Commit | Line | Data |
---|---|---|
c6d46f5f | 1 | /* Tags file maker to go with GNU Emacs |
3a22ee35 | 2 | Copyright (C) 1984,87,88,89,93,94 Free Software Foundation, Inc. and Ken Arnold |
ea6cd314 | 3 | This file is not considered part of GNU Emacs. |
c6d46f5f | 4 | |
ea6cd314 | 5 | This program is free software; you can redistribute it and/or modify |
c6d46f5f | 6 | it under the terms of the GNU General Public License as published by |
ea6cd314 RS |
7 | the Free Software Foundation; either version 2 of the License, or |
8 | (at your option) any later version. | |
c6d46f5f | 9 | |
ea6cd314 | 10 | This program is distributed in the hope that it will be useful, |
c6d46f5f JB |
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
13 | GNU General Public License for more details. | |
14 | ||
15 | You should have received a copy of the GNU General Public License | |
ea6cd314 RS |
16 | along with this program; if not, write to the Free Software |
17 | Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ | |
c6d46f5f JB |
18 | |
19 | /* | |
20 | * Authors: | |
21 | * Ctags originally by Ken Arnold. | |
6dd5561c | 22 | * Fortran added by Jim Kleckner. |
c6d46f5f JB |
23 | * Ed Pelegri-Llopart added C typedefs. |
24 | * Gnu Emacs TAGS format and modifications by RMS? | |
25 | * Sam Kendall added C++. | |
b9755a12 FP |
26 | * Francesco Potorti` reorganised C and C++ based on work by Joe Wells. |
27 | #ifdef ETAGS_REGEXPS | |
28 | * Regexp tags by Tom Tromey. | |
29 | #endif | |
31d4b314 | 30 | * |
46c145db | 31 | * Francesco Potorti` (pot@cnuce.cnr.it) is the current maintainer. |
c6d46f5f | 32 | */ |
2bd88040 | 33 | char pot_etags_version[] = "@(#) pot revision number is 11.21"; |
c6d46f5f | 34 | |
46c145db | 35 | |
c6880c90 RS |
36 | #ifdef MSDOS |
37 | #include <fcntl.h> | |
88f125fc | 38 | #include <sys/param.h> |
c6880c90 RS |
39 | #endif /* MSDOS */ |
40 | ||
c05b6df5 RS |
41 | #ifdef WINDOWSNT |
42 | #include <stdlib.h> | |
43 | #include <fcntl.h> | |
44 | #include <string.h> | |
45 | #define MAXPATHLEN _MAX_PATH | |
46 | #endif | |
47 | ||
72a339d7 | 48 | #ifdef HAVE_CONFIG_H |
b9755a12 | 49 | #include <config.h> |
aab1fdae FP |
50 | /* On some systems, Emacs defines static as nothing for the sake |
51 | of unexec. We don't want that here since we don't use unexec. */ | |
1ddff51c | 52 | #undef static |
1e134a5f RM |
53 | #endif |
54 | ||
918f9ad1 JB |
55 | #include <stdio.h> |
56 | #include <ctype.h> | |
dcc89e63 FP |
57 | #include <errno.h> |
58 | #ifndef errno | |
59 | extern int errno; | |
60 | #endif | |
918f9ad1 JB |
61 | #include <sys/types.h> |
62 | #include <sys/stat.h> | |
63 | ||
2b878b4c FP |
64 | #if !defined (S_ISREG) && defined (S_IFREG) |
65 | # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG) | |
66 | #endif | |
67 | ||
b9755a12 FP |
68 | #include <getopt.h> |
69 | ||
70 | #ifdef ETAGS_REGEXPS | |
71 | #include <regex.h> | |
72 | #endif /* ETAGS_REGEXPS */ | |
918f9ad1 | 73 | |
55597f90 FP |
74 | #define TRUE 1 |
75 | #define FALSE 0 | |
c6d46f5f | 76 | |
32daa216 FP |
77 | /* Define CTAGS to make the program "ctags" compatible with the usual one. |
78 | Let it undefined to make the program "etags", which makes emacs-style | |
79 | tag tables and tags typedefs, #defines and struct/union/enum by default. */ | |
80 | #ifdef CTAGS | |
81 | # undef CTAGS | |
82 | # define CTAGS TRUE | |
83 | #else | |
84 | # define CTAGS FALSE | |
c6d46f5f JB |
85 | #endif |
86 | ||
87 | /* Exit codes for success and failure. */ | |
88 | #ifdef VMS | |
32daa216 FP |
89 | #define GOOD 1 |
90 | #define BAD 0 | |
c6d46f5f | 91 | #else |
32daa216 FP |
92 | #define GOOD 0 |
93 | #define BAD 1 | |
c6d46f5f JB |
94 | #endif |
95 | ||
55597f90 FP |
96 | /* C extensions. */ |
97 | #define C_PLPL 0x00001 /* C++ */ | |
98 | #define C_STAR 0x00003 /* C* */ | |
99 | #define YACC 0x10000 /* yacc file */ | |
c6d46f5f | 100 | |
aab1fdae FP |
101 | #define streq(s,t) (strcmp (s, t) == 0) |
102 | #define strneq(s,t,n) (strncmp (s, t, n) == 0) | |
c6d46f5f | 103 | |
c6d46f5f JB |
104 | #define iswhite(arg) (_wht[arg]) /* T if char is white */ |
105 | #define begtoken(arg) (_btk[arg]) /* T if char can start token */ | |
106 | #define intoken(arg) (_itk[arg]) /* T if char can be in token */ | |
107 | #define endtoken(arg) (_etk[arg]) /* T if char ends tokens */ | |
c6d46f5f | 108 | |
55597f90 FP |
109 | /* |
110 | * xnew -- allocate storage | |
111 | * | |
112 | * SYNOPSIS: Type *xnew (int n, Type); | |
113 | */ | |
114 | #define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type))) | |
c6d46f5f | 115 | |
aab1fdae FP |
116 | typedef int logical; |
117 | ||
55597f90 | 118 | typedef struct nd_st |
c6d46f5f JB |
119 | { /* sorting structure */ |
120 | char *name; /* function or type name */ | |
121 | char *file; /* file name */ | |
122 | logical is_func; /* use pattern or line no */ | |
fe0b3356 | 123 | logical named; /* list name separately */ |
c6d46f5f JB |
124 | logical been_warned; /* set if noticed dup */ |
125 | int lno; /* line number tag is on */ | |
126 | long cno; /* character number line starts on */ | |
127 | char *pat; /* search pattern */ | |
128 | struct nd_st *left, *right; /* left and right sons */ | |
55597f90 | 129 | } NODE; |
c6d46f5f | 130 | |
55597f90 | 131 | extern char *getenv (); |
c6d46f5f JB |
132 | |
133 | char *concat (); | |
46c145db | 134 | char *savenstr (), *savestr (); |
b02c5fea FP |
135 | char *etags_strchr (), *etags_strrchr (); |
136 | char *etags_getcwd (); | |
46c145db FP |
137 | char *relative_filename (), *absolute_filename (), *absolute_dirname (); |
138 | char *xmalloc (), *xrealloc (); | |
b9755a12 FP |
139 | |
140 | typedef void Lang_function (); | |
cdc1f6a7 | 141 | #if FALSE /* many compilers barf on this */ |
b9755a12 FP |
142 | Lang_function Asm_labels; |
143 | Lang_function default_C_entries; | |
144 | Lang_function C_entries; | |
145 | Lang_function Cplusplus_entries; | |
146 | Lang_function Cstar_entries; | |
147 | Lang_function Fortran_functions; | |
148 | Lang_function Yacc_entries; | |
149 | Lang_function Lisp_functions; | |
150 | Lang_function Pascal_functions; | |
151 | Lang_function Prolog_functions; | |
152 | Lang_function Scheme_functions; | |
153 | Lang_function TeX_functions; | |
154 | Lang_function just_read_file; | |
aab1fdae FP |
155 | #else /* so let's write it this way */ |
156 | void Asm_labels (); | |
157 | void default_C_entries (); | |
158 | void C_entries (); | |
159 | void Cplusplus_entries (); | |
160 | void Cstar_entries (); | |
161 | void Fortran_functions (); | |
162 | void Yacc_entries (); | |
163 | void Lisp_functions (); | |
164 | void Pascal_functions (); | |
165 | void Prolog_functions (); | |
166 | void Scheme_functions (); | |
167 | void TeX_functions (); | |
168 | void just_read_file (); | |
169 | #endif | |
b9755a12 FP |
170 | |
171 | logical get_language (); | |
c6d46f5f | 172 | int total_size_of_entries (); |
c6d46f5f | 173 | long readline (); |
b9755a12 FP |
174 | long readline_internal (); |
175 | #ifdef ETAGS_REGEXPS | |
176 | void add_regex (); | |
177 | #endif | |
c6d46f5f JB |
178 | void add_node (); |
179 | void error (); | |
cdc1f6a7 | 180 | void fatal (), pfatal (); |
55597f90 | 181 | void find_entries (); |
c6d46f5f JB |
182 | void free_tree (); |
183 | void getit (); | |
c6d46f5f JB |
184 | void init (); |
185 | void initbuffer (); | |
c6d46f5f JB |
186 | void pfnote (); |
187 | void process_file (); | |
188 | void put_entries (); | |
189 | void takeprec (); | |
190 | ||
c6d46f5f | 191 | \f |
55597f90 | 192 | char searchar = '/'; /* use /.../ searches */ |
c6d46f5f | 193 | |
55597f90 FP |
194 | int lineno; /* line number of current line */ |
195 | long charno; /* current character number */ | |
c6d46f5f | 196 | |
55597f90 FP |
197 | long linecharno; /* charno of start of line; not used by C, |
198 | but by every other language. */ | |
c6d46f5f | 199 | |
55597f90 FP |
200 | char *curfile; /* current input file name */ |
201 | char *tagfile; /* output file */ | |
202 | char *progname; /* name this program was invoked with */ | |
203 | char *cwd; /* current working directory */ | |
204 | char *tagfiledir; /* directory of tagfile */ | |
c6d46f5f | 205 | |
55597f90 FP |
206 | FILE *tagf; /* ioptr for tags file */ |
207 | NODE *head; /* the head of the binary tree of tags */ | |
c6d46f5f | 208 | |
55597f90 FP |
209 | /* |
210 | * A `struct linebuffer' is a structure which holds a line of text. | |
211 | * `readline' reads a line from a stream into a linebuffer and works | |
212 | * regardless of the length of the line. | |
213 | */ | |
214 | struct linebuffer | |
215 | { | |
216 | long size; | |
217 | char *buffer; | |
218 | }; | |
c6d46f5f | 219 | |
55597f90 | 220 | struct linebuffer lb; /* the current line */ |
2bd88040 | 221 | struct linebuffer token_str; /* used by C_entries as temporary area */ |
55597f90 FP |
222 | struct |
223 | { | |
224 | long linepos; | |
225 | struct linebuffer lb; /* used by C_entries instead of lb */ | |
226 | } lbs[2]; | |
c6d46f5f | 227 | |
55597f90 FP |
228 | /* boolean "functions" (see init) */ |
229 | logical _wht[0177], _etk[0177], _itk[0177], _btk[0177]; | |
230 | char | |
ee70dba5 FP |
231 | *white = " \f\t\n\013", /* white chars */ |
232 | *endtk = " \t\n\013\"'#()[]{}=-+%*/&|^~!<>;,.:?", /* token ending chars */ | |
13fde0cd | 233 | /* token starting chars */ |
b12756c8 | 234 | *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~", |
46c145db | 235 | /* valid in-token chars */ |
13fde0cd | 236 | *intk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789"; |
c6d46f5f | 237 | |
55597f90 FP |
238 | logical append_to_tagfile; /* -a: append to tags */ |
239 | /* The following three default to TRUE for etags, but to FALSE for ctags. */ | |
240 | logical typedefs; /* -t: create tags for typedefs */ | |
241 | logical typedefs_and_cplusplus; /* -T: create tags for typedefs, level */ | |
c6d46f5f | 242 | /* 0 struct/enum/union decls, and C++ */ |
32daa216 | 243 | /* member functions. */ |
55597f90 | 244 | logical constantypedefs; /* -d: create tags for C #define and enum */ |
32daa216 | 245 | /* constants. Enum consts not implemented. */ |
c6d46f5f | 246 | /* -D: opposite of -d. Default under ctags. */ |
55597f90 FP |
247 | logical update; /* -u: update tags */ |
248 | logical vgrind_style; /* -v: create vgrind style index output */ | |
249 | logical no_warnings; /* -w: suppress warnings */ | |
250 | logical cxref_style; /* -x: create cxref style output */ | |
251 | logical cplusplus; /* .[hc] means C++, not C */ | |
201f9f2b | 252 | logical noindentypedefs; /* -I: ignore indentation in C */ |
55597f90 | 253 | #define permit_duplicates TRUE /* allow duplicate tags */ |
c6d46f5f | 254 | |
7537186d FP |
255 | struct option longopts[] = |
256 | { | |
4746118a JB |
257 | { "append", no_argument, NULL, 'a' }, |
258 | { "backward-search", no_argument, NULL, 'B' }, | |
259 | { "c++", no_argument, NULL, 'C' }, | |
260 | { "cxref", no_argument, NULL, 'x' }, | |
261 | { "defines", no_argument, NULL, 'd' }, | |
7537186d | 262 | { "help", no_argument, NULL, 'h' }, |
4746118a | 263 | { "help", no_argument, NULL, 'H' }, |
201f9f2b | 264 | { "ignore-indentation", no_argument, NULL, 'I' }, |
4746118a | 265 | { "include", required_argument, NULL, 'i' }, |
b9755a12 | 266 | { "language", required_argument, NULL, 'l' }, |
4746118a | 267 | { "no-defines", no_argument, NULL, 'D' }, |
b9755a12 | 268 | { "no-regex", no_argument, NULL, 'R' }, |
4746118a JB |
269 | { "no-warn", no_argument, NULL, 'w' }, |
270 | { "output", required_argument, NULL, 'o' }, | |
b9755a12 | 271 | { "regex", required_argument, NULL, 'r' }, |
4746118a JB |
272 | { "typedefs", no_argument, NULL, 't' }, |
273 | { "typedefs-and-c++", no_argument, NULL, 'T' }, | |
274 | { "update", no_argument, NULL, 'u' }, | |
275 | { "version", no_argument, NULL, 'V' }, | |
276 | { "vgrind", no_argument, NULL, 'v' }, | |
277 | { 0 } | |
278 | }; | |
279 | ||
b9755a12 FP |
280 | #ifdef ETAGS_REGEXPS |
281 | /* Structure defining a regular expression. Elements are | |
282 | the compiled pattern, and the name string. */ | |
283 | struct pattern | |
284 | { | |
285 | struct re_pattern_buffer *pattern; | |
286 | struct re_registers regs; | |
287 | char *name_pattern; | |
288 | logical error_signaled; | |
289 | }; | |
290 | ||
291 | /* Number of regexps found. */ | |
292 | int num_patterns = 0; | |
293 | ||
294 | /* Array of all regexps. */ | |
295 | struct pattern *patterns = NULL; | |
296 | #endif /* ETAGS_REGEXPS */ | |
297 | ||
298 | /* Language stuff. */ | |
299 | struct lang_entry | |
300 | { | |
301 | char *extension; | |
302 | Lang_function *function; | |
303 | }; | |
304 | ||
305 | /* Table of language names and corresponding functions. */ | |
306 | /* It is ok for a given function to be listed under more than one | |
307 | name. I just didn't. */ | |
308 | /* "auto" language reverts to default behavior. */ | |
309 | struct lang_entry lang_names[] = | |
310 | { | |
311 | { "asm", Asm_labels }, | |
312 | { "c", default_C_entries }, | |
313 | { "c++", Cplusplus_entries }, | |
314 | { "c*", Cstar_entries }, | |
315 | { "fortran", Fortran_functions }, | |
316 | { "lisp", Lisp_functions }, | |
317 | { "none", just_read_file }, | |
318 | { "pascal", Pascal_functions }, | |
319 | { "scheme" , Scheme_functions }, | |
320 | { "tex", TeX_functions }, | |
321 | { "auto", NULL }, | |
322 | { NULL, NULL } | |
323 | }; | |
324 | ||
325 | /* Table of file extensions and corresponding language functions. */ | |
326 | struct lang_entry lang_extensions[] = | |
327 | { | |
328 | /* Assume that ".s" or ".a" is assembly code. -wolfgang. | |
329 | Or even ".sa". */ | |
330 | { "a", Asm_labels }, /* Unix assembler */ | |
331 | { "asm", Asm_labels }, /* Microcontroller assembly */ | |
332 | { "def", Asm_labels }, /* BSO/Tasking definition includes */ | |
333 | { "inc", Asm_labels }, /* Microcontroller include files */ | |
334 | { "ins", Asm_labels }, /* Microcontroller include files */ | |
335 | { "s", Asm_labels }, | |
336 | { "sa", Asm_labels }, /* Unix assembler */ | |
337 | { "src", Asm_labels }, /* BSO/Tasking C compiler output */ | |
338 | ||
339 | /* .aux, .bbl, .clo, .cls, .dtx or .tex implies LaTeX source code. */ | |
340 | { "aux", TeX_functions }, | |
341 | { "bbl", TeX_functions }, | |
342 | { "clo", TeX_functions }, | |
343 | { "cls", TeX_functions }, | |
344 | { "dtx", TeX_functions }, | |
345 | { "sty", TeX_functions }, | |
346 | { "tex", TeX_functions }, | |
347 | ||
348 | /* .l or .el or .lisp (or .cl or .clisp or ...) implies lisp source code */ | |
349 | { "cl", Lisp_functions }, | |
350 | { "clisp", Lisp_functions }, | |
351 | { "el", Lisp_functions }, | |
352 | { "l", Lisp_functions }, | |
353 | { "lisp", Lisp_functions }, | |
354 | { "lsp", Lisp_functions }, | |
355 | ||
356 | /* .scm or .sm or .scheme implies scheme source code */ | |
357 | { "SCM", Scheme_functions }, | |
358 | { "SM", Scheme_functions }, | |
359 | { "oak", Scheme_functions }, | |
360 | { "sch", Scheme_functions }, | |
361 | { "scheme", Scheme_functions }, | |
362 | { "scm", Scheme_functions }, | |
363 | { "sm", Scheme_functions }, | |
364 | { "t", Scheme_functions }, | |
365 | /* FIXME Can't do the `SCM' or `scm' prefix with a version number */ | |
366 | ||
367 | /* Note that ".c" and ".h" can be considered C++, if the --c++ | |
368 | flag was given. That is why default_C_entries is called here. */ | |
369 | { "c", default_C_entries }, | |
370 | { "h", default_C_entries }, | |
371 | ||
2bd88040 FP |
372 | /* .C or .H or .c++ or .cc or .cpp or .cxx or .h++ or .hh or .hxx: |
373 | a C++ file */ | |
b9755a12 FP |
374 | { "C", Cplusplus_entries }, |
375 | { "H", Cplusplus_entries }, | |
2bd88040 | 376 | { "c++", Cplusplus_entries }, |
b9755a12 FP |
377 | { "cc", Cplusplus_entries }, |
378 | { "cpp", Cplusplus_entries }, | |
379 | { "cxx", Cplusplus_entries }, | |
2bd88040 | 380 | { "h++", Cplusplus_entries }, |
b9755a12 FP |
381 | { "hh", Cplusplus_entries }, |
382 | { "hxx", Cplusplus_entries }, | |
383 | ||
