
) : NULL
, TRUE
,
3763 s
, pos
, lineno
, linecharno
);
3772 * Handle attributes. Currently, tags are generated for defines
3775 * They are on the form:
3776 * -define(foo, bar).
3777 * -define(Foo(M, N), M+N).
3778 * -record(graph, {vtab = notable, cyclic = true}).
3781 erlang_attribute (s
)
3785 int erlang_white ();
3790 if ((strncmp(s
, "-define", 7) == 0) ||
3791 (strncmp(s
, "-record", 7) == 0))
3794 pos
+= erlang_white(s
, pos
);
3796 if (s
[pos
++] == '(')
3798 pos
+= erlang_white(s
, pos
);
3800 if (len
= erlang_atom(s
, pos
))
3802 pfnote ((CTAGS
) ? savenstr (& s
[pos
], len
) : NULL
, TRUE
,
3803 s
, pos
+ len
, lineno
, linecharno
);
3812 * Consume an Erlang atom (or variable).
3813 * Return the number of bytes consumed, or -1 if there was an error.
3816 erlang_atom (s
, pos
)
3824 if (isalpha (s
[pos
]) || s
[pos
] == '_')
3826 /* The atom is unquoted. */
3828 while (isalnum (s
[pos
]) || s
[pos
] == '_')
3830 return pos
- origpos
;
3832 else if (s
[pos
] == '\'')
3843 else if (s
[pos
] == '\0')
3844 /* Multiline quoted atoms are ignored. */
3846 else if (s
[pos
] == '\\')
3848 if (s
[pos
+1] == '\0')
3855 return pos
- origpos
;
3861 /* Consume whitespace. Return the number of bytes eaten */
3863 erlang_white (s
, pos
)
3871 while (isspace (s
[pos
]))
3874 return pos
- origpos
;
3877 #ifdef ETAGS_REGEXPS
3878 /* Take a string like "/blah/" and turn it into "blah", making sure
3879 that the first and last characters are the same, and handling
3880 quoted separator characters. Actually, stops on the occurrence of
3881 an unquoted separator. Also turns "\t" into a Tab character.
3882 Returns pointer to terminating separator. Works in place. Null
3883 terminates name string. */
3885 scan_separators (name
)
3889 char *copyto
= name
;
3890 logical quoted
= FALSE
;
3892 for (++name
; *name
!= '\0'; ++name
)
3898 else if (*name
== sep
)
3902 /* Something else is quoted, so preserve the quote. */
3908 else if (*name
== '\\')
3910 else if (*name
== sep
)
3916 /* Terminate copied string. */
3921 /* Turn a name, which is an ed-style (but Emacs syntax) regular
3922 expression, into a real regular expression by compiling it. */
3924 add_regex (regexp_pattern
)
3925 char *regexp_pattern
;
3929 struct re_pattern_buffer
*patbuf
;
3931 if (regexp_pattern
== NULL
)
3933 /* Remove existing regexps. */
3939 if (regexp_pattern
[0] == '\0')
3941 error ("missing regexp", 0);
3944 if (regexp_pattern
[strlen(regexp_pattern
)-1] != regexp_pattern
[0])
3946 error ("%s: unterminated regexp", regexp_pattern
);
3949 name
= scan_separators (regexp_pattern
);
3950 if (regexp_pattern
[0] == '\0')
3952 error ("null regexp", 0);
3955 (void) scan_separators (name
);
3957 patbuf
= xnew (1, struct re_pattern_buffer
);
3958 patbuf
->translate
= NULL
;
3959 patbuf
->fastmap
= NULL
;
3960 patbuf
->buffer
= NULL
;
3961 patbuf
->allocated
= 0;
3963 err
= re_compile_pattern (regexp_pattern
, strlen (regexp_pattern
), patbuf
);
3966 error ("%s while compiling pattern", err
);
3971 if (num_patterns
== 1)
3972 patterns
= xnew (1, struct pattern
);
3974 patterns
= ((struct pattern
*)
3976 (num_patterns
* sizeof (struct pattern
))));
3977 patterns
[num_patterns
- 1].pattern
= patbuf
;
3978 patterns
[num_patterns
- 1].name_pattern
= savestr (name
);
3979 patterns
[num_patterns
- 1].error_signaled
= FALSE
;
3983 * Do the substitutions indicated by the regular expression and
3987 substitute (in
, out
, regs
)
3989 struct re_registers
*regs
;
3991 char *result
= NULL
, *t
;
3994 /* Pass 1: figure out how much size to allocate. */
3995 for (t
= out
; *t
; ++t
)
4002 fprintf (stderr
, "%s: pattern substitution ends prematurely\n",
4009 size
+= regs
->end
[dig
] - regs
->start
[dig
];
4014 /* Allocate space and do the substitutions. */
4015 result
= xnew (size
+ 1, char);
4024 /* Using "dig2" satisfies my debugger. Bleah. */
4025 int dig2
= *out
- '0';
4026 strncpy (result
+ size
, in
+ regs
->start
[dig2
],
4027 regs
->end
[dig2
] - regs
->start
[dig2
]);
4028 size
+= regs
->end
[dig2
] - regs
->start
[dig2
];
4031 result
[size
++] = *out
;
4034 result
[size
++] = *out
;
4036 result
[size
] = '\0';
4041 #endif /* ETAGS_REGEXPS */
4042 /* Initialize a linebuffer for use */
4044 initbuffer (linebuffer
)
4045 struct linebuffer
*linebuffer
;
4047 linebuffer
->size
= 200;
4048 linebuffer
->buffer
= xnew (200, char);
4052 * Read a line of text from `stream' into `linebuffer'.