384 | /* .y: a yacc file */ | |
385 | { "y", Yacc_entries }, | |
386 | ||
387 | /* .cs or .hs: a C* file */ | |
388 | { "cs", Cstar_entries }, | |
389 | { "hs", Cstar_entries }, | |
390 | ||
391 | /* .f and .for are FORTRAN. */ | |
392 | { "F", Fortran_functions }, | |
393 | { "f", Fortran_functions }, | |
394 | { "for", Fortran_functions }, | |
395 | ||
396 | /* .pl implies prolog source code */ | |
397 | { "pl", Prolog_functions }, | |
398 | ||
399 | /* .p or .pas: a Pascal file */ | |
400 | { "p", Pascal_functions }, | |
401 | { "pas", Pascal_functions }, | |
402 | ||
403 | { NULL, NULL } | |
404 | }; | |
405 | ||
406 | /* Non-NULL if language fixed. */ | |
407 | Lang_function *lang_func = NULL; | |
408 | ||
c6d46f5f | 409 | \f |
b9755a12 FP |
410 | void |
411 | print_language_names () | |
412 | { | |
413 | struct lang_entry *name, *ext; | |
414 | ||
415 | puts ("\nThese are the currently supported languages, along with the\n\ | |
416 | default extensions for files:"); | |
417 | for (name = lang_names; name->extension; ++name) | |
418 | { | |
419 | printf ("\t%s\t", name->extension); | |
420 | for (ext = lang_extensions; ext->extension; ++ext) | |
421 | if (name->function == ext->function) | |
422 | printf (" .%s", ext->extension); | |
423 | puts (""); | |
424 | } | |
425 | puts ("Where `auto' means use default language for files based on filename\n\ | |
426 | extension, and `none' means only do regexp processing on files.\n\ | |
427 | If no language is specified and no extension is found for some file,\n\ | |
428 | Fortran is tried first; if no tags are found, C is tried next."); | |
429 | } | |
430 | ||
4746118a JB |
431 | void |
432 | print_version () | |
433 | { | |
32daa216 | 434 | #ifdef VERSION |
cc582743 | 435 | printf ("%s for Emacs version %s.\n", (CTAGS) ? "CTAGS" : "ETAGS", VERSION); |
32daa216 FP |
436 | #else |
437 | printf ("%s for Emacs version 19.\n", (CTAGS) ? "CTAGS" : "ETAGS"); | |
438 | #endif | |
4746118a | 439 | |
1a0d8c80 | 440 | exit (GOOD); |
4746118a JB |
441 | } |
442 | ||
443 | void | |
444 | print_help () | |
445 | { | |
446 | printf ("These are the options accepted by %s. You may use unambiguous\n\ | |
b9755a12 | 447 | abbreviations for the long option names. A - as file name means read\n\ |
1a0d8c80 | 448 | names from stdin.\n\n", progname); |
4746118a | 449 | |
52cc7c59 JB |
450 | puts ("-a, --append\n\ |
451 | Append tag entries to existing tags file."); | |
1a0d8c80 | 452 | |
32daa216 FP |
453 | if (CTAGS) |
454 | puts ("-B, --backward-search\n\ | |
1a0d8c80 | 455 | Write the search commands for the tag entries using '?', the\n\ |
3ad2882c | 456 | backward-search command instead of '/', the forward-search command."); |
1a0d8c80 | 457 | |
52cc7c59 | 458 | puts ("-C, --c++\n\ |
b9755a12 | 459 | Treat files whose extension defaults to C language as C++ files."); |
4746118a | 460 | |
32daa216 FP |
461 | if (CTAGS) |
462 | puts ("-d, --defines\n\ | |
ee70dba5 | 463 | Create tag entries for constant C #defines, too."); |
32daa216 FP |
464 | else |
465 | puts ("-D, --no-defines\n\ | |
ee70dba5 FP |
466 | Don't create tag entries for constant C #defines. This makes\n\ |
467 | the tags file smaller."); | |
4746118a | 468 | |
32daa216 | 469 | if (!CTAGS) |
b9755a12 FP |
470 | { |
471 | puts ("-i FILE, --include=FILE\n\ | |
1a0d8c80 FP |
472 | Include a note in tag file indicating that, when searching for\n\ |
473 | a tag, one should also consult the tags file FILE after\n\ | |
474 | checking the current file."); | |
b9755a12 FP |
475 | puts ("-l LANG, --language=LANG\n\ |
476 | Force the following files to be considered as written in the\n\ | |
477 | named language up to the next --language=LANG option."); | |
7537186d FP |
478 | } |
479 | ||
b9755a12 | 480 | #ifdef ETAGS_REGEXPS |
7537186d | 481 | puts ("-r /REGEXP/, --regex=/REGEXP/\n\ |
b9755a12 FP |
482 | Make a tag for each line matching pattern REGEXP in the\n\ |
483 | following files. REGEXP is anchored (as if preceded by ^).\n\ | |
484 | The form /REGEXP/NAME/ creates a named tag. For example Tcl\n\ | |
485 | named tags can be created with:\n\ | |
486 | --regex=/proc[ \\t]+\\([^ \\t]+\\)/\\1/."); | |
7537186d | 487 | puts ("-R, --no-regex\n\ |
b9755a12 FP |
488 | Don't create tags from regexps for the following files."); |
489 | #endif /* ETAGS_REGEXPS */ | |
1a0d8c80 FP |
490 | puts ("-o FILE, --output=FILE\n\ |
491 | Write the tags to FILE."); | |
201f9f2b | 492 | puts ("-I, --ignore-indentation\n\ |
4746118a JB |
493 | Don't rely on indentation quite as much as normal. Currently,\n\ |
494 | this means not to assume that a closing brace in the first\n\ | |
495 | column is the final brace of a function or structure\n\ | |
32daa216 | 496 | definition in C and C++."); |
4746118a | 497 | |
32daa216 FP |
498 | if (CTAGS) |
499 | { | |
500 | puts ("-t, --typedefs\n\ | |
501 | Generate tag entries for C typedefs."); | |
502 | puts ("-T, --typedefs-and-c++\n\ | |
503 | Generate tag entries for C typedefs, C struct/enum/union tags,\n\ | |
504 | and C++ member functions."); | |
32daa216 | 505 | puts ("-u, --update\n\ |
4746118a JB |
506 | Update the tag entries for the given files, leaving tag\n\ |
507 | entries for other files in place. Currently, this is\n\ | |
508 | implemented by deleting the existing entries for the given\n\ | |
509 | files and then rewriting the new entries at the end of the\n\ | |
510 | tags file. It is often faster to simply rebuild the entire\n\ | |
52cc7c59 | 511 | tag file than to use this."); |
32daa216 | 512 | puts ("-v, --vgrind\n\ |
4746118a JB |
513 | Generates an index of items intended for human consumption,\n\ |
514 | similar to the output of vgrind. The index is sorted, and\n\ | |
52cc7c59 | 515 | gives the page number of each item."); |
b9755a12 FP |
516 | puts ("-w, --no-warn\n\ |
517 | Suppress warning messages about entries defined in multiple\n\ | |
518 | files."); | |
32daa216 | 519 | puts ("-x, --cxref\n\ |
4746118a JB |
520 | Like --vgrind, but in the style of cxref, rather than vgrind.\n\ |
521 | The output uses line numbers instead of page numbers, but\n\ | |
522 | beyond that the differences are cosmetic; try both to see\n\ | |
52cc7c59 | 523 | which you like."); |
32daa216 | 524 | } |
4746118a JB |
525 | |
526 | puts ("-V, --version\n\ | |
527 | Print the version of the program.\n\ | |
7537186d | 528 | -h, --help\n\ |
4746118a JB |
529 | Print this help message."); |
530 | ||
b9755a12 FP |
531 | print_language_names (); |
532 | ||
1a0d8c80 | 533 | exit (GOOD); |
4746118a JB |
534 | } |
535 | ||
536 | \f | |
b9755a12 FP |
537 | enum argument_type |
538 | { | |
539 | at_language, | |
540 | at_regexp, | |
541 | at_filename | |
542 | }; | |
543 | ||
544 | /* This structure helps us allow mixing of --lang and filenames. */ | |
545 | typedef struct | |
546 | { | |
547 | enum argument_type arg_type; | |
548 | char *what; | |
549 | Lang_function *function; | |
550 | } ARGUMENT; | |
551 | ||
552 | #ifdef VMS /* VMS specific functions */ | |
553 | ||
554 | #define EOS '\0' | |
555 | ||
556 | /* This is a BUG! ANY arbitrary limit is a BUG! | |
557 | Won't someone please fix this? */ | |
558 | #define MAX_FILE_SPEC_LEN 255 | |
559 | typedef struct { | |
560 | short curlen; | |
561 | char body[MAX_FILE_SPEC_LEN + 1]; | |
562 | } vspec; | |
563 | ||
564 | /* | |
565 | v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names | |
566 | returning in each successive call the next filename matching the input | |
567 | spec. The function expects that each in_spec passed | |
568 | to it will be processed to completion; in particular, up to and | |
569 | including the call following that in which the last matching name | |
570 | is returned, the function ignores the value of in_spec, and will | |
571 | only start processing a new spec with the following call. | |
572 | If an error occurs, on return out_spec contains the value | |
573 | of in_spec when the error occurred. | |
574 | ||
575 | With each successive filename returned in out_spec, the | |
576 | function's return value is one. When there are no more matching | |
577 | names the function returns zero. If on the first call no file | |
578 | matches in_spec, or there is any other error, -1 is returned. | |
579 | */ | |
580 | ||
581 | #include <rmsdef.h> | |
582 | #include <descrip.h> | |
583 | #define OUTSIZE MAX_FILE_SPEC_LEN | |
584 | short | |
585 | fn_exp (out, in) | |
586 | vspec *out; | |
587 | char *in; | |
588 | { | |
589 | static long context = 0; | |
590 | static struct dsc$descriptor_s o; | |
591 | static struct dsc$descriptor_s i; | |
592 | static logical pass1 = TRUE; | |
593 | long status; | |
594 | short retval; | |
595 | ||
596 | if (pass1) | |
597 | { | |
598 | pass1 = FALSE; | |
599 | o.dsc$a_pointer = (char *) out; | |
600 | o.dsc$w_length = (short)OUTSIZE; | |
601 | i.dsc$a_pointer = in; | |
602 | i.dsc$w_length = (short)strlen(in); | |
603 | i.dsc$b_dtype = DSC$K_DTYPE_T; | |
604 | i.dsc$b_class = DSC$K_CLASS_S; | |
605 | o.dsc$b_dtype = DSC$K_DTYPE_VT; | |
606 | o.dsc$b_class = DSC$K_CLASS_VS; | |
607 | } | |
608 | if ((status = lib$find_file(&i, &o, &context, 0, 0)) == RMS$_NORMAL) | |
609 | { | |
610 | out->body[out->curlen] = EOS; | |
611 | return 1; | |
612 | } | |
613 | else if (status == RMS$_NMF) | |
614 | retval = 0; | |
615 | else | |
616 | { | |
617 | strcpy(out->body, in); | |
618 | retval = -1; | |
619 | } | |
620 | lib$find_file_end(&context); | |
621 | pass1 = TRUE; | |
622 | return retval; | |
623 | } | |
624 | ||
625 | /* | |
626 | v1.01 nmm 19-Aug-85 gfnames - return in successive calls the | |
627 | name of each file specified by the provided arg expanding wildcards. | |
628 | */ | |
629 | char * | |
630 | gfnames (arg, p_error) | |
631 | char *arg; | |
632 | logical *p_error; | |
633 | { | |
634 | static vspec filename = {MAX_FILE_SPEC_LEN, "\0"}; | |
635 | ||
636 | switch (fn_exp (&filename, arg)) | |
637 | { | |
638 | case 1: | |
639 | *p_error = FALSE; | |
640 | return filename.body; | |
641 | case 0: | |
642 | *p_error = FALSE; | |
643 | return NULL; | |
644 | default: | |
645 | *p_error = TRUE; | |
646 | return filename.body; | |
647 | } | |
648 | } | |
649 | ||
650 | #ifndef OLD /* Newer versions of VMS do provide `system'. */ | |
651 | system (cmd) | |
652 | char *cmd; | |
653 | { | |
654 | fprintf (stderr, "system() function not implemented under VMS\n"); | |
655 | } | |
656 | #endif | |
657 | ||
658 | #define VERSION_DELIM ';' | |
659 | char *massage_name (s) | |
660 | char *s; | |
661 | { | |
662 | char *start = s; | |
663 | ||
664 | for ( ; *s; s++) | |
665 | if (*s == VERSION_DELIM) | |
666 | { | |
667 | *s = EOS; | |
668 | break; | |
669 | } | |
670 | else | |
671 | *s = tolower(*s); | |
672 | return start; | |
673 | } | |
674 | #endif /* VMS */ | |
675 | ||
676 | \f | |
c6d46f5f JB |
677 | void |
678 | main (argc, argv) | |
679 | int argc; | |
680 | char *argv[]; | |
681 | { | |
c6d46f5f | 682 | int i; |
1e134a5f | 683 | unsigned int nincluded_files = 0; |
72a339d7 | 684 | char **included_files = xnew (argc, char *); |
c6d46f5f | 685 | char *this_file; |
b9755a12 FP |
686 | ARGUMENT *argbuffer; |
687 | int current_arg = 0, file_count = 0; | |
55597f90 | 688 | struct linebuffer filename_lb; |
c6d46f5f | 689 | #ifdef VMS |
b9755a12 | 690 | logical got_err; |
c6d46f5f | 691 | #endif |
46c145db | 692 | |
c05b6df5 | 693 | #ifdef DOS_NT |
42680d3c | 694 | _fmode = O_BINARY; /* all of files are treated as binary files */ |
c05b6df5 | 695 | #endif /* DOS_NT */ |
c6880c90 | 696 | |
c6d46f5f JB |
697 | progname = argv[0]; |
698 | ||
b9755a12 FP |
699 | /* Allocate enough no matter what happens. Overkill, but each one |
700 | is small. */ | |
701 | argbuffer = xnew (argc, ARGUMENT); | |
702 | ||
703 | #ifdef ETAGS_REGEXPS | |
704 | /* Set syntax for regular expression routines. */ | |
705 | re_set_syntax (RE_SYNTAX_EMACS); | |
706 | #endif /* ETAGS_REGEXPS */ | |
707 | ||
c6d46f5f JB |
708 | /* |
709 | * If etags, always find typedefs and structure tags. Why not? | |
710 | * Also default is to find macro constants. | |
711 | */ | |
32daa216 | 712 | if (!CTAGS) |
55597f90 | 713 | typedefs = typedefs_and_cplusplus = constantypedefs = TRUE; |
c6d46f5f | 714 | |
dcc89e63 | 715 | while (1) |
c6d46f5f | 716 | { |
b9755a12 | 717 | int opt = getopt_long (argc, argv, |
201f9f2b | 718 | "-aCdDf:Il:o:r:RStTi:BuvxwVhH", longopts, 0); |
4746118a JB |
719 | |
720 | if (opt == EOF) | |
721 | break; | |
722 | ||
723 | switch (opt) | |
c6d46f5f | 724 | { |
b02c5fea FP |
725 | case 0: |
726 | /* If getopt returns 0, then it has already processed a | |
4746118a JB |
727 | long-named option. We should do nothing. */ |
728 | break; | |
729 | ||
b9755a12 FP |
730 | case 1: |
731 | /* This means that a filename has been seen. Record it. */ | |
732 | argbuffer[current_arg].arg_type = at_filename; | |
733 | argbuffer[current_arg].what = optarg; | |
734 | ++current_arg; | |
735 | ++file_count; | |
736 | break; | |
737 | ||
4746118a JB |
738 | /* Common options. */ |
739 | case 'a': | |
55597f90 | 740 | append_to_tagfile = TRUE; |
4746118a JB |
741 | break; |
742 | case 'C': | |
55597f90 | 743 | cplusplus = TRUE; |
4746118a JB |
744 | break; |
745 | case 'd': | |
55597f90 | 746 | constantypedefs = TRUE; |
4746118a JB |
747 | break; |
748 | case 'D': | |
55597f90 | 749 | constantypedefs = FALSE; |
4746118a | 750 | break; |
32daa216 | 751 | case 'f': /* for compatibility with old makefiles */ |
4746118a | 752 | case 'o': |
6dd5561c | 753 | if (tagfile) |
c6d46f5f | 754 | { |
201f9f2b FP |
755 | fprintf (stderr, "%s: -%c option may only be given once.\n", |
756 | progname, opt); | |
c6d46f5f JB |
757 | goto usage; |
758 | } | |
6dd5561c | 759 | tagfile = optarg; |
4746118a | 760 | break; |
201f9f2b FP |
761 | case 'I': |
762 | case 'S': /* for backward compatibility */ | |
763 | noindentypedefs = TRUE; | |
764 | break; | |
b9755a12 FP |
765 | case 'l': |
766 | if (!get_language (optarg, &argbuffer[current_arg].function)) | |
767 | { | |
768 | fprintf (stderr, "%s: language \"%s\" not recognized.\n", | |
769 | progname, optarg); | |
770 | goto usage; | |
771 | } | |
772 | argbuffer[current_arg].arg_type = at_language; | |
773 | ++current_arg; | |
774 | break; | |
775 | #ifdef ETAGS_REGEXPS | |
776 | case 'r': | |
777 | argbuffer[current_arg].arg_type = at_regexp; | |
778 | argbuffer[current_arg].what = optarg; | |
779 | ++current_arg; | |
780 | break; | |
781 | case 'R': | |
782 | argbuffer[current_arg].arg_type = at_regexp; | |
783 | argbuffer[current_arg].what = NULL; | |
784 | ++current_arg; | |
785 | break; | |
786 | #endif /* ETAGS_REGEXPS */ | |
4746118a JB |
787 | case 'V': |
788 | print_version (); | |
789 | break; | |
7537186d | 790 | case 'h': |
4746118a JB |
791 | case 'H': |
792 | print_help (); | |
793 | break; | |
0c1fd2e3 | 794 | case 't': |
55597f90 | 795 | typedefs = TRUE; |
0c1fd2e3 FP |
796 | break; |
797 | case 'T': | |
55597f90 | 798 | typedefs = typedefs_and_cplusplus = TRUE; |
0c1fd2e3 | 799 | break; |
b02c5fea | 800 | #if (!CTAGS) |
4746118a JB |
801 | /* Etags options */ |
802 | case 'i': | |
4746118a JB |
803 | included_files[nincluded_files++] = optarg; |
804 | break; | |
b02c5fea | 805 | #else /* CTAGS */ |
4746118a JB |
806 | /* Ctags options. */ |
807 | case 'B': | |
808 | searchar = '?'; | |
4746118a | 809 | break; |
4746118a | 810 | case 'u': |
55597f90 | 811 | update = TRUE; |
4746118a JB |
812 | break; |
813 | case 'v': | |
55597f90 | 814 | vgrind_style = TRUE; |
4746118a JB |
815 | /*FALLTHRU*/ |
816 | case 'x': | |
55597f90 | 817 | cxref_style = TRUE; |
4746118a JB |
818 | break; |
819 | case 'w': | |
55597f90 | 820 | no_warnings = TRUE; |
4746118a | 821 | break; |
b02c5fea | 822 | #endif /* CTAGS */ |
4746118a JB |
823 | default: |
824 | goto usage; | |
c6d46f5f | 825 | } |
c6d46f5f JB |
826 | } |
827 | ||
b9755a12 FP |
828 | for (; optind < argc; ++optind) |
829 | { | |
830 | argbuffer[current_arg].arg_type = at_filename; | |
831 | argbuffer[current_arg].what = argv[optind]; | |
832 | ++current_arg; | |
833 | ++file_count; | |
834 | } | |
835 | ||
836 | if (nincluded_files == 0 && file_count == 0) | |
c6d46f5f | 837 | { |
4746118a JB |
838 | fprintf (stderr, "%s: No input files specified.\n", progname); |
839 | ||
c6d46f5f | 840 | usage: |
201f9f2b FP |
841 | fprintf (stderr, "\tTry `%s --help' for a complete list of options.\n", |
842 | progname); | |
c6d46f5f JB |
843 | exit (BAD); |
844 | } | |
845 | ||
6dd5561c | 846 | if (tagfile == NULL) |
c6d46f5f | 847 | { |
6dd5561c | 848 | tagfile = CTAGS ? "tags" : "TAGS"; |
c6d46f5f | 849 | } |
b02c5fea | 850 | cwd = etags_getcwd (); /* the current working directory */ |
46c145db | 851 | strcat (cwd, "/"); |
6dd5561c | 852 | if (streq (tagfile, "-")) |
46c145db | 853 | { |
6dd5561c | 854 | tagfiledir = cwd; |
46c145db FP |
855 | } |
856 | else | |
857 | { | |
6dd5561c | 858 | tagfiledir = absolute_dirname (tagfile, cwd); |
46c145db | 859 | } |
c6d46f5f | 860 | |
b9755a12 | 861 | init (); /* set up boolean "functions" */ |
c6d46f5f JB |
862 | |
863 | initbuffer (&lb); | |
2bd88040 | 864 | initbuffer (&token_str); |
13fde0cd RS |
865 | initbuffer (&lbs[0].lb); |
866 | initbuffer (&lbs[1].lb); | |
c6d46f5f | 867 | initbuffer (&filename_lb); |
b9755a12 | 868 | |
32daa216 | 869 | if (!CTAGS) |
c6d46f5f | 870 | { |
6dd5561c FP |
871 | if (streq (tagfile, "-")) |
872 | tagf = stdout; | |
c6d46f5f | 873 | else |
6dd5561c FP |
874 | tagf = fopen (tagfile, append_to_tagfile ? "a" : "w"); |
875 | if (tagf == NULL) | |
cdc1f6a7 | 876 | pfatal (tagfile); |
c6d46f5f JB |
877 | } |
878 | ||
b9755a12 FP |
879 | /* |
880 | * Loop through files finding functions. | |
881 | */ | |
882 | for (i = 0; i < current_arg; ++i) | |
c6d46f5f | 883 | { |
b9755a12 | 884 | switch (argbuffer[i].arg_type) |
c6d46f5f | 885 | { |
b9755a12 FP |
886 | case at_language: |
887 | lang_func = argbuffer[i].function; | |
888 | break; | |
889 | #ifdef ETAGS_REGEXPS | |
890 | case at_regexp: | |
891 | add_regex (argbuffer[i].what); | |
892 | break; | |
c6d46f5f | 893 | #endif |
b9755a12 FP |
894 | case at_filename: |
895 | #ifdef VMS | |
896 | while ((this_file = gfnames (argbuffer[i].what, &got_err)) != NULL) | |
897 | { | |
898 | if (got_err) | |
899 | { | |
900 | error ("Can't find file %s\n", this_file); | |
901 | argc--, argv++; | |
902 | } | |
903 | else | |
904 | { | |
905 | this_file = massage_name (this_file); | |
906 | } | |
c6d46f5f | 907 | #else |
b9755a12 | 908 | this_file = argbuffer[i].what; |
c6d46f5f | 909 | #endif |
b9755a12 FP |
910 | /* Input file named "-" means read file names from stdin |
911 | and use them. */ | |
912 | if (streq (this_file, "-")) | |
913 | { | |
914 | while (!feof (stdin)) | |
915 | { | |
b9755a12 FP |
916 | (void) readline_internal (&filename_lb, stdin); |
917 | if (strlen (filename_lb.buffer) > 0) | |
918 | process_file (filename_lb.buffer); | |
919 | } | |
920 | } | |
921 | else | |
922 | process_file (this_file); | |
923 | #ifdef VMS | |
c6d46f5f | 924 | } |
b9755a12 FP |
925 | #endif |
926 | break; | |
c6d46f5f | 927 | } |
46c145db | 928 | } |
32daa216 | 929 | if (!CTAGS) |
c6d46f5f | 930 | { |
1e134a5f | 931 | while (nincluded_files-- > 0) |
6dd5561c | 932 | fprintf (tagf, "\f\n%s,include\n", *included_files++); |
1e134a5f | 933 | |
55597f90 | 934 | fclose (tagf); |
1a0d8c80 | 935 | exit (GOOD); |
c6d46f5f JB |
936 | } |
937 | ||
55597f90 FP |
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. */ | |
c6d46f5f JB |
940 | if (cxref_style) |
941 | { | |
55597f90 FP |
942 | tagf = fopen (tagfile, append_to_tagfile ? "a" : "w"); |
943 | if (tagf == NULL) | |
944 | pfatal (tagfile); | |
c6d46f5f JB |
945 | put_entries (head); |
946 | exit (GOOD); | |
947 | } | |
55597f90 | 948 | |
4746118a | 949 | if (update) |
c6d46f5f | 950 | { |
55597f90 | 951 | char cmd[BUFSIZ]; |
b9755a12 | 952 | for (i = 0; i < current_arg; ++i) |
c6d46f5f | 953 | { |
55597f90 | 954 | if (argbuffer[i].arg_type != at_filename) |
b9755a12 | 955 | continue; |
c6d46f5f JB |
956 | sprintf (cmd, |
957 | "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS", | |
b9755a12 | 958 | tagfile, argbuffer[i].what, tagfile); |
55597f90 FP |
959 | if (system (cmd) != GOOD) |
960 | fatal ("failed to execute shell command"); | |
c6d46f5f | 961 | } |
55597f90 | 962 | append_to_tagfile = TRUE; |
c6d46f5f | 963 | } |
55597f90 | 964 | |
6dd5561c FP |
965 | tagf = fopen (tagfile, append_to_tagfile ? "a" : "w"); |
966 | if (tagf == NULL) | |
55597f90 | 967 | pfatal (tagfile); |
c6d46f5f | 968 | put_entries (head); |
55597f90 FP |
969 | fclose (tagf); |
970 | ||
c6d46f5f JB |
971 | if (update) |
972 | { | |
55597f90 | 973 | char cmd[BUFSIZ]; |
6dd5561c | 974 | sprintf (cmd, "sort %s -o %s", tagfile, tagfile); |
55597f90 | 975 | exit (system (cmd)); |
c6d46f5f JB |
976 | } |
977 | exit (GOOD); | |
978 | } | |
979 | ||
980 | ||
b9755a12 FP |
981 | /* |
982 | * Set the language, given the name. | |
983 | */ | |
984 | logical | |
985 | get_language (language, func) | |
986 | char *language; | |
987 | Lang_function **func; | |
988 | { | |
989 | struct lang_entry *lang; | |
990 | ||
991 | for (lang = lang_names; lang->extension; ++lang) | |
992 | { | |
993 | if (streq (language, lang->extension)) | |
994 | { | |
995 | *func = lang->function; | |
996 | return TRUE; | |
997 | } | |
998 | } | |
999 | ||
1000 | return FALSE; | |
1001 | } | |
1002 | ||
1003 | ||
c6d46f5f JB |
1004 | /* |
1005 | * This routine is called on each file argument. | |
1006 | */ | |
1007 | void | |
1008 | process_file (file) | |
1009 | char *file; | |
1010 | { | |
1011 | struct stat stat_buf; | |
55597f90 | 1012 | FILE *inf; |
c6d46f5f | 1013 | |
42680d3c | 1014 | if (stat (file, &stat_buf) == 0 && !S_ISREG (stat_buf.st_mode)) |
c6d46f5f JB |
1015 | { |
1016 | fprintf (stderr, "Skipping %s: it is not a regular file.\n", file); | |
1017 | return; | |
1018 | } | |
6dd5561c | 1019 | if (streq (file, tagfile) && !streq (tagfile, "-")) |
c6d46f5f JB |
1020 | { |
1021 | fprintf (stderr, "Skipping inclusion of %s in self.\n", file); | |
1022 | return; | |
1023 | } | |
55597f90 FP |
1024 | inf = fopen (file, "r"); |
1025 | if (inf == NULL) | |
42680d3c | 1026 | { |
55597f90 | 1027 | perror (file); |
42680d3c FP |
1028 | return; |
1029 | } | |
55597f90 FP |
1030 | |
1031 | find_entries (file, inf); | |
1032 | ||
32daa216 | 1033 | if (!CTAGS) |
c6d46f5f | 1034 | { |
46c145db FP |
1035 | char *filename; |
1036 | ||
1037 | if (file[0] == '/') | |
1038 | { | |
1039 | /* file is an absolute filename. Canonicalise it. */ | |
1040 | filename = absolute_filename (file, cwd); | |
1041 | } | |
1042 | else | |
1043 | { | |
1044 | /* file is a filename relative to cwd. Make it relative | |
1045 | to the directory of the tags file. */ | |
6dd5561c | 1046 | filename = relative_filename (file, tagfiledir); |
46c145db | 1047 | } |
6dd5561c | 1048 | fprintf (tagf, "\f\n%s,%d\n", filename, total_size_of_entries (head)); |
c6d46f5f JB |
1049 | put_entries (head); |
1050 | free_tree (head); | |
1051 | head = NULL; | |
1052 | } | |
1053 | } | |
1054 | ||
1055 | /* | |
eb8c3be9 | 1056 | * This routine sets up the boolean pseudo-functions which work |
99e0a2e0 | 1057 | * by setting boolean flags dependent upon the corresponding character |
c6d46f5f JB |
1058 | * Every char which is NOT in that string is not a white char. Therefore, |
1059 | * all of the array "_wht" is set to FALSE, and then the elements | |
1060 | * subscripted by the chars in "white" are set to TRUE. Thus "_wht" | |
1061 | * of a char is TRUE if it is the string "white", else FALSE. | |
1062 | */ | |
1063 | void | |
1064 | init () | |
1065 | { | |
13fde0cd RS |
1066 | register char *sp; |
1067 | register int i; | |
c6d46f5f JB |
1068 | |
1069 | for (i = 0; i < 0177; i++) | |
13fde0cd | 1070 | _wht[i] = _etk[i] = _itk[i] = _btk[i] = FALSE; |
c6d46f5f JB |
1071 | for (sp = white; *sp; sp++) |
1072 | _wht[*sp] = TRUE; | |
1073 | for (sp = endtk; *sp; sp++) | |
1074 | _etk[*sp] = TRUE; | |
1075 | for (sp = intk; *sp; sp++) | |
1076 | _itk[*sp] = TRUE; | |
1077 | for (sp = begtk; *sp; sp++) | |
1078 | _btk[*sp] = TRUE; | |
c6d46f5f JB |
1079 | _wht[0] = _wht['\n']; |
1080 | _etk[0] = _etk['\n']; | |
1081 | _btk[0] = _btk['\n']; | |
1082 | _itk[0] = _itk['\n']; | |
c6d46f5f JB |
1083 | } |
1084 | ||
1085 | /* | |
1086 | * This routine opens the specified file and calls the function | |
1087 | * which finds the function and type definitions. | |
1088 | */ | |
55597f90 FP |
1089 | void |
1090 | find_entries (file, inf) | |
c6d46f5f | 1091 | char *file; |
55597f90 | 1092 | FILE *inf; |
c6d46f5f | 1093 | { |
b9755a12 FP |
1094 | char *cp; |
1095 | struct lang_entry *lang; | |
1096 | NODE *old_last_node; | |
1097 | extern NODE *last_node; | |
c6d46f5f | 1098 | |
c6d46f5f | 1099 | curfile = savestr (file); |
b02c5fea | 1100 | cp = etags_strrchr (file, '.'); |
c6d46f5f | 1101 | |
b9755a12 FP |
1102 | /* If user specified a language, use it. */ |
1103 | if (lang_func != NULL) | |
13fde0cd | 1104 | { |
b9755a12 FP |
1105 | lang_func (inf); |
1106 | fclose (inf); | |
55597f90 | 1107 | return; |
13fde0cd | 1108 | } |
b9755a12 FP |
1109 | |
1110 | if (cp) | |
c6d46f5f | 1111 | { |
b9755a12 FP |
1112 | ++cp; |
1113 | for (lang = lang_extensions; lang->extension; ++lang) | |
1114 | { | |
1115 | if (streq (cp, lang->extension)) | |
1116 | { | |
1117 | lang->function (inf); | |
1118 | fclose (inf); | |
55597f90 | 1119 | return; |
b9755a12 FP |
1120 | } |
1121 | } | |
c6d46f5f | 1122 | } |
c6d46f5f | 1123 | |
b9755a12 FP |
1124 | /* Try Fortran. */ |
1125 | old_last_node = last_node; | |
1126 | Fortran_functions (inf); | |
c6d46f5f | 1127 | |
b9755a12 FP |
1128 | /* No Fortran entries found. Try C. */ |
1129 | if (old_last_node == last_node) | |
1130 | default_C_entries (inf); | |
1131 | fclose (inf); | |
c6d46f5f JB |
1132 | } |
1133 | \f | |
1134 | /* Record a tag. */ | |
1135 | /* Should take a TOKEN* instead!! */ | |
c6d46f5f | 1136 | void |
fe0b3356 | 1137 | pfnote (name, is_func, named, linestart, linelen, lno, cno) |
c6d46f5f | 1138 | char *name; /* tag name */ |
55597f90 FP |
1139 | logical is_func; /* tag is a function */ |
1140 | logical named; /* tag different from text of definition */ | |
1141 | char *linestart; /* start of the line where tag is */ | |
1142 | int linelen; /* length of the line where tag is */ | |
1143 | int lno; /* line number */ | |
1144 | long cno; /* character number */ | |
1145 | { | |
1146 | register NODE *np = xnew (1, NODE); | |
c6d46f5f | 1147 | register char *fp; |
c6d46f5f | 1148 | |
c6d46f5f | 1149 | /* If ctags mode, change name "main" to M<thisfilename>. */ |
32daa216 | 1150 | if (CTAGS && !cxref_style && streq (name, "main")) |
c6d46f5f | 1151 | { |
b02c5fea | 1152 | fp = etags_strrchr (curfile, '/'); |
55597f90 FP |
1153 | np->name = concat ("M", fp == 0 ? curfile : fp + 1, ""); |
1154 | fp = etags_strrchr (np->name, '.'); | |
c6d46f5f | 1155 | if (fp && fp[1] != '\0' && fp[2] == '\0') |
55597f90 FP |
1156 | fp[0] = 0; |
1157 | np->named = TRUE; | |
1158 | } | |
1159 | else | |
1160 | { | |
1161 | np->name = name; | |
1162 | np->named = named; | |
c6d46f5f | 1163 | } |
4b533b5b | 1164 | np->been_warned = FALSE; |
c6d46f5f JB |
1165 | np->file = curfile; |
1166 | np->is_func = is_func; | |
c6d46f5f | 1167 | np->lno = lno; |
aab1fdae FP |
1168 | /* Our char numbers are 0-base, because of C language tradition? |
1169 | ctags compatibility? old versions compatibility? I don't know. | |
1170 | Anyway, since emacs's are 1-base we espect etags.el to take care | |
1171 | of the difference. If we wanted to have 1-based numbers, we would | |
1172 | uncomment the +1 below. */ | |
1173 | np->cno = cno /* + 1 */ ; | |
55597f90 FP |
1174 | np->left = np->right = NULL; |
1175 | np->pat = savenstr (linestart, ((CTAGS && !cxref_style) ? 50 : linelen)); | |
c6d46f5f JB |
1176 | |
1177 | add_node (np, &head); | |
1178 | } | |
1179 | ||
1180 | /* | |
1181 | * free_tree () | |
1182 | * recurse on left children, iterate on right children. | |
1183 | */ | |
1184 | void | |
1185 | free_tree (node) | |
1186 | register NODE *node; | |
1187 | { | |
1188 | while (node) | |
1189 | { | |
1190 | register NODE *node_right = node->right; | |
1191 | free_tree (node->left); | |
55597f90 FP |
1192 | if (node->named) |
1193 | free (node->name); | |
c6d46f5f JB |
1194 | free (node->pat); |
1195 | free ((char *) node); | |
1196 | node = node_right; | |
1197 | } | |
1198 | } | |
1199 | ||
1200 | /* | |
1201 | * add_node () | |
1202 | * Adds a node to the tree of nodes. In etags mode, we don't keep | |
1203 | * it sorted; we just keep a linear list. In ctags mode, maintain | |
1204 | * an ordered tree, with no attempt at balancing. | |
1205 | * | |
1206 | * add_node is the only function allowed to add nodes, so it can | |
1207 | * maintain state. | |
1208 | */ | |
6dd5561c | 1209 | NODE *last_node = NULL; |
c6d46f5f JB |
1210 | void |
1211 | add_node (node, cur_node_p) | |
1212 | NODE *node, **cur_node_p; | |
1213 | { | |
1214 | register int dif; | |
1215 | register NODE *cur_node = *cur_node_p; | |
c6d46f5f JB |
1216 | |
1217 | if (cur_node == NULL) | |
1218 | { | |
1219 | *cur_node_p = node; | |
1220 | last_node = node; | |
1221 | return; | |
1222 | } | |
1223 | ||
32daa216 | 1224 | if (!CTAGS) |
c6d46f5f JB |
1225 | { |
1226 | /* Etags Mode */ | |
1a0d8c80 FP |
1227 | if (last_node == NULL) |
1228 | fatal ("internal error in add_node", 0); | |
c6d46f5f JB |
1229 | last_node->right = node; |
1230 | last_node = node; | |
1231 | } | |
1232 | else | |
1233 | { | |
1234 | /* Ctags Mode */ | |
1235 | dif = strcmp (node->name, cur_node->name); | |
1236 | ||
1237 | /* | |
1238 | * If this tag name matches an existing one, then | |
1239 | * do not add the node, but maybe print a warning. | |
1240 | */ | |
1241 | if (!dif) | |
1242 | { | |
1243 | if (node->file == cur_node->file) | |
1244 | { | |
1245 | if (!no_warnings) | |
1246 | { | |
1247 | fprintf (stderr, "Duplicate entry in file %s, line %d: %s\n", | |
1248 | node->file, lineno, node->name); | |
1249 | fprintf (stderr, "Second entry ignored\n"); | |
1250 | } | |
1251 | return; | |
1252 | } | |
1253 | if (!cur_node->been_warned && !no_warnings) | |
1254 | { | |
1255 | fprintf (stderr, | |
1256 | "Duplicate entry in files %s and %s: %s (Warning only)\n", | |
1257 | node->file, cur_node->file, node->name); | |
1258 | } | |
1259 | cur_node->been_warned = TRUE; | |
1260 | return; | |
1261 | } | |
1262 | ||
1263 | /* Maybe refuse to add duplicate nodes. */ | |
1264 | if (!permit_duplicates) | |
1265 | { | |
1a0d8c80 FP |
1266 | if (streq (node->name, cur_node->name) |
1267 | && streq (node->file, cur_node->file)) | |
c6d46f5f JB |
1268 | return; |
1269 | } | |
1270 | ||
1271 | /* Actually add the node */ | |
1272 | add_node (node, dif < 0 ? &cur_node->left : &cur_node->right); | |
1273 | } | |
1274 | } | |
1275 | \f | |
1276 | void | |
1277 | put_entries (node) | |
13fde0cd | 1278 | register NODE *node; |
c6d46f5f | 1279 | { |
13fde0cd | 1280 | register char *sp; |
c6d46f5f JB |
1281 | |
1282 | if (node == NULL) | |
1283 | return; | |
1284 | ||
1285 | /* Output subentries that precede this one */ | |
1286 | put_entries (node->left); | |
1287 | ||
1288 | /* Output this entry */ | |
1289 | ||
32daa216 | 1290 | if (!CTAGS) |
c6d46f5f | 1291 | { |
fe0b3356 | 1292 | if (node->named) |
c6d46f5f | 1293 | { |
6dd5561c | 1294 | fprintf (tagf, "%s\177%s\001%d,%d\n", |
cc6d6e58 RM |
1295 | node->pat, node->name, |
1296 | node->lno, node->cno); | |
c6d46f5f JB |
1297 | } |
1298 | else | |
1299 | { | |
6dd5561c | 1300 | fprintf (tagf, "%s\177%d,%d\n", |
cc6d6e58 RM |
1301 | node->pat, |
1302 | node->lno, node->cno); | |
c6d46f5f JB |
1303 | } |
1304 | } | |
1305 | else if (!cxref_style) | |
1306 | { | |
6dd5561c | 1307 | fprintf (tagf, "%s\t%s\t", |
c6d46f5f JB |
1308 | node->name, node->file); |
1309 | ||
1310 | if (node->is_func) | |
1311 | { /* a function */ | |
6dd5561c FP |
1312 | putc (searchar, tagf); |
1313 | putc ('^', tagf); | |
c6d46f5f JB |
1314 | |
1315 | for (sp = node->pat; *sp; sp++) | |
1316 | { | |