4053 * Return the number of characters read from `stream',
4054 * which is the length of the line including the newline, if any.
4057 readline_internal (linebuffer
, stream
)
4058 struct linebuffer
*linebuffer
;
4059 register FILE *stream
;
4061 char *buffer
= linebuffer
->buffer
;
4062 register char *p
= linebuffer
->buffer
;
4063 register char *pend
;
4066 pend
= p
+ linebuffer
->size
; /* Separate to avoid 386/IX compiler bug. */
4070 register int c
= getc (stream
);
4073 linebuffer
->size
*= 2;
4074 buffer
= (char *) xrealloc (buffer
, linebuffer
->size
);
4075 p
+= buffer
- linebuffer
->buffer
;
4076 pend
= buffer
+ linebuffer
->size
;
4077 linebuffer
->buffer
= buffer
;
4087 if (p
> buffer
&& p
[-1] == '\r')
4102 return p
- buffer
+ chars_deleted
;
4106 * Like readline_internal, above, but try to match the input
4107 * line against any existing regular expressions.
4110 readline (linebuffer
, stream
)
4111 struct linebuffer
*linebuffer
;
4114 /* Read new line. */
4115 long result
= readline_internal (linebuffer
, stream
);
4116 #ifdef ETAGS_REGEXPS
4119 /* Match against all listed patterns. */
4120 for (i
= 0; i
< num_patterns
; ++i
)
4122 int match
= re_match (patterns
[i
].pattern
, linebuffer
->buffer
,
4123 (int)result
, 0, &patterns
[i
].regs
);
4128 if (!patterns
[i
].error_signaled
)
4130 error ("error while matching pattern %d", i
);
4131 patterns
[i
].error_signaled
= TRUE
;
4138 /* Match occurred. Construct a tag. */
4139 if (patterns
[i
].name_pattern
[0] != '\0')
4141 /* Make a named tag. */
4142 char *name
= substitute (linebuffer
->buffer
,
4143 patterns
[i
].name_pattern
,
4147 linebuffer
->buffer
, match
, lineno
, linecharno
);
4151 /* Make an unnamed tag. */
4153 linebuffer
->buffer
, match
, lineno
, linecharno
);
4158 #endif /* ETAGS_REGEXPS */
4164 * Read a file, but do no processing. This is used to do regexp
4165 * matching on files that have no language defined.
4168 just_read_file (inf
)
4177 linecharno
= charno
;
4178 charno
+= readline (&lb
, inf
) + 1;
4184 * Return a pointer to a space of size strlen(cp)+1 allocated
4185 * with xnew where the string CP has been copied.
4191 return savenstr (cp
, strlen (cp
));
4195 * Return a pointer to a space of size LEN+1 allocated with xnew where
4196 * the string CP has been copied for at most the first LEN characters.
4205 dp
= xnew (len
+ 1, char);
4206 strncpy (dp
, cp
, len
);
4212 * Return the ptr in sp at which the character c last
4213 * appears; NULL if not found
4215 * Identical to System V strrchr, included for portability.
4218 etags_strrchr (sp
, c
)
4219 register char *sp
, c
;
4234 * Return the ptr in sp at which the character c first
4235 * appears; NULL if not found
4237 * Identical to System V strchr, included for portability.