1317 | if (*sp == '\\' || *sp == searchar) | |
6dd5561c FP |
1318 | putc ('\\', tagf); |
1319 | putc (*sp, tagf); | |
c6d46f5f | 1320 | } |
6dd5561c | 1321 | putc (searchar, tagf); |
c6d46f5f JB |
1322 | } |
1323 | else | |
1324 | { /* a typedef; text pattern inadequate */ | |
6dd5561c | 1325 | fprintf (tagf, "%d", node->lno); |
c6d46f5f | 1326 | } |
6dd5561c | 1327 | putc ('\n', tagf); |
c6d46f5f JB |
1328 | } |
1329 | else if (vgrind_style) | |
1330 | fprintf (stdout, "%s %s %d\n", | |
1331 | node->name, node->file, (node->lno + 63) / 64); | |
1332 | else | |
daa37602 | 1333 | fprintf (stdout, "%-16s %3d %-16s %s\n", |
c6d46f5f JB |
1334 | node->name, node->lno, node->file, node->pat); |
1335 | ||
1336 | /* Output subentries that follow this one */ | |
1337 | put_entries (node->right); | |
1338 | } | |
1339 | ||
1340 | /* Length of a number's decimal representation. */ | |
1341 | int | |
1342 | number_len (num) | |
1343 | long num; | |
1344 | { | |
1345 | int len = 0; | |
1346 | if (!num) | |
1347 | return 1; | |
1348 | for (; num; num /= 10) | |
1349 | ++len; | |
1350 | return len; | |
1351 | } | |
1352 | ||
1353 | /* | |
1354 | * Return total number of characters that put_entries will output for | |
32daa216 FP |
1355 | * the nodes in the subtree of the specified node. Works only if |
1356 | * we are not ctags, but called only in that case. This count | |
1357 | * is irrelevant with the new tags.el, but is still supplied for | |
1358 | * backward compatibility. | |
c6d46f5f JB |
1359 | */ |
1360 | int | |
1361 | total_size_of_entries (node) | |
13fde0cd | 1362 | register NODE *node; |
c6d46f5f | 1363 | { |
13fde0cd | 1364 | register int total; |
c6d46f5f JB |
1365 | |
1366 | if (node == NULL) | |
1367 | return 0; | |
1368 | ||
1369 | total = 0; | |
1370 | for (; node; node = node->right) | |
1371 | { | |
1372 | /* Count left subentries. */ | |
1373 | total += total_size_of_entries (node->left); | |
1374 | ||
1375 | /* Count this entry */ | |
1376 | total += strlen (node->pat) + 1; | |
1377 | total += number_len ((long) node->lno) + 1 + number_len (node->cno) + 1; | |
fe0b3356 | 1378 | if (node->named) |
c6d46f5f JB |
1379 | total += 1 + strlen (node->name); /* \001name */ |
1380 | } | |
1381 | ||
1382 | return total; | |
1383 | } | |
1384 | \f | |
1385 | /* | |
1386 | * The C symbol tables. | |
1387 | */ | |
55597f90 FP |
1388 | enum sym_type |
1389 | { | |
1390 | st_none, st_C_struct, st_C_enum, st_C_define, st_C_typedef, st_C_typespec | |
1391 | }; | |
c6d46f5f | 1392 | |
42680d3c FP |
1393 | /* Feed stuff between (but not including) %[ and %] lines to: |
1394 | gperf -c -k1,3 -o -p -r -t | |
1395 | %[ | |
1396 | struct C_stab_entry { char *name; int c_ext; enum sym_type type; } | |
1397 | %% | |
1398 | class, C_PLPL, st_C_struct | |
1399 | domain, C_STAR, st_C_struct | |
1400 | union, 0, st_C_struct | |
1401 | struct, 0, st_C_struct | |
1402 | enum, 0, st_C_enum | |
1403 | typedef, 0, st_C_typedef | |
1404 | define, 0, st_C_define | |
1405 | long, 0, st_C_typespec | |
1406 | short, 0, st_C_typespec | |
1407 | int, 0, st_C_typespec | |
1408 | char, 0, st_C_typespec | |
1409 | float, 0, st_C_typespec | |
1410 | double, 0, st_C_typespec | |
1411 | signed, 0, st_C_typespec | |
1412 | unsigned, 0, st_C_typespec | |
1413 | auto, 0, st_C_typespec | |
1414 | void, 0, st_C_typespec | |
1415 | extern, 0, st_C_typespec | |
1416 | static, 0, st_C_typespec | |
1417 | const, 0, st_C_typespec | |
1418 | volatile, 0, st_C_typespec | |
1419 | %] | |
1420 | and replace lines between %< and %> with its output. */ | |
1421 | /*%<*/ | |
1422 | /* C code produced by gperf version 1.8.1 (K&R C version) */ | |
1423 | /* Command-line: gperf -c -k1,3 -o -p -r -t */ | |
1424 | ||
1425 | ||
1426 | struct C_stab_entry { char *name; int c_ext; enum sym_type type; }; | |
1427 | ||
1428 | #define MIN_WORD_LENGTH 3 | |
1429 | #define MAX_WORD_LENGTH 8 | |
1430 | #define MIN_HASH_VALUE 10 | |
1431 | #define MAX_HASH_VALUE 62 | |
c6d46f5f | 1432 | /* |
42680d3c FP |
1433 | 21 keywords |
1434 | 53 is the maximum key range | |
1435 | */ | |
1436 | ||
1437 | static int | |
1438 | hash (str, len) | |
1439 | register char *str; | |
1440 | register int len; | |
1441 | { | |
1442 | static unsigned char hash_table[] = | |
1443 | { | |
1444 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1445 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1446 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1447 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1448 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1449 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1450 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1451 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1452 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1453 | 62, 62, 62, 62, 62, 62, 62, 2, 62, 7, | |
1454 | 6, 9, 15, 30, 62, 24, 62, 62, 1, 24, | |
1455 | 7, 27, 13, 62, 19, 26, 18, 27, 1, 62, | |
1456 | 62, 62, 62, 62, 62, 62, 62, 62, | |
1457 | }; | |
1458 | return len + hash_table[str[2]] + hash_table[str[0]]; | |
1459 | } | |
c6d46f5f | 1460 | |
42680d3c FP |
1461 | struct C_stab_entry * |
1462 | in_word_set (str, len) | |
1463 | register char *str; | |
1464 | register int len; | |
1465 | { | |
1466 | ||
1467 | static struct C_stab_entry wordlist[] = | |
1468 | { | |
1469 | {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, | |
1470 | {"",}, | |
1471 | {"volatile", 0, st_C_typespec}, | |
1472 | {"",}, | |
1473 | {"long", 0, st_C_typespec}, | |
1474 | {"char", 0, st_C_typespec}, | |
1475 | {"class", C_PLPL, st_C_struct}, | |
1476 | {"",}, {"",}, {"",}, {"",}, | |
1477 | {"const", 0, st_C_typespec}, | |
1478 | {"",}, {"",}, {"",}, {"",}, | |
1479 | {"auto", 0, st_C_typespec}, | |
1480 | {"",}, {"",}, | |
1481 | {"define", 0, st_C_define}, | |
1482 | {"",}, | |
1483 | {"void", 0, st_C_typespec}, | |
1484 | {"",}, {"",}, {"",}, | |
1485 | {"extern", 0, st_C_typespec}, | |
1486 | {"static", 0, st_C_typespec}, | |
1487 | {"",}, | |
1488 | {"domain", C_STAR, st_C_struct}, | |
1489 | {"",}, | |
1490 | {"typedef", 0, st_C_typedef}, | |
1491 | {"double", 0, st_C_typespec}, | |
1492 | {"enum", 0, st_C_enum}, | |
1493 | {"",}, {"",}, {"",}, {"",}, | |
1494 | {"int", 0, st_C_typespec}, | |
1495 | {"",}, | |
1496 | {"float", 0, st_C_typespec}, | |
1497 | {"",}, {"",}, {"",}, | |
1498 | {"struct", 0, st_C_struct}, | |
1499 | {"",}, {"",}, {"",}, {"",}, | |
1500 | {"union", 0, st_C_struct}, | |
1501 | {"",}, | |
1502 | {"short", 0, st_C_typespec}, | |
1503 | {"",}, {"",}, | |
1504 | {"unsigned", 0, st_C_typespec}, | |
1505 | {"signed", 0, st_C_typespec}, | |
1506 | }; | |
1507 | ||
1508 | if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH) | |
1509 | { | |
1510 | register int key = hash (str, len); | |
1511 | ||
1512 | if (key <= MAX_HASH_VALUE && key >= MIN_HASH_VALUE) | |
1513 | { | |
1514 | register char *s = wordlist[key].name; | |
1515 | ||
1a0d8c80 | 1516 | if (*s == *str && strneq (str + 1, s + 1, len - 1)) |
42680d3c FP |
1517 | return &wordlist[key]; |
1518 | } | |
1519 | } | |
1520 | return 0; | |
c6d46f5f | 1521 | } |
42680d3c | 1522 | /*%>*/ |
c6d46f5f | 1523 | |
42680d3c FP |
1524 | enum sym_type |
1525 | C_symtype(str, len, c_ext) | |
1526 | char *str; | |
1527 | int len; | |
c6d46f5f JB |
1528 | int c_ext; |
1529 | { | |
42680d3c | 1530 | register struct C_stab_entry *se = in_word_set(str, len); |
c6d46f5f | 1531 | |
42680d3c FP |
1532 | if (se == NULL || (se->c_ext && !(c_ext & se->c_ext))) |
1533 | return st_none; | |
1534 | return se->type; | |
c6d46f5f JB |
1535 | } |
1536 | \f | |
13fde0cd | 1537 | /* |
13fde0cd RS |
1538 | * C functions are recognized using a simple finite automaton. |
1539 | * funcdef is its state variable. | |
1540 | */ | |
1541 | typedef enum | |
1542 | { | |
31d4b314 FP |
1543 | fnone, /* nothing seen */ |
1544 | ftagseen, /* function-like tag seen */ | |
b12756c8 | 1545 | fstartlist, /* just after open parenthesis */ |
31d4b314 FP |
1546 | finlist, /* in parameter list */ |
1547 | flistseen, /* after parameter list */ | |
46e4cb76 | 1548 | fignore /* before open brace */ |
13fde0cd RS |
1549 | } FUNCST; |
1550 | FUNCST funcdef; | |
1551 | ||
1552 | ||
46c145db FP |
1553 | /* |
1554 | * typedefs are recognized using a simple finite automaton. | |
13fde0cd RS |
1555 | * typeddef is its state variable. |
1556 | */ | |
1557 | typedef enum | |
1558 | { | |
31d4b314 FP |
1559 | tnone, /* nothing seen */ |
1560 | ttypedseen, /* typedef keyword seen */ | |
1561 | tinbody, /* inside typedef body */ | |
46c145db FP |
1562 | tend, /* just before typedef tag */ |
1563 | tignore /* junk after typedef tag */ | |
13fde0cd RS |
1564 | } TYPEDST; |
1565 | TYPEDST typdef; | |
1566 | ||
1567 | ||
46c145db FP |
1568 | /* |
1569 | * struct-like structures (enum, struct and union) are recognized | |
1570 | * using another simple finite automaton. `structdef' is its state | |
1571 | * variable. | |
13fde0cd RS |
1572 | */ |
1573 | typedef enum | |
1574 | { | |
1575 | snone, /* nothing seen yet */ | |
1576 | skeyseen, /* struct-like keyword seen */ | |
1577 | stagseen, /* struct-like tag seen */ | |
1578 | scolonseen, /* colon seen after struct-like tag */ | |
46e4cb76 | 1579 | sinbody /* in struct body: recognize member func defs*/ |
13fde0cd RS |
1580 | } STRUCTST; |
1581 | STRUCTST structdef; | |
46c145db | 1582 | |
13fde0cd RS |
1583 | /* |
1584 | * When structdef is stagseen, scolonseen, or sinbody, structtag is the | |
42680d3c FP |
1585 | * struct tag, and structtype is the type of the preceding struct-like |
1586 | * keyword. | |
13fde0cd | 1587 | */ |
55597f90 | 1588 | char *structtag = "<uninited>"; |
42680d3c | 1589 | enum sym_type structtype; |
13fde0cd RS |
1590 | |
1591 | /* | |
1592 | * Yet another little state machine to deal with preprocessor lines. | |
1593 | */ | |
1594 | typedef enum | |
1595 | { | |
1596 | dnone, /* nothing seen */ | |
1597 | dsharpseen, /* '#' seen as first char on line */ | |
1598 | ddefineseen, /* '#' and 'define' seen */ | |
46e4cb76 | 1599 | dignorerest /* ignore rest of line */ |
13fde0cd RS |
1600 | } DEFINEST; |
1601 | DEFINEST definedef; | |
1602 | ||
1603 | /* | |
1604 | * Set this to TRUE, and the next token considered is called a function. | |
cdc1f6a7 | 1605 | * Used only for GNU emacs's function-defining macros. |
13fde0cd RS |
1606 | */ |
1607 | logical next_token_is_func; | |
1608 | ||
1609 | /* | |
1610 | * TRUE in the rules part of a yacc file, FALSE outside (parse as C). | |
1611 | */ | |
1612 | logical yacc_rules; | |
1613 | ||
6dd5561c FP |
1614 | /* |
1615 | * consider_token () | |
1616 | * checks to see if the current token is at the start of a | |
1617 | * function, or corresponds to a typedef, or is a struct/union/enum | |
1618 | * tag. | |
1619 | * | |
1620 | * *IS_FUNC gets TRUE iff the token is a function or macro with args. | |
1621 | * C_EXT is which language we are looking at. | |
1622 | * | |
1623 | * In the future we will need some way to adjust where the end of | |
1624 | * the token is; for instance, implementing the C++ keyword | |
1625 | * `operator' properly will adjust the end of the token to be after | |
1626 | * whatever follows `operator'. | |
1627 | * | |
1628 | * Globals | |
1629 | * funcdef IN OUT | |
1630 | * structdef IN OUT | |
1631 | * definedef IN OUT | |
1632 | * typdef IN OUT | |
1633 | * next_token_is_func IN OUT | |
1634 | */ | |
1635 | ||
1636 | logical | |
55597f90 FP |
1637 | consider_token (str, len, c, c_ext, cblev, is_func) |
1638 | register char *str; /* IN: token pointer */ | |
1639 | register int len; /* IN: token length */ | |
6dd5561c | 1640 | register char c; /* IN: first char after the token */ |
6dd5561c FP |
1641 | int c_ext; /* IN: C extensions mask */ |
1642 | int cblev; /* IN: curly brace level */ | |
715b6f8c | 1643 | logical *is_func; /* OUT: function found */ |
6dd5561c | 1644 | { |
55597f90 | 1645 | enum sym_type toktype = C_symtype (str, len, c_ext); |
6dd5561c FP |
1646 | |
1647 | /* | |
1648 | * Advance the definedef state machine. | |
1649 | */ | |
1650 | switch (definedef) | |
1651 | { | |
1652 | case dnone: | |
1653 | /* We're not on a preprocessor line. */ | |
1654 | break; | |
1655 | case dsharpseen: | |
1656 | if (toktype == st_C_define) | |
1657 | { | |
1658 | definedef = ddefineseen; | |
1659 | } | |
1660 | else | |
1661 | { | |
1662 | definedef = dignorerest; | |
1663 | } | |
b9755a12 | 1664 | return FALSE; |
6dd5561c FP |
1665 | case ddefineseen: |
1666 | /* | |
ee70dba5 FP |
1667 | * Make a tag for any macro, unless it is a constant |
1668 | * and constantypedefs is FALSE. | |
6dd5561c FP |
1669 | */ |
1670 | definedef = dignorerest; | |
1671 | *is_func = (c == '('); | |
1672 | if (!*is_func && !constantypedefs) | |
b9755a12 | 1673 | return FALSE; |
6dd5561c | 1674 | else |
b9755a12 | 1675 | return TRUE; |
6dd5561c | 1676 | case dignorerest: |
b9755a12 | 1677 | return FALSE; |
6dd5561c FP |
1678 | default: |
1679 | error ("internal error: definedef value.", 0); | |
1680 | } | |
1681 | ||
1682 | /* | |
1683 | * Now typedefs | |
1684 | */ | |
1685 | switch (typdef) | |
1686 | { | |
1687 | case tnone: | |
1688 | if (toktype == st_C_typedef) | |
1689 | { | |
1690 | if (typedefs) | |
1691 | typdef = ttypedseen; | |
1692 | funcdef = fnone; | |
b9755a12 | 1693 | return FALSE; |
6dd5561c FP |
1694 | } |
1695 | break; | |
1696 | case ttypedseen: | |
1697 | switch (toktype) | |
1698 | { | |
1699 | case st_none: | |
1700 | case st_C_typespec: | |
1701 | typdef = tend; | |
1702 | break; | |
1703 | case st_C_struct: | |
1704 | case st_C_enum: | |
1705 | break; | |
1706 | } | |
1707 | /* Do not return here, so the structdef stuff has a chance. */ | |
1708 | break; | |
1709 | case tend: | |
1710 | switch (toktype) | |
1711 | { | |
1712 | case st_C_typespec: | |
1713 | case st_C_struct: | |
1714 | case st_C_enum: | |
b9755a12 | 1715 | return FALSE; |
6dd5561c | 1716 | } |
b9755a12 | 1717 | return TRUE; |
6dd5561c FP |
1718 | } |
1719 | ||
1720 | /* | |
1721 | * This structdef business is currently only invoked when cblev==0. | |
1722 | * It should be recursively invoked whatever the curly brace level, | |
1723 | * and a stack of states kept, to allow for definitions of structs | |
1724 | * within structs. | |
1725 | * | |
1726 | * This structdef business is NOT invoked when we are ctags and the | |
1727 | * file is plain C. This is because a struct tag may have the same | |
1728 | * name as another tag, and this loses with ctags. | |
1729 | * | |
1730 | * This if statement deals with the typdef state machine as | |
1731 | * follows: if typdef==ttypedseen and token is struct/union/class/enum, | |
b9755a12 | 1732 | * return FALSE. All the other code here is for the structdef |
6dd5561c FP |
1733 | * state machine. |
1734 | */ | |
1735 | switch (toktype) | |
1736 | { | |
1737 | case st_C_struct: | |
1738 | case st_C_enum: | |
1739 | if (typdef == ttypedseen | |
1740 | || (typedefs_and_cplusplus && cblev == 0 && structdef == snone)) | |
1741 | { | |
1742 | structdef = skeyseen; | |
1743 | structtype = toktype; | |
1744 | } | |
b9755a12 | 1745 | return FALSE; |
6dd5561c FP |
1746 | } |
1747 | if (structdef == skeyseen) | |
1748 | { | |
55597f90 FP |
1749 | /* Save the tag for struct/union/class, for functions that may be |
1750 | defined inside. */ | |
6dd5561c | 1751 | if (structtype == st_C_struct) |
55597f90 | 1752 | structtag = savenstr (str, len); |
6dd5561c | 1753 | else |
55597f90 | 1754 | structtag = "<enum>"; |
6dd5561c | 1755 | structdef = stagseen; |
b9755a12 | 1756 | return TRUE; |
6dd5561c FP |
1757 | } |
1758 | ||
1759 | /* Avoid entering funcdef stuff if typdef is going on. */ | |
1760 | if (typdef != tnone) | |
1761 | { | |
1762 | definedef = dnone; | |
b9755a12 | 1763 | return FALSE; |
6dd5561c FP |
1764 | } |
1765 | ||
715b6f8c | 1766 | /* Detect GNU macros. */ |
6dd5561c | 1767 | if (definedef == dnone) |
55597f90 | 1768 | if (strneq (str, "DEFUN", 5) /* Used in emacs */ |
715b6f8c FP |
1769 | #if FALSE |
1770 | These are defined inside C functions, so currently they | |
1771 | are not met anyway. | |
55597f90 FP |
1772 | || strneq (str, "EXFUN", 5) /* Used in glibc */ |
1773 | || strneq (str, "DEFVAR_", 7) /* Used in emacs */ | |
715b6f8c | 1774 | #endif |
55597f90 FP |
1775 | || strneq (str, "SYSCALL", 7) /* Used in glibc (mach) */ |
1776 | || strneq (str, "ENTRY", 5) /* Used in glibc */ | |
1777 | || strneq (str, "PSEUDO", 6)) /* Used in glibc */ | |
715b6f8c FP |
1778 | |
1779 | { | |
1780 | next_token_is_func = TRUE; | |
b9755a12 | 1781 | return FALSE; |
715b6f8c | 1782 | } |
6dd5561c FP |
1783 | if (next_token_is_func) |
1784 | { | |
1785 | next_token_is_func = FALSE; | |
715b6f8c FP |
1786 | funcdef = fignore; |
1787 | *is_func = TRUE; | |
b9755a12 | 1788 | return TRUE; |
6dd5561c FP |
1789 | } |
1790 | ||
1791 | /* A function? */ | |
1792 | switch (toktype) | |
1793 | { | |
1794 | case st_C_typespec: | |
1795 | if (funcdef != finlist && funcdef != fignore) | |
1796 | funcdef = fnone; /* should be useless */ | |
b9755a12 | 1797 | return FALSE; |
6dd5561c FP |
1798 | default: |
1799 | if (funcdef == fnone) | |
1800 | { | |
1801 | funcdef = ftagseen; | |
1802 | *is_func = TRUE; | |
b9755a12 | 1803 | return TRUE; |
6dd5561c FP |
1804 | } |
1805 | } | |
1806 | ||
b9755a12 | 1807 | return FALSE; |
6dd5561c FP |
1808 | } |
1809 | ||
c6d46f5f JB |
1810 | /* |
1811 | * C_entries () | |
13fde0cd RS |
1812 | * This routine finds functions, typedefs, #define's and |
1813 | * struct/union/enum definitions in C syntax and adds them | |
c6d46f5f JB |
1814 | * to the list. |
1815 | */ | |
55597f90 FP |
1816 | typedef struct |
1817 | { | |
1818 | char *str; | |
1819 | logical named; | |
1820 | int linelen; | |
1821 | int lineno; | |
2bd88040 FP |
1822 | long linepos; |
1823 | char *buffer; | |
55597f90 FP |
1824 | } TOKEN; |
1825 | ||
1826 | #define current_lb_is_new (newndx == curndx) | |
1827 | #define switch_line_buffers() (curndx = 1 - curndx) | |
c6d46f5f | 1828 | |
13fde0cd RS |
1829 | #define curlb (lbs[curndx].lb) |
1830 | #define othlb (lbs[1-curndx].lb) | |
1831 | #define newlb (lbs[newndx].lb) | |
1832 | #define curlinepos (lbs[curndx].linepos) | |
1833 | #define othlinepos (lbs[1-curndx].linepos) | |
1834 | #define newlinepos (lbs[newndx].linepos) | |
1835 | ||
c6d46f5f | 1836 | #define CNL_SAVE_DEFINEDEF \ |
13fde0cd | 1837 | do { \ |
55597f90 | 1838 | curlinepos = charno; \ |
c6d46f5f | 1839 | lineno++; \ |
13fde0cd RS |
1840 | charno += readline (&curlb, inf); \ |
1841 | lp = curlb.buffer; \ | |
1842 | quotednl = FALSE; \ | |
1843 | newndx = curndx; \ | |
b9755a12 | 1844 | } while (0) |
c6d46f5f JB |
1845 | |
1846 | #define CNL \ | |
13fde0cd | 1847 | do { \ |
c6d46f5f | 1848 | CNL_SAVE_DEFINEDEF; \ |
55597f90 FP |
1849 | if (token_saved) \ |
1850 | { \ | |
1851 | tok = savetok; \ | |
1852 | token_saved = FALSE; \ | |
1853 | } \ | |
c6d46f5f | 1854 | definedef = dnone; \ |
b9755a12 | 1855 | } while (0) |
13fde0cd | 1856 | |
2bd88040 FP |
1857 | #define make_tag(isfun) pfnote (savestr (token_str.buffer), isfun, \ |
1858 | tok.named, tok.buffer, tok.linelen, tok.lineno, tok.linepos) | |
c6d46f5f JB |
1859 | |
1860 | void | |
6dd5561c | 1861 | C_entries (c_ext, inf) |
b9755a12 FP |
1862 | int c_ext; /* extension of C */ |
1863 | FILE *inf; /* input file */ | |
c6d46f5f | 1864 | { |
13fde0cd | 1865 | register char c; /* latest char read; '\0' for end of line */ |
c6d46f5f | 1866 | register char *lp; /* pointer one beyond the character `c' */ |
13fde0cd | 1867 | int curndx, newndx; /* indices for current and new lb */ |
55597f90 FP |
1868 | TOKEN tok; /* latest token read */ |
1869 | register int tokoff; /* offset in line of start of current token */ | |
1870 | register int toklen; /* length of current token */ | |
591fa824 | 1871 | int cblev; /* current curly brace level */ |
b12756c8 | 1872 | int parlev; /* current parenthesis level */ |
13fde0cd RS |
1873 | logical incomm, inquote, inchar, quotednl, midtoken; |
1874 | logical cplpl; | |
55597f90 FP |
1875 | logical token_saved; /* token saved */ |
1876 | TOKEN savetok; /* token saved during preprocessor handling */ | |
c6d46f5f | 1877 | |
13fde0cd | 1878 | curndx = newndx = 0; |
c6d46f5f JB |
1879 | lineno = 0; |
1880 | charno = 0; | |
13fde0cd | 1881 | lp = curlb.buffer; |
c6d46f5f JB |
1882 | *lp = 0; |
1883 | ||
46c145db | 1884 | definedef = dnone; funcdef = fnone; typdef = tnone; structdef = snone; |
55597f90 | 1885 | next_token_is_func = yacc_rules = token_saved = FALSE; |
13fde0cd | 1886 | midtoken = inquote = inchar = incomm = quotednl = FALSE; |
591fa824 | 1887 | cblev = 0; |
b12756c8 | 1888 | parlev = 0; |
13fde0cd | 1889 | cplpl = c_ext & C_PLPL; |
c6d46f5f | 1890 | |
c6d46f5f JB |
1891 | while (!feof (inf)) |
1892 | { | |
1893 | c = *lp++; | |
c6d46f5f JB |
1894 | if (c == '\\') |
1895 | { | |
4746118a JB |
1896 | /* If we're at the end of the line, the next character is a |
1897 | '\0'; don't skip it, because it's the thing that tells us | |
1898 | to read the next line. */ | |
13fde0cd | 1899 | if (*lp == '\0') |
99e0a2e0 | 1900 | { |
13fde0cd | 1901 | quotednl = TRUE; |
99e0a2e0 RS |
1902 | continue; |
1903 | } | |
1e134a5f | 1904 | lp++; |
c6d46f5f JB |
1905 | c = ' '; |
1906 | } | |
1907 | else if (incomm) | |
1908 | { | |
13fde0cd | 1909 | switch (c) |
c6d46f5f | 1910 | { |
13fde0cd RS |
1911 | case '*': |
1912 | if (*lp == '/') | |
1913 | { | |
1914 | c = *lp++; | |
1915 | incomm = FALSE; | |
1916 | } | |
1917 | break; | |
1918 | case '\0': | |
1919 | /* Newlines inside comments do not end macro definitions in | |
1920 | traditional cpp. */ | |
1921 | CNL_SAVE_DEFINEDEF; | |
1922 | break; | |
c6d46f5f | 1923 | } |
13fde0cd | 1924 | continue; |
c6d46f5f JB |
1925 | } |
1926 | else if (inquote) | |
1927 | { | |
13fde0cd RS |
1928 | switch (c) |
1929 | { | |
1930 | case '"': | |
1931 | inquote = FALSE; | |
1932 | break; | |
1933 | case '\0': | |
42680d3c | 1934 | /* Newlines inside strings do not end macro definitions |
13fde0cd RS |
1935 | in traditional cpp, even though compilers don't |
1936 | usually accept them. */ | |
1937 | CNL_SAVE_DEFINEDEF; | |
1938 | break; | |
1939 | } | |
1940 | continue; | |
c6d46f5f JB |
1941 | } |
1942 | else if (inchar) | |
1943 | { | |
42680d3c FP |
1944 | switch (c) |
1945 | { | |
1946 | case '\0': | |
1947 | /* Hmmm, something went wrong. */ | |
1948 | CNL; | |
1949 | /* FALLTHRU */ | |
1950 | case '\'': | |
46c145db | 1951 | inchar = FALSE; |
42680d3c FP |
1952 | break; |
1953 | } | |
c6d46f5f JB |
1954 | continue; |
1955 | } | |
13fde0cd | 1956 | else |
c6d46f5f JB |
1957 | switch (c) |
1958 | { | |
1959 | case '"': | |
1960 | inquote = TRUE; | |
b12756c8 FP |
1961 | if (funcdef != finlist && funcdef != fignore) |
1962 | funcdef = fnone; | |
c6d46f5f JB |
1963 | continue; |
1964 | case '\'': | |
1965 | inchar = TRUE; | |
b12756c8 FP |
1966 | if (funcdef != finlist && funcdef != fignore) |
1967 | funcdef = fnone; | |
c6d46f5f JB |
1968 | continue; |
1969 | case '/': | |
1970 | if (*lp == '*') | |
1971 | { | |
1972 | lp++; | |
1973 | incomm = TRUE; | |
13fde0cd | 1974 | continue; |
c6d46f5f | 1975 | } |
13fde0cd | 1976 | else if (cplpl && *lp == '/') |
c6d46f5f | 1977 | { |
daa37602 JB |
1978 | c = 0; |
1979 | break; | |
c6d46f5f | 1980 | } |
b12756c8 FP |
1981 | else |
1982 | break; | |
13fde0cd RS |
1983 | case '%': |
1984 | if ((c_ext & YACC) && *lp == '%') | |
1985 | { | |
1986 | /* entering or exiting rules section in yacc file */ | |
1987 | lp++; | |
1988 | definedef = dnone; funcdef = fnone; | |
46c145db | 1989 | typdef = tnone; structdef = snone; |
13fde0cd RS |
1990 | next_token_is_func = FALSE; |
1991 | midtoken = inquote = inchar = incomm = quotednl = FALSE; | |
591fa824 | 1992 | cblev = 0; |
13fde0cd RS |
1993 | yacc_rules = !yacc_rules; |
1994 | continue; | |
591fa824 | 1995 | } |
b12756c8 FP |
1996 | else |
1997 | break; | |
c6d46f5f | 1998 | case '#': |
ee70dba5 FP |
1999 | if (definedef == dnone) |
2000 | { | |
2001 | char *cp; | |
2002 | logical cpptoken = TRUE; | |
2003 | ||
2004 | /* Look back on this line. If all blanks, or nonblanks | |
2005 | followed by an end of comment, this is a preprocessor | |
2006 | token. */ | |
2007 | for (cp = newlb.buffer; cp < lp-1; cp++) | |
2008 | if (!iswhite (*cp)) | |
2009 | { | |
2010 | if (*cp == '*' && *(cp+1) == '/') | |
2011 | { | |
2012 | cp++; | |
2013 | cpptoken = TRUE; | |
2014 | } | |
2015 | else | |
2016 | cpptoken = FALSE; | |
2017 | } | |
2018 | if (cpptoken) | |
2019 | definedef = dsharpseen; | |
2020 | } /* if (definedef == dnone) */ | |
2021 | ||
c6d46f5f | 2022 | continue; |
13fde0cd | 2023 | } /* switch (c) */ |
c6d46f5f | 2024 | |
c6d46f5f | 2025 | |
591fa824 | 2026 | /* Consider token only if some complicated conditions are satisfied. */ |
ee70dba5 FP |
2027 | if ((definedef != dnone |
2028 | || (cblev == 0 && structdef != scolonseen) | |
591fa824 | 2029 | || (cblev == 1 && cplpl && structdef == sinbody)) |
46c145db | 2030 | && typdef != tignore |
13fde0cd | 2031 | && definedef != dignorerest |
ee70dba5 | 2032 | && funcdef != finlist) |
c6d46f5f JB |
2033 | { |
2034 | if (midtoken) | |
2035 | { | |
2036 | if (endtoken (c)) | |
2037 | { | |
b12756c8 | 2038 | if (cplpl && c == ':' && *lp == ':' && begtoken(*(lp + 1))) |
c6d46f5f JB |
2039 | { |
2040 | /* | |
ee70dba5 FP |
2041 | * This handles :: in the middle, but not at the |
2042 | * beginning of an identifier. | |
c6d46f5f JB |
2043 | */ |
2044 | lp += 2; | |
2045 | toklen += 3; | |
2046 | } | |
2047 | else | |
2048 | { | |
fe0b3356 | 2049 | logical is_func = FALSE; |
c6d46f5f | 2050 | |
13fde0cd | 2051 | if (yacc_rules |
55597f90 FP |
2052 | || consider_token (newlb.buffer + tokoff, toklen, |
2053 | c, c_ext, cblev, &is_func)) | |
c6d46f5f | 2054 | { |
99e0a2e0 | 2055 | if (structdef == sinbody |
fe0b3356 FP |
2056 | && definedef == dnone |
2057 | && is_func) | |
2058 | /* function defined in C++ class body */ | |
2059 | { | |
2bd88040 FP |
2060 | int strsize = strlen(structtag) + 2 + toklen + 1; |
2061 | while (token_str.size < strsize) | |
2062 | { | |
2063 | token_str.size *= 2; | |
2064 | xrealloc (token_str.buffer, token_str.size); | |
2065 | } | |
2066 | strcpy (token_str.buffer, structtag); | |
2067 | strcat (token_str.buffer, "::"); | |
2068 | strncat (token_str.buffer, | |
2069 | newlb.buffer+tokoff, toklen); | |
ee70dba5 | 2070 | tok.named = TRUE; |
c6d46f5f JB |
2071 | } |
2072 | else | |
2073 | { | |
2bd88040 FP |
2074 | while (token_str.size < toklen + 1) |
2075 | { | |
2076 | token_str.size *= 2; | |
2077 | xrealloc (token_str.buffer, token_str.size); | |
2078 | } | |
2079 | strncpy (token_str.buffer, | |
2080 | newlb.buffer+tokoff, toklen); | |
2081 | token_str.buffer[toklen] = '\0'; | |
55597f90 FP |
2082 | if (structdef == stagseen |
2083 | || typdef == tend | |
2084 | || (is_func | |
2085 | && definedef == dignorerest)) /* macro */ | |
2086 | tok.named = TRUE; | |
2087 | else | |
2088 | tok.named = FALSE; | |
c6d46f5f | 2089 | } |
55597f90 FP |
2090 | tok.lineno = lineno; |
2091 | tok.linelen = tokoff + toklen + 1; | |
2bd88040 FP |
2092 | tok.buffer = newlb.buffer; |
2093 | tok.linepos = newlinepos; | |
fe0b3356 | 2094 | |
b12756c8 FP |
2095 | if (definedef == dnone |
2096 | && (funcdef == ftagseen | |
2097 | || structdef == stagseen | |
2098 | || typdef == tend)) | |
13fde0cd | 2099 | { |
55597f90 FP |
2100 | if (current_lb_is_new) |
2101 | switch_line_buffers (); | |
13fde0cd RS |
2102 | } |
2103 | else | |
2bd88040 | 2104 | make_tag (is_func); |
c6d46f5f JB |
2105 | } |
2106 | midtoken = FALSE; | |
2107 | } | |
13fde0cd | 2108 | } /* if (endtoken (c)) */ |
c6d46f5f | 2109 | else if (intoken (c)) |
13fde0cd RS |
2110 | { |
2111 | toklen++; | |
2112 | continue; | |
2113 | } | |
2114 | } /* if (midtoken) */ | |
c6d46f5f JB |
2115 | else if (begtoken (c)) |
2116 | { | |
b12756c8 | 2117 | switch (definedef) |
13fde0cd | 2118 | { |
b12756c8 FP |
2119 | case dnone: |
2120 | switch (funcdef) | |
2121 | { | |
2122 | case fstartlist: | |
2123 | funcdef = finlist; | |
2124 | continue; | |
2125 | case flistseen: | |
2bd88040 | 2126 | make_tag (TRUE); |
b12756c8 FP |
2127 | funcdef = fignore; |
2128 | break; | |
2129 | case ftagseen: | |
2130 | funcdef = fnone; | |
2131 | break; | |
2132 | } | |
2133 | if (structdef == stagseen) | |
2134 | structdef = snone; | |
13fde0cd | 2135 | break; |
b12756c8 | 2136 | case dsharpseen: |
4b533b5b FP |
2137 | savetok = tok; |
2138 | token_saved = TRUE; | |
13fde0cd | 2139 | } |
13fde0cd RS |
2140 | if (!yacc_rules || lp == newlb.buffer + 1) |
2141 | { | |
2142 | tokoff = lp - 1 - newlb.buffer; | |
2143 | toklen = 1; | |
2144 | midtoken = TRUE; | |
2145 | } | |
2146 | continue; | |
4b533b5b | 2147 | } /* if (begtoken) */ |
13fde0cd RS |
2148 | } /* if must look at token */ |
2149 | ||
2150 | ||
2151 | /* Detect end of line, colon, comma, semicolon and various braces | |
b12756c8 | 2152 | after having handled a token.*/ |
13fde0cd | 2153 | switch (c) |
1e134a5f | 2154 | { |
13fde0cd | 2155 | case ':': |
b12756c8 FP |
2156 | if (definedef != dnone) |
2157 | break; | |
13fde0cd RS |
2158 | if (structdef == stagseen) |
2159 | structdef = scolonseen; | |
b12756c8 FP |
2160 | else |
2161 | switch (funcdef) | |
2162 | { | |
2163 | case ftagseen: | |
2164 | if (yacc_rules) | |
2165 | { | |
2bd88040 | 2166 | make_tag (FALSE); |
b12756c8 FP |
2167 | funcdef = fignore; |
2168 | } | |
2169 | break; | |
2170 | case fstartlist: | |
2171 | funcdef = fnone; | |
2172 | break; | |
2173 | } | |
13fde0cd RS |
2174 | break; |
2175 | case ';': | |
b12756c8 FP |
2176 | if (definedef != dnone) |
2177 | break; | |
46c145db FP |
2178 | if (cblev == 0) |
2179 | switch (typdef) | |
2180 | { | |
2181 | case tend: | |
2bd88040 | 2182 | make_tag (FALSE); |
46c145db FP |
2183 | /* FALLTHRU */ |
2184 | default: | |
2185 | typdef = tnone; | |
2186 | } | |
31d4b314 FP |
2187 | if (funcdef != fignore) |
2188 | funcdef = fnone; | |
46c145db FP |
2189 | if (structdef == stagseen) |
2190 | structdef = snone; | |
2191 | break; | |
13fde0cd | 2192 | case ',': |
46c145db FP |
2193 | if (definedef != dnone) |
2194 | break; | |
2195 | if (funcdef != finlist && funcdef != fignore) | |
2196 | funcdef = fnone; | |
2197 | if (structdef == stagseen) | |
2198 | structdef = snone; | |
2199 | break; | |
13fde0cd | 2200 | case '[': |
b12756c8 FP |
2201 | if (definedef != dnone) |
2202 | break; | |
46c145db FP |
2203 | if (cblev == 0 && typdef == tend) |
2204 | { | |
2205 | typdef = tignore; | |
2bd88040 | 2206 | make_tag (FALSE); |
46c145db FP |
2207 | break; |
2208 | } | |
31d4b314 | 2209 | if (funcdef != finlist && funcdef != fignore) |
13fde0cd RS |
2210 | funcdef = fnone; |
2211 | if (structdef == stagseen) | |
2212 | structdef = snone; | |
2213 | break; | |
2214 | case '(': | |
b12756c8 FP |
2215 | if (definedef != dnone) |
2216 | break; | |
13fde0cd | 2217 | switch (funcdef) |
57e83cfe | 2218 | { |
ee70dba5 FP |
2219 | case fnone: |
2220 | switch (typdef) | |
2221 | { | |
2222 | case ttypedseen: | |
2223 | case tend: | |
2224 | /* Make sure that the next char is not a '*'. | |
2225 | This handles constructs like: | |
2226 | typedef void OperatorFun (int fun); */ | |
2227 | if (*lp != '*') | |
2228 | { | |
2229 | typdef = tignore; | |
2bd88040 | 2230 | make_tag (FALSE); |
ee70dba5 FP |
2231 | } |
2232 | break; | |
2233 | } /* switch (typdef) */ | |
2234 | break; | |
13fde0cd | 2235 | case ftagseen: |
b12756c8 | 2236 | funcdef = fstartlist; |
13fde0cd | 2237 | break; |
13fde0cd | 2238 | case flistseen: |
b12756c8 | 2239 | funcdef = finlist; |
13fde0cd | 2240 | break; |
57e83cfe | 2241 | } |
b12756c8 | 2242 | parlev++; |
13fde0cd RS |
2243 | break; |
2244 | case ')': | |
b12756c8 FP |
2245 | if (definedef != dnone) |
2246 | break; | |
2247 | if (--parlev == 0) | |
2248 | { | |
2249 | switch (funcdef) | |
2250 | { | |
2251 | case fstartlist: | |
2252 | case finlist: | |
2253 | funcdef = flistseen; | |
2254 | break; | |
2255 | } | |
46c145db FP |
2256 | if (cblev == 0 && typdef == tend) |
2257 | { | |
2258 | typdef = tignore; | |