4240 etags_strchr (sp
, c
)
4241 register char *sp
, c
;
4251 /* Print error message and exit. */
4269 suggest_asking_for_help ()
4271 fprintf (stderr
, "\tTry `%s --help' for a complete list of options.\n",
4276 /* Print error message. `s1' is printf control string, `s2' is arg for it. */
4281 fprintf (stderr
, "%s: ", progname
);
4282 fprintf (stderr
, s1
, s2
);
4283 fprintf (stderr
, "\n");
4286 /* Return a newly-allocated string whose contents
4287 concatenate those of s1, s2, s3. */
4292 int len1
= strlen (s1
), len2
= strlen (s2
), len3
= strlen (s3
);
4293 char *result
= xnew (len1
+ len2
+ len3
+ 1, char);
4295 strcpy (result
, s1
);
4296 strcpy (result
+ len1
, s2
);
4297 strcpy (result
+ len1
+ len2
, s3
);
4298 result
[len1
+ len2
+ len3
] = '\0';
4303 /* Does the same work as the system V getcwd, but does not need to
4304 guess the buffer size in advance. */
4309 char *p
, path
[MAXPATHLEN
+ 1]; /* Fixed size is safe on MSDOS. */
4312 for (p
= path
; *p
!= '\0'; p
++)
4318 return strdup (path
);
4319 #else /* not MSDOS */
4322 char *path
= xnew (bufsize
, char);
4324 while (getcwd (path
, bufsize
) == NULL
)
4326 if (errno
!= ERANGE
)
4329 path
= xnew (bufsize
, char);
4333 #else /* not MSDOS and not HAVE_GETCWD */
4334 struct linebuffer path
;
4338 pipe
= (FILE *) popen ("pwd 2>/dev/null", "r");
4339 if (pipe
== NULL
|| readline_internal (&path
, pipe
) == 0)
4344 #endif /* not HAVE_GETCWD */
4345 #endif /* not MSDOS */
4348 /* Return a newly allocated string containing the filename
4349 of FILE relative to the absolute directory DIR (which
4350 should end with a slash). */
4352 relative_filename (file
, dir
)
4355 char *fp
, *dp
, *abs
, *res
;
4357 /* Find the common root of file and dir. */
4358 abs
= absolute_filename (file
, cwd
);
4361 while (*fp
++ == *dp
++)
4370 /* Build a sequence of "../" strings for the resulting relative filename. */
4371 for (dp
= etags_strchr (dp
+ 1, '/'), res
= "";
4373 dp
= etags_strchr (dp
+ 1, '/'))
4375 res
= concat (res
, "../", "");
4378 /* Add the filename relative to the common root of file and dir. */
4379 res
= concat (res
, fp
+ 1, "");
4385 /* Return a newly allocated string containing the
4386 absolute filename of FILE given CWD (which should
4387 end with a slash). */
4389 absolute_filename (file
, cwd
)
4392 char *slashp
, *cp
, *res
;
4394 if (absolutefn (file
))
4395 res
= concat (file
, "", "");
4397 /* We don't support non-absolute filenames with a drive
4398 letter, like `d:NAME' (it's too much hassle). */
4399 else if (file
[1] == ':')
4400 fatal ("%s: relative filenames with drive letters not supported", file
);
4403 res
= concat (cwd
, file
, "");
4405 /* Delete the "/dirname/.." and "/." substrings. */
4406 slashp
= etags_strchr (res
, '/');
4407 while (slashp
!= NULL
&& slashp
[0] != '\0')
4409 if (slashp
[1] == '.')
4411 if (slashp
[2] == '.'
4412 && (slashp
[3] == '/' || slashp
[3] == '\0'))
4417 while (cp
>= res
&& !absolutefn (cp
));
4420 strcpy (cp
, slashp
+ 3);
4423 /* Under MSDOS and NT we get `d:/NAME' as absolute
4424 filename, so the luser could say `d:/../NAME'.
4425 We silently treat this as `d:/NAME'. */
4426 else if (cp
[1] == ':')
4427 strcpy (cp
+ 3, slashp
+ 4);
4429 else /* else (cp == res) */
4431 if (slashp
[3] != '\0')
4432 strcpy (cp
, slashp
+ 4);
4439 else if (slashp
[2] == '/' || slashp
[2] == '\0')
4441 strcpy (slashp
, slashp
+ 2);
4446 slashp
= etags_strchr (slashp
+ 1, '/');
4452 /* Return a newly allocated string containing the absolute
4453 filename of dir where FILE resides given CWD (which should
4454 end with a slash). */
4456 absolute_dirname (file
, cwd
)
4464 for (p
= file
; *p
!= '\0'; p
++)
4469 slashp
= etags_strrchr (file
, '/');
4474 res
= absolute_filename (file
, cwd
);
4480 /* Like malloc but get fatal error if memory is exhausted. */
4485 long *result
= (long *) malloc (size
);
4487 fatal ("virtual memory exhausted", 0);
4492 xrealloc (ptr
, size
)
4496 long *result
= (long *) realloc (ptr
, size
);
4498 fatal ("virtual memory exhausted");