2bd88040 | 2259 | make_tag (FALSE); |
46c145db | 2260 | } |
b12756c8 FP |
2261 | } |
2262 | else if (parlev < 0) /* can happen due to ill-conceived #if's. */ | |
2263 | parlev = 0; | |
13fde0cd RS |
2264 | break; |
2265 | case '{': | |
b12756c8 FP |
2266 | if (definedef != dnone) |
2267 | break; | |
13fde0cd RS |
2268 | if (typdef == ttypedseen) |
2269 | typdef = tinbody; | |
2270 | switch (structdef) | |
2271 | { | |
2272 | case skeyseen: /* unnamed struct */ | |
55597f90 | 2273 | structtag = "_anonymous_"; |
13fde0cd RS |
2274 | structdef = sinbody; |
2275 | break; | |
2276 | case stagseen: | |
2277 | case scolonseen: /* named struct */ | |
2278 | structdef = sinbody; | |
2bd88040 | 2279 | make_tag (FALSE); |
13fde0cd RS |
2280 | break; |
2281 | } | |
31d4b314 FP |
2282 | switch (funcdef) |
2283 | { | |
2284 | case flistseen: | |
2bd88040 | 2285 | make_tag (TRUE); |
31d4b314 FP |
2286 | /* FALLTHRU */ |
2287 | case fignore: | |
2288 | funcdef = fnone; | |
46c145db FP |
2289 | break; |
2290 | case fnone: | |
715b6f8c | 2291 | /* Neutralize `extern "C" {' grot and look inside structs. */ |
46c145db | 2292 | if (cblev == 0 && structdef == snone && typdef == tnone) |
715b6f8c | 2293 | cblev = -1; |
31d4b314 | 2294 | } |
591fa824 | 2295 | cblev++; |
31d4b314 | 2296 | break; |
13fde0cd | 2297 | case '*': |
b12756c8 FP |
2298 | if (definedef != dnone) |
2299 | break; | |
2300 | if (funcdef == fstartlist) | |
2301 | funcdef = fnone; /* avoid tagging `foo' in `foo (*bar()) ()' */ | |
13fde0cd RS |
2302 | break; |
2303 | case '}': | |
b12756c8 FP |
2304 | if (definedef != dnone) |
2305 | break; | |
13fde0cd | 2306 | if (!noindentypedefs && lp == newlb.buffer + 1) |
b12756c8 FP |
2307 | { |
2308 | cblev = 0; /* reset curly brace level if first column */ | |
2309 | parlev = 0; /* also reset paren level, just in case... */ | |
2310 | } | |
591fa824 RS |
2311 | else if (cblev > 0) |
2312 | cblev--; | |
2313 | if (cblev == 0) | |
13fde0cd RS |
2314 | { |
2315 | if (typdef == tinbody) | |
2316 | typdef = tend; | |
55597f90 FP |
2317 | #if FALSE /* too risky */ |
2318 | if (structdef == sinbody) | |
2319 | free (structtag); | |
2320 | #endif | |
13fde0cd | 2321 | structdef = snone; |
55597f90 | 2322 | structtag = "<error>"; |
13fde0cd RS |
2323 | } |
2324 | break; | |
b12756c8 | 2325 | case '=': |
42680d3c FP |
2326 | case '#': case '+': case '-': case '~': case '&': case '%': case '/': |
2327 | case '|': case '^': case '!': case '<': case '>': case '.': case '?': | |
b12756c8 FP |
2328 | if (definedef != dnone) |
2329 | break; | |
2330 | /* These surely cannot follow a function tag. */ | |
2331 | if (funcdef != finlist && funcdef != fignore) | |
2332 | funcdef = fnone; | |
2333 | break; | |
13fde0cd RS |
2334 | case '\0': |
2335 | /* If a macro spans multiple lines don't reset its state. */ | |
2336 | if (quotednl) | |
2337 | CNL_SAVE_DEFINEDEF; | |
2338 | else | |
2339 | CNL; | |
2340 | break; | |
2341 | } /* switch (c) */ | |
2342 | ||
2343 | } /* while not eof */ | |
c6d46f5f | 2344 | } |
b9755a12 FP |
2345 | |
2346 | /* | |
2347 | * Process either a C++ file or a C file depending on the setting | |
2348 | * of a global flag. | |
2349 | */ | |
2350 | void | |
2351 | default_C_entries (inf) | |
2352 | FILE *inf; | |
2353 | { | |
2354 | C_entries (cplusplus ? C_PLPL : 0, inf); | |
2355 | } | |
2356 | ||
2357 | /* Always do C++. */ | |
2358 | void | |
2359 | Cplusplus_entries (inf) | |
2360 | FILE *inf; | |
2361 | { | |
2362 | C_entries (C_PLPL, inf); | |
2363 | } | |
2364 | ||
2365 | /* Always do C*. */ | |
2366 | void | |
2367 | Cstar_entries (inf) | |
2368 | FILE *inf; | |
2369 | { | |
2370 | C_entries (C_STAR, inf); | |
2371 | } | |
2372 | ||
2373 | /* Always do Yacc. */ | |
2374 | void | |
2375 | Yacc_entries (inf) | |
2376 | FILE *inf; | |
2377 | { | |
2378 | C_entries (YACC, inf); | |
2379 | } | |
6dd5561c FP |
2380 | \f |
2381 | /* Fortran parsing */ | |
c6d46f5f | 2382 | |
6dd5561c | 2383 | char *dbp; |
c6d46f5f JB |
2384 | |
2385 | logical | |
6dd5561c FP |
2386 | tail (cp) |
2387 | char *cp; | |
c6d46f5f | 2388 | { |
6dd5561c | 2389 | register int len = 0; |
c6d46f5f | 2390 | |
6dd5561c FP |
2391 | while (*cp && (*cp | ' ') == (dbp[len] | ' ')) |
2392 | cp++, len++; | |
2393 | if (*cp == 0) | |
c6d46f5f | 2394 | { |
6dd5561c | 2395 | dbp += len; |
b9755a12 | 2396 | return TRUE; |
c6d46f5f | 2397 | } |
b9755a12 | 2398 | return FALSE; |
6dd5561c | 2399 | } |
13fde0cd | 2400 | |
6dd5561c FP |
2401 | void |
2402 | takeprec () | |
2403 | { | |
2404 | while (isspace (*dbp)) | |
2405 | dbp++; | |
2406 | if (*dbp != '*') | |
2407 | return; | |
2408 | dbp++; | |
2409 | while (isspace (*dbp)) | |
2410 | dbp++; | |
0c1fd2e3 FP |
2411 | if (tail ("(*)")) |
2412 | return; | |
6dd5561c | 2413 | if (!isdigit (*dbp)) |
c6d46f5f | 2414 | { |
6dd5561c FP |
2415 | --dbp; /* force failure */ |
2416 | return; | |
c6d46f5f | 2417 | } |
6dd5561c FP |
2418 | do |
2419 | dbp++; | |
2420 | while (isdigit (*dbp)); | |
2421 | } | |
13fde0cd | 2422 | |
6dd5561c FP |
2423 | void |
2424 | getit (inf) | |
2425 | FILE *inf; | |
2426 | { | |
2427 | register char *cp; | |
13fde0cd | 2428 | |
6dd5561c FP |
2429 | while (isspace (*dbp)) |
2430 | dbp++; | |
2431 | if (*dbp == '\0') | |
c6d46f5f | 2432 | { |
6dd5561c FP |
2433 | lineno++; |
2434 | linecharno = charno; | |
2435 | charno += readline (&lb, inf); | |
2436 | dbp = lb.buffer; | |
2437 | if (dbp[5] != '&') | |
2438 | return; | |
2439 | dbp += 6; | |
2440 | while (isspace (*dbp)) | |
2441 | dbp++; | |
c6d46f5f | 2442 | } |
6dd5561c FP |
2443 | if (!isalpha (*dbp) |
2444 | && *dbp != '_' | |
2445 | && *dbp != '$') | |
2446 | return; | |
2447 | for (cp = dbp + 1; | |
2448 | (*cp | |
2449 | && (isalpha (*cp) || isdigit (*cp) || (*cp == '_') || (*cp == '$'))); | |
2450 | cp++) | |
2451 | continue; | |
55597f90 | 2452 | pfnote (savenstr (dbp, cp-dbp), TRUE, FALSE, lb.buffer, |
6dd5561c | 2453 | cp - lb.buffer + 1, lineno, linecharno); |
c6d46f5f | 2454 | } |
c6d46f5f | 2455 | |
b9755a12 | 2456 | void |
6dd5561c FP |
2457 | Fortran_functions (inf) |
2458 | FILE *inf; | |
c6d46f5f JB |
2459 | { |
2460 | lineno = 0; | |
2461 | charno = 0; | |
c6d46f5f | 2462 | |
6dd5561c | 2463 | while (!feof (inf)) |
c6d46f5f JB |
2464 | { |
2465 | lineno++; | |
2466 | linecharno = charno; | |
6dd5561c | 2467 | charno += readline (&lb, inf); |
c6d46f5f JB |
2468 | dbp = lb.buffer; |
2469 | if (*dbp == '%') | |
2470 | dbp++; /* Ratfor escape to fortran */ | |
2471 | while (isspace (*dbp)) | |
2472 | dbp++; | |
2473 | if (*dbp == 0) | |
2474 | continue; | |
2475 | switch (*dbp | ' ') | |
2476 | { | |
2477 | case 'i': | |
2478 | if (tail ("integer")) | |
2479 | takeprec (); | |
2480 | break; | |
2481 | case 'r': | |
2482 | if (tail ("real")) | |
2483 | takeprec (); | |
2484 | break; | |
2485 | case 'l': | |
2486 | if (tail ("logical")) | |
2487 | takeprec (); | |
2488 | break; | |
2489 | case 'c': | |
2490 | if (tail ("complex") || tail ("character")) | |
2491 | takeprec (); | |
2492 | break; | |
2493 | case 'd': | |
2494 | if (tail ("double")) | |
2495 | { | |
2496 | while (isspace (*dbp)) | |
2497 | dbp++; | |
2498 | if (*dbp == 0) | |
2499 | continue; | |
2500 | if (tail ("precision")) | |
2501 | break; | |
2502 | continue; | |
2503 | } | |
2504 | break; | |
2505 | } | |
2506 | while (isspace (*dbp)) | |
2507 | dbp++; | |
2508 | if (*dbp == 0) | |
2509 | continue; | |
2510 | switch (*dbp | ' ') | |
2511 | { | |
2512 | case 'f': | |
2513 | if (tail ("function")) | |
6dd5561c | 2514 | getit (inf); |
c6d46f5f JB |
2515 | continue; |
2516 | case 's': | |
2517 | if (tail ("subroutine")) | |
6dd5561c | 2518 | getit (inf); |
c6d46f5f | 2519 | continue; |
8a6c8bcf RS |
2520 | case 'e': |
2521 | if (tail ("entry")) | |
6dd5561c | 2522 | getit (inf); |
8a6c8bcf | 2523 | continue; |
c6d46f5f JB |
2524 | case 'p': |
2525 | if (tail ("program")) | |
2526 | { | |
6dd5561c | 2527 | getit (inf); |
c6d46f5f JB |
2528 | continue; |
2529 | } | |
2530 | if (tail ("procedure")) | |
6dd5561c | 2531 | getit (inf); |
c6d46f5f JB |
2532 | continue; |
2533 | } | |
2534 | } | |
c6d46f5f | 2535 | } |
6dd5561c FP |
2536 | \f |
2537 | /* | |
2538 | * Bob Weiner, Motorola Inc., 4/3/94 | |
2539 | * Unix and microcontroller assembly tag handling | |
2540 | * look for '^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]' | |
2541 | */ | |
c6d46f5f | 2542 | void |
6dd5561c FP |
2543 | Asm_labels (inf) |
2544 | FILE *inf; | |
c6d46f5f JB |
2545 | { |
2546 | register char *cp; | |
c6d46f5f JB |
2547 | |
2548 | lineno = 0; | |
2549 | charno = 0; | |
c6d46f5f | 2550 | |
6dd5561c | 2551 | while (!feof (inf)) |
c6d46f5f JB |
2552 | { |
2553 | lineno++; | |
2554 | linecharno = charno; | |
6dd5561c FP |
2555 | charno += readline (&lb, inf); |
2556 | cp = lb.buffer; | |
2557 | ||
2558 | /* If first char is alphabetic or one of [_.$], test for colon | |
2559 | following identifier. */ | |
2560 | if (isalpha (*cp) || *cp == '_' || *cp == '.' || *cp == '$') | |
2561 | { | |
2562 | /* Read past label. */ | |
2563 | cp++; | |
2564 | while (isalnum (*cp) || *cp == '_' || *cp == '.' || *cp == '$') | |
2565 | cp++; | |
2566 | if (*cp == ':' || isspace (*cp)) | |
2567 | { | |
2568 | /* Found end of label, so copy it and add it to the table. */ | |
55597f90 FP |
2569 | pfnote (savenstr (lb.buffer, cp-lb.buffer), TRUE, FALSE, |
2570 | lb.buffer, cp - lb.buffer + 1, lineno, linecharno); | |
6dd5561c FP |
2571 | } |
2572 | } | |
c6d46f5f JB |
2573 | } |
2574 | } | |
2575 | \f | |
2576 | /* Added by Mosur Mohan, 4/22/88 */ | |
2577 | /* Pascal parsing */ | |
2578 | ||
2579 | #define GET_NEW_LINE \ | |
2580 | { \ | |
2581 | linecharno = charno; lineno++; \ | |
2582 | charno += 1 + readline (&lb, inf); \ | |
2583 | dbp = lb.buffer; \ | |
2584 | } | |
2585 | ||
aab1fdae FP |
2586 | /* |
2587 | * Locates tags for procedures & functions. Doesn't do any type- or | |
2588 | * var-definitions. It does look for the keyword "extern" or | |
2589 | * "forward" immediately following the procedure statement; if found, | |
2590 | * the tag is skipped. | |
c6d46f5f | 2591 | */ |
c6d46f5f | 2592 | void |
6dd5561c FP |
2593 | Pascal_functions (inf) |
2594 | FILE *inf; | |
c6d46f5f JB |
2595 | { |
2596 | struct linebuffer tline; /* mostly copied from C_entries */ | |
2597 | long save_lcno; | |
2598 | int save_lineno; | |
2599 | char c, *cp; | |
55597f90 | 2600 | char *nambuf; |
c6d46f5f JB |
2601 | |
2602 | logical /* each of these flags is TRUE iff: */ | |
b9755a12 | 2603 | incomment, /* point is inside a comment */ |
c6d46f5f JB |
2604 | inquote, /* point is inside '..' string */ |
2605 | get_tagname, /* point is after PROCEDURE/FUNCTION */ | |
2606 | /* keyword, so next item = potential tag */ | |
2607 | found_tag, /* point is after a potential tag */ | |
2608 | inparms, /* point is within parameter-list */ | |
2609 | verify_tag; /* point has passed the parm-list, so the */ | |
2610 | /* next token will determine whether */ | |
2611 | /* this is a FORWARD/EXTERN to be */ | |
2612 | /* ignored, or whether it is a real tag */ | |
2613 | ||
2614 | lineno = 0; | |
2615 | charno = 0; | |
2616 | dbp = lb.buffer; | |
2617 | *dbp = 0; | |
2618 | initbuffer (&tline); | |
2619 | ||
b9755a12 | 2620 | incomment = inquote = FALSE; |
c6d46f5f JB |
2621 | found_tag = FALSE; /* have a proc name; check if extern */ |
2622 | get_tagname = FALSE; /* have found "procedure" keyword */ | |
2623 | inparms = FALSE; /* found '(' after "proc" */ | |
2624 | verify_tag = FALSE; /* check if "extern" is ahead */ | |
2625 | ||
2626 | /* long main loop to get next char */ | |
6dd5561c | 2627 | while (!feof (inf)) |
c6d46f5f JB |
2628 | { |
2629 | c = *dbp++; | |
55597f90 | 2630 | if (c == '\0') /* if end of line */ |
c6d46f5f JB |
2631 | { |
2632 | GET_NEW_LINE; | |
55597f90 | 2633 | if (*dbp == '\0') |
c6d46f5f JB |
2634 | continue; |
2635 | if (!((found_tag && verify_tag) || | |
2636 | get_tagname)) | |
2637 | c = *dbp++; /* only if don't need *dbp pointing */ | |
55597f90 FP |
2638 | /* to the beginning of the name of */ |
2639 | /* the procedure or function */ | |
c6d46f5f | 2640 | } |
b9755a12 | 2641 | if (incomment) |
c6d46f5f | 2642 | { |
b9755a12 FP |
2643 | if (c == '}') /* within { - } comments */ |
2644 | incomment = FALSE; | |
2645 | else if (c == '*' && dbp[1] == ')') /* within (* - *) comments */ | |
c6d46f5f | 2646 | { |
b9755a12 FP |
2647 | dbp++; |
2648 | incomment = FALSE; | |
c6d46f5f JB |
2649 | } |
2650 | continue; | |
2651 | } | |
2652 | else if (inquote) | |
2653 | { | |
2654 | if (c == '\'') | |
2655 | inquote = FALSE; | |
2656 | continue; | |
2657 | } | |
55597f90 | 2658 | else |
c6d46f5f JB |
2659 | switch (c) |
2660 | { | |
2661 | case '\'': | |
2662 | inquote = TRUE; /* found first quote */ | |
2663 | continue; | |
2664 | case '{': /* found open-{-comment */ | |
b9755a12 | 2665 | incomment = TRUE; |
c6d46f5f JB |
2666 | continue; |
2667 | case '(': | |
2668 | if (*dbp == '*') /* found open-(*-comment */ | |
2669 | { | |
b9755a12 | 2670 | incomment = TRUE; |
c6d46f5f JB |
2671 | dbp++; |
2672 | } | |
2673 | else if (found_tag) /* found '(' after tag, i.e., parm-list */ | |
2674 | inparms = TRUE; | |
2675 | continue; | |
2676 | case ')': /* end of parms list */ | |
2677 | if (inparms) | |
2678 | inparms = FALSE; | |
2679 | continue; | |
2680 | case ';': | |
2681 | if ((found_tag) && (!inparms)) /* end of proc or fn stmt */ | |
2682 | { | |
2683 | verify_tag = TRUE; | |
2684 | break; | |
2685 | } | |
2686 | continue; | |
2687 | } | |
2688 | if ((found_tag) && (verify_tag) && (*dbp != ' ')) | |
2689 | { | |
2690 | /* check if this is an "extern" declaration */ | |
2691 | if (*dbp == 0) | |
2692 | continue; | |
2693 | if ((*dbp == 'e') || (*dbp == 'E')) | |
2694 | { | |
2695 | if (tail ("extern")) /* superfluous, really! */ | |
2696 | { | |
2697 | found_tag = FALSE; | |
2698 | verify_tag = FALSE; | |
2699 | } | |
2700 | } | |
2701 | else if ((*dbp == 'f') || (*dbp == 'F')) | |
2702 | { | |
2703 | if (tail ("forward")) /* check for forward reference */ | |
2704 | { | |
2705 | found_tag = FALSE; | |
2706 | verify_tag = FALSE; | |
2707 | } | |
2708 | } | |
46c145db | 2709 | if ((found_tag) && (verify_tag)) /* not external proc, so make tag */ |
c6d46f5f JB |
2710 | { |
2711 | found_tag = FALSE; | |
2712 | verify_tag = FALSE; | |
55597f90 FP |
2713 | pfnote (nambuf, TRUE, FALSE, tline.buffer, |
2714 | cp - tline.buffer + 1, save_lineno, save_lcno); | |
c6d46f5f JB |
2715 | continue; |
2716 | } | |
2717 | } | |
2718 | if (get_tagname) /* grab name of proc or fn */ | |
2719 | { | |
2720 | if (*dbp == 0) | |
2721 | continue; | |
2722 | ||
2723 | /* save all values for later tagging */ | |
2724 | tline.size = lb.size; | |
2725 | strcpy (tline.buffer, lb.buffer); | |
2726 | save_lineno = lineno; | |
2727 | save_lcno = linecharno; | |
2728 | ||
2729 | /* grab block name */ | |
2730 | for (cp = dbp + 1; *cp && (!endtoken (*cp)); cp++) | |
2731 | continue; | |
55597f90 | 2732 | nambuf = savenstr (dbp, cp-dbp); |
c6d46f5f JB |
2733 | dbp = cp; /* restore dbp to e-o-token */ |
2734 | get_tagname = FALSE; | |
2735 | found_tag = TRUE; | |
2736 | continue; | |
2737 | ||
2738 | /* and proceed to check for "extern" */ | |
2739 | } | |
55597f90 | 2740 | else if (!incomment && !inquote && !found_tag) |
c6d46f5f JB |
2741 | { |
2742 | /* check for proc/fn keywords */ | |
2743 | switch (c | ' ') | |
2744 | { | |
2745 | case 'p': | |
2746 | if (tail ("rocedure")) /* c = 'p', dbp has advanced */ | |
2747 | get_tagname = TRUE; | |
2748 | continue; | |
2749 | case 'f': | |
2750 | if (tail ("unction")) | |
2751 | get_tagname = TRUE; | |
2752 | continue; | |
2753 | } | |
2754 | } | |
6dd5561c | 2755 | } /* while not eof */ |
c6d46f5f JB |
2756 | } |
2757 | \f | |
2758 | /* | |
2759 | * lisp tag functions | |
55597f90 | 2760 | * look for (def or (DEF, quote or QUOTE |
c6d46f5f | 2761 | */ |
c6d46f5f | 2762 | int |
55597f90 FP |
2763 | L_isdef (strp) |
2764 | register char *strp; | |
c6d46f5f | 2765 | { |
55597f90 FP |
2766 | return ((strp[1] == 'd' || strp[1] == 'D') |
2767 | && (strp[2] == 'e' || strp[2] == 'E') | |
2768 | && (strp[3] == 'f' || strp[3] == 'F')); | |
31d4b314 FP |
2769 | } |
2770 | ||
2771 | int | |
55597f90 FP |
2772 | L_isquote (strp) |
2773 | register char *strp; | |
2774 | { | |
2775 | return ((*(++strp) == 'q' || *strp == 'Q') | |
2776 | && (*(++strp) == 'u' || *strp == 'U') | |
2777 | && (*(++strp) == 'o' || *strp == 'O') | |
2778 | && (*(++strp) == 't' || *strp == 'T') | |
2779 | && (*(++strp) == 'e' || *strp == 'E') | |
2780 | && isspace(*(++strp))); | |
c6d46f5f JB |
2781 | } |
2782 | ||
2783 | void | |
2784 | L_getit () | |
2785 | { | |
2786 | register char *cp; | |
c6d46f5f | 2787 | |
31d4b314 FP |
2788 | if (*dbp == '\'') /* Skip prefix quote */ |
2789 | dbp++; | |
2790 | else if (*dbp == '(' && L_isquote (dbp)) /* Skip "(quote " */ | |
2791 | { | |
2792 | dbp += 7; | |
2793 | while (isspace(*dbp)) | |
2794 | dbp++; | |
2795 | } | |
55597f90 FP |
2796 | for (cp = dbp /*+1*/; |
2797 | *cp && *cp != '(' && *cp != ' ' && *cp != ')'; | |
2798 | cp++) | |
c6d46f5f | 2799 | continue; |
31d4b314 FP |
2800 | if (cp == dbp) |
2801 | return; | |
2802 | ||
55597f90 | 2803 | pfnote (savenstr (dbp, cp-dbp), TRUE, FALSE, lb.buffer, |
591fa824 | 2804 | cp - lb.buffer + 1, lineno, linecharno); |
c6d46f5f | 2805 | } |
6dd5561c FP |
2806 | |
2807 | void | |
2808 | Lisp_functions (inf) | |
2809 | FILE *inf; | |
2810 | { | |
2811 | lineno = 0; | |
2812 | charno = 0; | |
6dd5561c FP |
2813 | |
2814 | while (!feof (inf)) | |
2815 | { | |
2816 | lineno++; | |
2817 | linecharno = charno; | |
2818 | charno += readline (&lb, inf); | |
2819 | dbp = lb.buffer; | |
2820 | if (dbp[0] == '(') | |
2821 | { | |
2822 | if (L_isdef (dbp)) | |
2823 | { | |
2824 | while (!isspace (*dbp)) | |
2825 | dbp++; | |
2826 | while (isspace (*dbp)) | |
2827 | dbp++; | |
2828 | L_getit (); | |
2829 | } | |
2830 | else | |
2831 | { | |
2832 | /* Check for (foo::defmumble name-defined ... */ | |
2833 | do | |
2834 | dbp++; | |
2835 | while (*dbp && !isspace (*dbp) | |
2836 | && *dbp != ':' && *dbp != '(' && *dbp != ')'); | |
2837 | if (*dbp == ':') | |
2838 | { | |
2839 | do | |
2840 | dbp++; | |
2841 | while (*dbp == ':'); | |
2842 | ||
2843 | if (L_isdef (dbp - 1)) | |
2844 | { | |
2845 | while (!isspace (*dbp)) | |
2846 | dbp++; | |
2847 | while (isspace (*dbp)) | |
2848 | dbp++; | |
2849 | L_getit (); | |
2850 | } | |
2851 | } | |
2852 | } | |
2853 | } | |
2854 | } | |
2855 | } | |
c6d46f5f JB |
2856 | \f |
2857 | /* | |
2858 | * Scheme tag functions | |
2859 | * look for (def... xyzzy | |
2860 | * look for (def... (xyzzy | |
2861 | * look for (def ... ((...(xyzzy .... | |
2862 | * look for (set! xyzzy | |
2863 | */ | |
2864 | ||
6dd5561c | 2865 | void get_scheme (); |
c6d46f5f JB |
2866 | |
2867 | void | |
6dd5561c FP |
2868 | Scheme_functions (inf) |
2869 | FILE *inf; | |
c6d46f5f JB |
2870 | { |
2871 | lineno = 0; | |
2872 | charno = 0; | |
c6d46f5f | 2873 | |
6dd5561c | 2874 | while (!feof (inf)) |
c6d46f5f JB |
2875 | { |
2876 | lineno++; | |
2877 | linecharno = charno; | |
6dd5561c | 2878 | charno += readline (&lb, inf); |
c6d46f5f JB |
2879 | dbp = lb.buffer; |
2880 | if (dbp[0] == '(' && | |
2881 | (dbp[1] == 'D' || dbp[1] == 'd') && | |
2882 | (dbp[2] == 'E' || dbp[2] == 'e') && | |
2883 | (dbp[3] == 'F' || dbp[3] == 'f')) | |
2884 | { | |
2885 | while (!isspace (*dbp)) | |
2886 | dbp++; | |
2887 | /* Skip over open parens and white space */ | |
2888 | while (*dbp && (isspace (*dbp) || *dbp == '(')) | |
2889 | dbp++; | |
2890 | get_scheme (); | |
2891 | } | |
2892 | if (dbp[0] == '(' && | |
2893 | (dbp[1] == 'S' || dbp[1] == 's') && | |
2894 | (dbp[2] == 'E' || dbp[2] == 'e') && | |
2895 | (dbp[3] == 'T' || dbp[3] == 't') && | |
2896 | (dbp[4] == '!' || dbp[4] == '!') && | |
2897 | (isspace (dbp[5]))) | |
2898 | { | |
2899 | while (!isspace (*dbp)) | |
2900 | dbp++; | |
2901 | /* Skip over white space */ | |
2902 | while (isspace (*dbp)) | |
2903 | dbp++; | |
2904 | get_scheme (); | |
2905 | } | |
2906 | } | |
2907 | } | |
2908 | ||
6dd5561c | 2909 | void |
c6d46f5f JB |
2910 | get_scheme () |
2911 | { | |
2912 | register char *cp; | |
c6d46f5f JB |
2913 | |
2914 | if (*dbp == 0) | |
2915 | return; | |
2916 | /* Go till you get to white space or a syntactic break */ | |
55597f90 FP |
2917 | for (cp = dbp + 1; |
2918 | *cp && *cp != '(' && *cp != ')' && !isspace (*cp); | |
2919 | cp++) | |
c6d46f5f | 2920 | continue; |
55597f90 FP |
2921 | pfnote (savenstr (dbp, cp-dbp), TRUE, FALSE, |
2922 | lb.buffer, cp - lb.buffer + 1, lineno, linecharno); | |
c6d46f5f JB |
2923 | } |
2924 | \f | |
2925 | /* Find tags in TeX and LaTeX input files. */ | |
2926 | ||
2927 | /* TEX_toktab is a table of TeX control sequences that define tags. | |
2928 | Each TEX_tabent records one such control sequence. | |
2929 | CONVERT THIS TO USE THE Stab TYPE!! */ | |
c6d46f5f JB |
2930 | struct TEX_tabent |
2931 | { | |
2932 | char *name; | |
2933 | int len; | |
2934 | }; | |
2935 | ||
2936 | struct TEX_tabent *TEX_toktab = NULL; /* Table with tag tokens */ | |
2937 | ||
2938 | /* Default set of control sequences to put into TEX_toktab. | |
2939 | The value of environment var TEXTAGS is prepended to this. */ | |
2940 | ||
6dd5561c FP |
2941 | char *TEX_defenv = "\ |
2942 | :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem:typeout"; | |
c6d46f5f JB |
2943 | |
2944 | void TEX_mode (); | |
2945 | struct TEX_tabent *TEX_decode_env (); | |
2946 | void TEX_getit (); | |
2947 | int TEX_Token (); | |
2948 | ||
6dd5561c FP |
2949 | char TEX_esc = '\\'; |
2950 | char TEX_opgrp = '{'; | |
2951 | char TEX_clgrp = '}'; | |
c6d46f5f JB |
2952 | |
2953 | /* | |
2954 | * TeX/LaTeX scanning loop. | |
2955 | */ | |
c6d46f5f | 2956 | void |
6dd5561c FP |
2957 | TeX_functions (inf) |
2958 | FILE *inf; | |
c6d46f5f JB |
2959 | { |
2960 | char *lasthit; | |
2961 | ||
2962 | lineno = 0; | |
2963 | charno = 0; | |
c6d46f5f JB |
2964 | |
2965 | /* Select either \ or ! as escape character. */ | |
6dd5561c | 2966 | TEX_mode (inf); |
c6d46f5f JB |
2967 | |
2968 | /* Initialize token table once from environment. */ | |
2969 | if (!TEX_toktab) | |
2970 | TEX_toktab = TEX_decode_env ("TEXTAGS", TEX_defenv); | |
2971 | ||
6dd5561c | 2972 | while (!feof (inf)) |
d2729198 | 2973 | { /* Scan each line in file */ |
c6d46f5f JB |
2974 | lineno++; |
2975 | linecharno = charno; | |
6dd5561c | 2976 | charno += readline (&lb, inf); |
c6d46f5f JB |
2977 | dbp = lb.buffer; |
2978 | lasthit = dbp; | |
b02c5fea | 2979 | while (dbp = etags_strchr (dbp, TEX_esc)) /* Look at each esc in line */ |
8a6c8bcf RS |
2980 | { |
2981 | register int i; | |
c6d46f5f | 2982 | |
8a6c8bcf RS |
2983 | if (!*(++dbp)) |
2984 | break; | |
2985 | linecharno += dbp - lasthit; | |
c6d46f5f | 2986 | lasthit = dbp; |
8a6c8bcf RS |
2987 | i = TEX_Token (lasthit); |
2988 | if (0 <= i) | |
c6d46f5f | 2989 | { |
8a6c8bcf | 2990 | TEX_getit (lasthit, TEX_toktab[i].len); |
d2729198 | 2991 | break; /* We only save a line once */ |
c6d46f5f JB |
2992 | } |
2993 | } | |
2994 | } | |
2995 | } | |
2996 | ||
2997 | #define TEX_LESC '\\' | |
2998 | #define TEX_SESC '!' | |
2999 | #define TEX_cmt '%' | |
3000 | ||
aab1fdae FP |
3001 | /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping |
3002 | chars accordingly. */ | |
c6d46f5f | 3003 | void |
6dd5561c FP |
3004 | TEX_mode (inf) |
3005 | FILE *inf; | |
c6d46f5f JB |
3006 | { |
3007 | int c; | |
3008 | ||
6dd5561c | 3009 | while ((c = getc (inf)) != EOF) |
c6d46f5f JB |
3010 | { |
3011 | /* Skip to next line if we hit the TeX comment char. */ | |
3012 | if (c == TEX_cmt) | |
3013 | while (c != '\n') | |
6dd5561c | 3014 | c = getc (inf); |
c6d46f5f JB |
3015 | else if (c == TEX_LESC || c == TEX_SESC ) |
3016 | break; | |
3017 | } | |
3018 | ||
3019 | if (c == TEX_LESC) | |
3020 | { | |
3021 | TEX_esc = TEX_LESC; | |
3022 | TEX_opgrp = '{'; | |
3023 | TEX_clgrp = '}'; | |
3024 | } | |
3025 | else | |
3026 | { | |
3027 | TEX_esc = TEX_SESC; | |
3028 | TEX_opgrp = '<'; | |
3029 | TEX_clgrp = '>'; | |
3030 | } | |
6dd5561c | 3031 | rewind (inf); |
c6d46f5f JB |
3032 | } |
3033 | ||
aab1fdae FP |
3034 | /* Read environment and prepend it to the default string. |
3035 | Build token table. */ | |
c6d46f5f JB |
3036 | struct TEX_tabent * |
3037 | TEX_decode_env (evarname, defenv) | |
3038 | char *evarname; | |
3039 | char *defenv; | |
3040 | { | |
3041 | register char *env, *p; | |
c6d46f5f JB |
3042 | |
3043 | struct TEX_tabent *tab; | |
3044 | int size, i; | |
3045 | ||
3046 | /* Append default string to environment. */ | |
3047 | env = getenv (evarname); | |
3048 | if (!env) | |
3049 | env = defenv; | |
3050 | else | |
3051 | env = concat (env, defenv, ""); | |
3052 | ||
3053 | /* Allocate a token table */ | |
3054 | for (size = 1, p = env; p;) | |
b02c5fea | 3055 | if ((p = etags_strchr (p, ':')) && *(++p)) |
c6d46f5f | 3056 | size++; |
8a6c8bcf RS |
3057 | /* Add 1 to leave room for null terminator. */ |
3058 | tab = xnew (size + 1, struct TEX_tabent); | |
c6d46f5f JB |
3059 | |
3060 | /* Unpack environment string into token table. Be careful about */ | |
3061 | /* zero-length strings (leading ':', "::" and trailing ':') */ | |
3062 | for (i = 0; *env;) | |
3063 | { | |
b02c5fea | 3064 | p = etags_strchr (env, ':'); |
c6d46f5f JB |
3065 | if (!p) /* End of environment string. */ |
3066 | p = env + strlen (env); | |
3067 | if (p - env > 0) | |
3068 | { /* Only non-zero strings. */ | |
3069 | tab[i].name = savenstr (env, p - env); | |
3070 | tab[i].len = strlen (tab[i].name); | |
3071 | i++; | |
3072 | } | |
3073 | if (*p) | |
3074 | env = p + 1; | |
3075 | else | |
3076 | { | |
3077 | tab[i].name = NULL; /* Mark end of table. */ | |
3078 | tab[i].len = 0; | |
3079 | break; | |
3080 | } | |
3081 | } | |
3082 | return tab; | |
3083 | } | |
3084 | ||
3085 | /* Record a tag defined by a TeX command of length LEN and starting at NAME. | |
3086 | The name being defined actually starts at (NAME + LEN + 1). | |
3087 | But we seem to include the TeX command in the tag name. */ | |
c6d46f5f JB |
3088 | void |
3089 | TEX_getit (name, len) | |
3090 | char *name; | |
3091 | int len; | |
3092 | { | |
3093 | char *p = name + len; | |
c6d46f5f JB |
3094 | |
3095 | if (*name == 0) | |
3096 | return; | |
3097 | ||
3098 | /* Let tag name extend to next group close (or end of line) */ | |
3099 | while (*p && *p != TEX_clgrp) | |
3100 | p++; | |
55597f90 FP |
3101 | pfnote (savenstr (name, p-name), TRUE, FALSE, lb.buffer, |
3102 | strlen (lb.buffer), lineno, linecharno); | |
c6d46f5f JB |
3103 | } |
3104 | ||
3105 | /* If the text at CP matches one of the tag-defining TeX command names, | |
b02c5fea | 3106 | return the pointer to the first occurrence of that command in TEX_toktab. |
aab1fdae FP |
3107 | Otherwise return -1. |
3108 | Keep the capital `T' in `Token' for dumb truncating compilers | |
c6d46f5f JB |
3109 | (this distinguishes it from `TEX_toktab' */ |
3110 | int | |
3111 | TEX_Token (cp) | |
3112 | char *cp; | |
3113 | { | |
3114 | int i; | |
3115 | ||
3116 | for (i = 0; TEX_toktab[i].len > 0; i++) | |
1a0d8c80 | 3117 | if (strneq (TEX_toktab[i].name, cp, TEX_toktab[i].len)) |
c6d46f5f JB |
3118 | return i; |
3119 | return -1; | |
3120 | } | |
3121 | \f | |
3122 | /* Support for Prolog. */ | |
3123 | ||
aab1fdae | 3124 | /* Whole head (not only functor, but also arguments) |
c6d46f5f | 3125 | is gotten in compound term. */ |
c6d46f5f | 3126 | void |
b9755a12 | 3127 | prolog_getit (s) |
c6d46f5f | 3128 | char *s; |
c6d46f5f | 3129 | { |
55597f90 | 3130 | char *save_s; |
c6d46f5f JB |
3131 | int insquote, npar; |
3132 | ||
3133 | save_s = s; | |
3134 | insquote = FALSE; | |
3135 | npar = 0; | |
3136 | while (1) | |
3137 | { | |
55597f90 | 3138 | if (s[0] == '\0') /* syntax error. */ |
c6d46f5f | 3139 | return; |
55597f90 | 3140 | else if (insquote && s[0] == '\'' && s[1] == '\'') |
c6d46f5f | 3141 | s += 2; |
55597f90 | 3142 | else if (s[0] == '\'') |
c6d46f5f JB |
3143 | { |
3144 | insquote = !insquote; | |
3145 | s++; | |
3146 | } | |
55597f90 | 3147 | else if (!insquote && s[0] == '(') |
c6d46f5f JB |
3148 | { |
3149 | npar++; | |
3150 | s++; | |
3151 | } | |
55597f90 | 3152 | else if (!insquote && s[0] == ')') |
c6d46f5f JB |
3153 | { |
3154 | npar--; | |
3155 | s++; | |
3156 | if (npar == 0) | |
3157 | break; | |
3158 | else if (npar < 0) /* syntax error. */ | |
3159 | return; | |
3160 | } | |
55597f90 FP |
3161 | else if (!insquote && s[0] == '.' |
3162 | && (isspace (s[1]) || s[1] == '\0')) | |
c6d46f5f JB |
3163 | { /* fullstop. */ |
3164 | if (npar != 0) /* syntax error. */ | |
3165 | return; | |
3166 | s++; | |
3167 | break; | |
3168 | } | |
3169 | else | |
3170 | s++; | |
3171 | } | |
55597f90 FP |
3172 | pfnote (savenstr (save_s, s-save_s), TRUE, FALSE, |
3173 | save_s, s-save_s, lineno, linecharno); | |
c6d46f5f JB |
3174 | } |
3175 | ||
3176 | /* It is assumed that prolog predicate starts from column 0. */ | |
c6d46f5f | 3177 | void |
6dd5561c FP |
3178 | Prolog_functions (inf) |
3179 | FILE *inf; | |
c6d46f5f JB |
3180 | { |
3181 | void skip_comment (), prolog_getit (); | |
3182 | ||
3183 | lineno = linecharno = charno = 0; | |
6dd5561c | 3184 | while (!feof (inf)) |
c6d46f5f JB |
3185 | { |
3186 | lineno++; | |
3187 | linecharno += charno; | |
6dd5561c | 3188 | charno = readline (&lb, inf) + 1; /* 1 for newline. */ |
c6d46f5f JB |
3189 | dbp = lb.buffer; |
3190 | if (isspace (dbp[0])) /* not predicate header. */ | |
3191 | continue; | |
3192 | else if (dbp[0] == '%') /* comment. */ | |
3193 | continue; | |
3194 | else if (dbp[0] == '/' && dbp[1] == '*') /* comment. */ | |
6dd5561c | 3195 | skip_comment (&lb, inf, &lineno, &linecharno); |
c6d46f5f | 3196 | else /* found. */ |
b9755a12 | 3197 | prolog_getit (dbp); |
c6d46f5f JB |
3198 | } |
3199 | } | |
3200 | ||
3201 | void | |
6dd5561c | 3202 | skip_comment (plb, inf, plineno, plinecharno) |
c6d46f5f | 3203 | struct linebuffer *plb; |
6dd5561c | 3204 | FILE *inf; |
c6d46f5f JB |
3205 | int *plineno; /* result */ |
3206 | long *plinecharno; /* result */ | |
3207 | { | |
b9755a12 FP |
3208 | char *cp; |
3209 | ||
3210 | do | |
c6d46f5f | 3211 | { |
b9755a12 FP |
3212 | for (cp = plb->buffer; *cp != '\0'; cp++) |
3213 | if (cp[0] == '*' && cp[1] == '/') | |
3214 | return; | |
c6d46f5f | 3215 | (*plineno)++; |
b9755a12 FP |
3216 | *plinecharno += readline (plb, inf) + 1; /* 1 for newline. */ |
3217 | } | |
3218 | while (!feof(inf)); | |
c6d46f5f | 3219 | } |
b9755a12 FP |
3220 | \f |
3221 | #ifdef ETAGS_REGEXPS | |
3222 | /* Take a string like "/blah/" and turn it into "blah", making sure | |
3223 | that the first and last characters are the same, and handling | |
3224 | quoted separator characters. Actually, stops on the occurence of | |
3225 | an unquoted separator. Also turns "\t" into a Tab character. | |
3226 | Returns pointer to terminating separator. Works in place. Null | |
3227 | terminates name string. */ | |
3228 | char * | |
3229 | scan_separators (name) | |
3230 | char *name; | |
3231 | { | |
3232 | char sep = name[0]; | |
3233 | char *copyto = name; | |
3234 | logical quoted = FALSE; | |
3235 | ||
3236 | for (++name; *name != '\0'; ++name) | |
3237 | { | |
3238 | if (quoted) | |
3239 | { | |
3240 | if (*name == 't') | |
3241 | *copyto++ = '\t'; | |
3242 | else if (*name == sep) | |
3243 | *copyto++ = sep; | |
3244 | else | |
3245 | { | |
3246 | /* Something else is quoted, so preserve the quote. */ | |
3247 | *copyto++ = '\\'; | |
3248 | *copyto++ = *name; | |
3249 | } | |
3250 | quoted = FALSE; | |
3251 | } | |
3252 | else if (*name == '\\') | |
3253 | quoted = TRUE; | |
3254 | else if (*name == sep) | |
3255 | break; | |
3256 | else | |
3257 | *copyto++ = *name; | |
3258 | } | |
c6d46f5f | 3259 | |
b9755a12 FP |
3260 | /* Terminate copied string. */ |
3261 | *copyto = '\0'; | |
3262 | return name; | |
3263 | } | |
c6d46f5f | 3264 | |
b9755a12 FP |
3265 | /* Turn a name, which is an ed-style (but Emacs syntax) regular |
3266 | expression, into a real regular expression by compiling it. */ | |
3267 | void | |
3268 | add_regex (regexp_pattern) | |
3269 | char *regexp_pattern; | |
c6d46f5f | 3270 | { |
b9755a12 FP |
3271 | char *name; |
3272 | const char *err; | |
3273 | struct re_pattern_buffer *patbuf; | |
c6d46f5f | 3274 | |
b9755a12 FP |
3275 | if (regexp_pattern == NULL) |
3276 | { | |
3277 | /* Remove existing regexps. */ | |
3278 | num_patterns = 0; | |
3279 | patterns = NULL; | |
3280 | return; | |
3281 | } | |
c6d46f5f | 3282 | |
b9755a12 FP |
3283 | if (regexp_pattern[0] == '\0') |
3284 | { | |
3285 | error ("missing regexp", 0); | |
3286 | return; | |
3287 | } | |
3288 | if (regexp_pattern[strlen(regexp_pattern)-1] != regexp_pattern[0]) | |
3289 | { | |
3290 | error ("%s: unterminated regexp", regexp_pattern); | |
3291 | return; | |
3292 | } | |
3293 | name = scan_separators (regexp_pattern); | |
3294 | if (regexp_pattern[0] == '\0') | |
3295 | { | |
3296 | error ("null regexp", 0); | |
3297 | return; | |
3298 | } | |
3299 | (void) scan_separators (name); | |
3300 | ||
3301 | patbuf = xnew (1, struct re_pattern_buffer); | |
3302 | patbuf->translate = NULL; | |
3303 | patbuf->fastmap = NULL; | |
3304 | patbuf->buffer = NULL; | |
3305 | patbuf->allocated = 0; | |
3306 | ||
3307 | err = re_compile_pattern (regexp_pattern, strlen (regexp_pattern), patbuf); | |
3308 | if (err != NULL) | |
3309 | { | |
3310 | error ("%s while compiling pattern", err); | |
3311 | return; | |
3312 | } | |
3313 | ||
3314 | num_patterns += 1; | |
3315 | if (num_patterns == 1) | |
3316 | patterns = xnew (1, struct pattern); | |
c6d46f5f | 3317 | else |
b9755a12 FP |
3318 | patterns = ((struct pattern *) |
3319 | xrealloc (patterns, | |
3320 | (num_patterns * sizeof (struct pattern)))); | |
3321 | patterns[num_patterns - 1].pattern = patbuf; | |
3322 | patterns[num_patterns - 1].name_pattern = savestr (name); | |
3323 | patterns[num_patterns - 1].error_signaled = FALSE; | |
3324 | } | |
3325 | ||
3326 | /* | |
3327 | * Do the subtitutions indicated by the regular expression and | |
3328 | * arguments. | |
3329 | */ | |
3330 | char * | |
3331 | substitute (in, out, regs) | |
3332 | char *in, *out; | |
3333 | struct re_registers *regs; | |
3334 | { | |
3335 | char *result = NULL, *t; | |
3336 | int size = 0; | |
3337 | ||
3338 | /* Pass 1: figure out how much size to allocate. */ | |
3339 | for (t = out; *t; ++t) | |
3340 | { | |
3341 | if (*t == '\\') | |
3342 | { | |
3343 | ++t; | |
3344 | if (!*t) | |
3345 | { | |
3346 | fprintf (stderr, "%s: pattern subtitution ends prematurely\n", | |
3347 | progname); | |
3348 | return NULL; | |
3349 | } | |
3350 | if (isdigit (*t)) | |
3351 | { | |
3352 | int dig = *t - '0'; | |
3353 | size += regs->end[dig] - regs->start[dig]; | |
3354 | } | |
3355 | } | |
3356 | } | |
3357 | ||
3358 | /* Allocate space and do the substitutions. */ | |
3359 | result = xnew (size + 1, char); | |
3360 | size = 0; | |
3361 | for (; *out; ++out) | |
3362 | { | |
3363 | if (*out == '\\') | |
3364 | { | |
3365 | ++out; | |
3366 | if (isdigit (*out)) | |
3367 | { | |
3368 | /* Using "dig2" satisfies my debugger. Bleah. */ | |
3369 | int dig2 = *out - '0'; | |
3370 | strncpy (result + size, in + regs->start[dig2], | |
3371 | regs->end[dig2] - regs->start[dig2]); | |
3372 | size += regs->end[dig2] - regs->start[dig2]; | |
3373 | } | |
3374 | else | |
3375 | { | |
3376 | switch (*out) | |
3377 | { | |
3378 | case '\t': | |
3379 | result[size++] = '\t'; | |
3380 | break; | |
3381 | case '\\': | |
3382 | *out = '\\'; | |
3383 | break; | |
3384 | default: | |
3385 | result[size++] = *out; | |
3386 | break; | |
3387 | } | |
3388 | } | |
3389 | } | |
3390 | else | |
3391 | result[size++] = *out; | |
3392 | } | |
3393 | result[size] = '\0'; | |
3394 | ||
3395 | return result; | |
c6d46f5f JB |
3396 | } |
3397 | \f | |
b9755a12 | 3398 | #endif /* ETAGS_REGEXPS */ |
c6d46f5f | 3399 | /* Initialize a linebuffer for use */ |
c6d46f5f JB |
3400 | void |
3401 | initbuffer (linebuffer) | |
3402 | struct linebuffer *linebuffer; | |
3403 | { | |
3404 | linebuffer->size = 200; | |
3405 | linebuffer->buffer = xnew (200, char); | |
3406 | } | |
3407 | ||
3408 | /* | |
3409 | * Read a line of text from `stream' into `linebuffer'. | |
3410 | * Return the number of characters read from `stream', | |
3411 | * which is the length of the line including the newline, if any. | |
3412 | */ | |
3413 | long | |
b9755a12 | 3414 | readline_internal (linebuffer, stream) |
c6d46f5f JB |
3415 | struct linebuffer *linebuffer; |
3416 | register FILE *stream; | |
3417 | { | |
3418 | char *buffer = linebuffer->buffer; | |
3419 | register char *p = linebuffer->buffer; | |
3420 | register char *pend; | |
aab1fdae | 3421 | int chars_deleted; |
c6d46f5f | 3422 | |
eb8c3be9 | 3423 | pend = p + linebuffer->size; /* Separate to avoid 386/IX compiler bug. */ |
c6d46f5f JB |
3424 | |
3425 | while (1) | |
3426 | { | |
3427 | register int c = getc (stream); | |
3428 | if (p == pend) | |
3429 | { | |
3430 | linebuffer->size *= 2; | |
3431 | buffer = (char *) xrealloc (buffer, linebuffer->size); | |
3432 | p += buffer - linebuffer->buffer; | |
3433 | pend = buffer + linebuffer->size; | |
3434 | linebuffer->buffer = buffer; | |
3435 | } | |
aab1fdae | 3436 | if (c == EOF) |
c6d46f5f | 3437 | { |
aab1fdae FP |
3438 | chars_deleted = 0; |
3439 | break; | |
3440 | } | |
3441 | if (c == '\n') | |
3442 | { | |
3443 | if (p[-1] == '\r' && p > buffer) | |
3444 | { | |
3445 | *--p = '\0'; | |
3446 | chars_deleted = 2; | |
3447 | } | |
3448 | else | |
3449 | { | |
3450 | *p = '\0'; | |
3451 | chars_deleted = 1; | |
3452 | } | |
c6d46f5f JB |
3453 | break; |
3454 | } | |
3455 | *p++ = c; | |
3456 | } | |
3457 | ||
aab1fdae | 3458 | return p - buffer + chars_deleted; |
c6d46f5f | 3459 | } |
b9755a12 FP |
3460 | |
3461 | /* | |
3462 | * Like readline_internal, above, but try to match the input | |
3463 | * line against any existing regular expressions. | |
3464 | */ | |
3465 | long | |
3466 | readline (linebuffer, stream) | |
3467 | struct linebuffer *linebuffer; | |
3468 | FILE *stream; | |
3469 | { | |
3470 | /* Read new line. */ | |
3471 | int i; | |
3472 | long result = readline_internal (linebuffer, stream); | |
3473 | ||
3474 | #ifdef ETAGS_REGEXPS | |
3475 | /* Match against all listed patterns. */ | |
3476 | for (i = 0; i < num_patterns; ++i) | |
3477 | { | |
3478 | int match = re_match (patterns[i].pattern, linebuffer->buffer, | |
3479 | (int)result, 0, &patterns[i].regs); | |
3480 | switch (match) | |
3481 | { | |
3482 | case -2: | |
3483 | /* Some error. */ | |
3484 | if (!patterns[i].error_signaled) | |
3485 | { | |
3486 | error ("error while matching pattern %d", i); | |
3487 | patterns[i].error_signaled = TRUE; | |
3488 | } | |
3489 | break; | |
3490 | case -1: | |
3491 | /* No match. */ | |
3492 | break; | |
3493 | default: | |
3494 | /* Match occurred. Construct a tag. */ | |
3495 | if (patterns[i].name_pattern[0] != '\0') | |
3496 | { | |
3497 | /* Make a named tag. */ | |
3498 | char *name = substitute (linebuffer->buffer, | |
3499 | patterns[i].name_pattern, | |
3500 | &patterns[i].regs); | |
3501 | if (name != NULL) | |
3502 | pfnote (name, TRUE, TRUE, linebuffer->buffer, | |
3503 | match, lineno, linecharno); | |
3504 | } | |
3505 | else | |
3506 | { | |
3507 | /* Make an unnamed tag. */ | |
3508 | pfnote (NULL, TRUE, FALSE, linebuffer->buffer, | |
3509 | match, lineno, linecharno); | |
3510 | } | |
3511 | break; | |
3512 | } | |
3513 | } | |
3514 | #endif /* ETAGS_REGEXPS */ | |
3515 | ||
3516 | return result; | |
3517 | } | |
3518 | ||
3519 | /* | |
3520 | * Read a file, but do no processing. This is used to do regexp | |
3521 | * matching on files that have no language defined. | |
3522 | */ | |
3523 | void | |
3524 | just_read_file (inf) | |
3525 | FILE *inf; | |
3526 | { | |
3527 | while (!feof (inf)) | |
3528 | { | |
3529 | ++lineno; | |
3530 | linecharno = charno; | |
3531 | charno += readline (&lb, inf) + 1; | |
3532 | } | |
3533 | } | |
3534 | ||
c6d46f5f | 3535 | \f |
55597f90 FP |
3536 | /* |
3537 | * Return a pointer to a space of size strlen(cp)+1 allocated | |
3538 | * with xnew where the string CP has been copied. | |
3539 | */ | |
c6d46f5f JB |
3540 | char * |
3541 | savestr (cp) | |
3542 | char *cp; | |
3543 | { | |
3544 | return savenstr (cp, strlen (cp)); | |
3545 | } | |
3546 | ||
55597f90 FP |
3547 | /* |
3548 | * Return a pointer to a space of size LEN+1 allocated with xnew where | |
3549 | * the string CP has been copied for at most the first LEN characters. | |
3550 | */ | |
c6d46f5f JB |
3551 | char * |
3552 | savenstr (cp, len) | |
3553 | char *cp; | |
3554 | int len; | |
3555 | { | |
3556 | register char *dp; | |
3557 | ||
3558 | dp = xnew (len + 1, char); | |
1a0d8c80 | 3559 | strncpy (dp, cp, len); |
c6d46f5f JB |
3560 | dp[len] = '\0'; |
3561 | return dp; | |
3562 | } | |
3563 | ||
c6d46f5f JB |
3564 | /* |
3565 | * Return the ptr in sp at which the character c last | |
3566 | * appears; NULL if not found | |
3567 | * | |
b02c5fea | 3568 | * Identical to System V strrchr, included for portability. |
c6d46f5f | 3569 | */ |
c6d46f5f | 3570 | char * |
b02c5fea | 3571 | etags_strrchr (sp, c) |
c6d46f5f JB |
3572 | register char *sp, c; |
3573 | { | |
3574 | register char *r; | |
3575 | ||
3576 | r = NULL; | |
3577 | do | |
3578 | { | |
3579 | if (*sp == c) | |
3580 | r = sp; | |
3581 | } while (*sp++); | |
b9755a12 | 3582 | return r; |
c6d46f5f JB |
3583 | } |
3584 | ||
9d7ad1b3 | 3585 | |
c6d46f5f JB |
3586 | /* |
3587 | * Return the ptr in sp at which the character c first | |
3588 | * appears; NULL if not found | |
3589 | * | |
b02c5fea | 3590 | * Identical to System V strchr, included for portability. |
c6d46f5f | 3591 | */ |
c6d46f5f | 3592 | char * |
b02c5fea | 3593 | etags_strchr (sp, c) |
c6d46f5f JB |
3594 | register char *sp, c; |
3595 | { | |
3596 | do | |
3597 | { | |
3598 | if (*sp == c) | |
b9755a12 FP |
3599 | return sp; |
3600 | } while (*sp++); | |
3601 | return NULL; | |
c6d46f5f JB |
3602 | } |
3603 | ||
c6d46f5f | 3604 | /* Print error message and exit. */ |
c6d46f5f JB |
3605 | void |
3606 | fatal (s1, s2) | |
3607 | char *s1, *s2; | |
3608 | { | |
3609 | error (s1, s2); | |
1a0d8c80 | 3610 | exit (BAD); |
c6d46f5f JB |
3611 | } |
3612 | ||
cdc1f6a7 FP |
3613 | void |
3614 | pfatal (s1) | |
3615 | char *s1; | |
3616 | { | |
3617 | perror (s1); | |
3618 | exit (BAD); | |
3619 | } | |
3620 | ||
c6d46f5f | 3621 | /* Print error message. `s1' is printf control string, `s2' is arg for it. */ |
c6d46f5f JB |
3622 | void |
3623 | error (s1, s2) | |
3624 | char *s1, *s2; | |
3625 | { | |
3626 | fprintf (stderr, "%s: ", progname); | |
3627 | fprintf (stderr, s1, s2); | |
3628 | fprintf (stderr, "\n"); | |
3629 | } | |
3630 | ||
46c145db FP |
3631 | /* Return a newly-allocated string whose contents |
3632 | concatenate those of s1, s2, s3. */ | |
c6d46f5f JB |
3633 | char * |
3634 | concat (s1, s2, s3) | |
3635 | char *s1, *s2, *s3; | |
3636 | { | |
3637 | int len1 = strlen (s1), len2 = strlen (s2), len3 = strlen (s3); | |
3638 | char *result = xnew (len1 + len2 + len3 + 1, char); | |
3639 | ||
1a0d8c80 FP |
3640 | strcpy (result, s1); |
3641 | strcpy (result + len1, s2); | |
3642 | strcpy (result + len1 + len2, s3); | |
46c145db | 3643 | result[len1 + len2 + len3] = '\0'; |
c6d46f5f JB |
3644 | |
3645 | return result; | |
3646 | } | |
b02c5fea | 3647 | \f |
cdc1f6a7 FP |
3648 | /* Does the same work as the system V getcwd, but does not need to |
3649 | guess buffer size in advance. */ | |
88f125fc RS |
3650 | char * |
3651 | etags_getcwd () | |
cdc1f6a7 | 3652 | #ifdef DOS_NT |
88f125fc | 3653 | { |
cdc1f6a7 FP |
3654 | char *p, path[MAXPATHLEN + 1]; /* Fixed size is safe on MSDOS. */ |
3655 | ||
3656 | getwd (path); | |
3657 | p = path; | |
88f125fc RS |
3658 | while (*p) |
3659 | if (*p == '\\') | |
3660 | *p++ = '/'; | |
3661 | else | |
3662 | *p++ = tolower (*p); | |
cdc1f6a7 FP |
3663 | |
3664 | return strdup (path); | |
88f125fc | 3665 | } |
cdc1f6a7 | 3666 | #elif HAVE_GETCWD /* not DOS_NT */ |
b02c5fea | 3667 | { |
cdc1f6a7 FP |
3668 | int bufsize = 200; |
3669 | char *path = xnew (bufsize, char); | |
c6d46f5f | 3670 | |
cdc1f6a7 | 3671 | while (getcwd (path, bufsize) == NULL) |
b02c5fea | 3672 | { |
dcc89e63 | 3673 | if (errno != ERANGE) |
cdc1f6a7 | 3674 | pfatal ("pwd"); |
5e9c8296 | 3675 | bufsize *= 2; |
cdc1f6a7 | 3676 | path = xnew (bufsize, char); |
5e9c8296 | 3677 | } |
b02c5fea | 3678 | |
cdc1f6a7 FP |
3679 | return path; |
3680 | } | |
3681 | #else /* not DOS_NT and not HAVE_GETCWD */ | |
3682 | { | |
3683 | struct linebuffer path; | |
3684 | FILE *pipe; | |
b02c5fea | 3685 | |
cdc1f6a7 FP |
3686 | initbuffer (&path); |
3687 | pipe = (FILE *) popen ("pwd 2>/dev/null", "r"); | |
3688 | if (pipe == NULL || readline_internal (&path, pipe) == 0) | |
3689 | pfatal ("pwd"); | |
3690 | pclose (pipe); | |
b02c5fea | 3691 | |
cdc1f6a7 | 3692 | return path.buffer; |
b02c5fea | 3693 | } |
cdc1f6a7 | 3694 | #endif /* not DOS_NT and not HAVE_GETCWD */ |
b02c5fea FP |
3695 | |
3696 | /* Return a newly allocated string containing the filename | |
3697 | of FILE relative to the absolute directory DIR (which | |
3698 | should end with a slash). */ | |
46c145db FP |
3699 | char * |
3700 | relative_filename (file, dir) | |
3701 | char *file, *dir; | |
3702 | { | |
3703 | char *fp, *dp, *res; | |
3704 | ||
3705 | /* Find the common root of file and dir. */ | |
3706 | fp = absolute_filename (file, cwd); | |
3707 | dp = dir; | |
3708 | while (*fp++ == *dp++) | |
3709 | continue; | |
3710 | do | |
3711 | { | |
3712 | fp--; | |
3713 | dp--; | |
3714 | } | |
3715 | while (*fp != '/'); | |
3716 | ||
3717 | /* Build a sequence of "../" strings for the resulting relative filename. */ | |
b02c5fea | 3718 | for (dp = etags_strchr (dp + 1, '/'), res = ""; |
46c145db | 3719 | dp != NULL; |
b02c5fea | 3720 | dp = etags_strchr (dp + 1, '/')) |
46c145db FP |
3721 | { |
3722 | res = concat (res, "../", ""); | |
3723 | } | |
3724 | ||
3725 | /* Add the filename relative to the common root of file and dir. */ | |
3726 | res = concat (res, fp + 1, ""); | |
3727 | ||
3728 | return res; /* temporary stub */ | |
3729 | } | |
3730 | ||
3731 | /* Return a newly allocated string containing the | |
b02c5fea FP |
3732 | absolute filename of FILE given CWD (which should |
3733 | end with a slash). */ | |
46c145db FP |
3734 | char * |
3735 | absolute_filename (file, cwd) | |
3736 | char *file, *cwd; | |
3737 | { | |
3738 | char *slashp, *cp, *res; | |
3739 | ||
3740 | if (file[0] == '/') | |
3741 | res = concat (file, "", ""); | |
3742 | else | |
3743 | res = concat (cwd, file, ""); | |
3744 | ||
3745 | /* Delete the "/dirname/.." and "/." substrings. */ | |
b02c5fea | 3746 | slashp = etags_strchr (res, '/'); |
46c145db FP |
3747 | while (slashp != NULL && slashp[0] != '\0') |
3748 | { | |
3749 | if (slashp[1] == '.') | |
3750 | { | |
3751 | if (slashp[2] == '.' | |
3752 | && (slashp[3] == '/' || slashp[3] == '\0')) | |
3753 | { | |
3754 | cp = slashp; | |
3755 | do | |
3756 | cp--; | |
3757 | while (cp >= res && *cp != '/'); | |
3758 | if (*cp == '/') | |
3759 | { | |
3760 | strcpy (cp, slashp + 3); | |
3761 | } | |
3762 | else /* else (cp == res) */ | |
3763 | { | |
1875d994 | 3764 | if (slashp[3] != '\0') |
46c145db FP |
3765 | strcpy (cp, slashp + 4); |
3766 | else | |
3767 | return "."; | |
3768 | } | |
3769 | slashp = cp; | |
e9b2b94c | 3770 | continue; |
46c145db FP |
3771 | } |
3772 | else if (slashp[2] == '/' || slashp[2] == '\0') | |
3773 | { | |
3774 | strcpy (slashp, slashp + 2); | |
e9b2b94c | 3775 | continue; |
46c145db FP |
3776 | } |
3777 | } | |
e9b2b94c FP |
3778 | |
3779 | slashp = etags_strchr (slashp + 1, '/'); | |
46c145db FP |
3780 | } |
3781 | ||
3782 | return res; | |
3783 | } | |
3784 | ||
b02c5fea FP |
3785 | /* Return a newly allocated string containing the absolute |
3786 | filename of dir where FILE resides given CWD (which should | |
3787 | end with a slash). */ | |
46c145db FP |
3788 | char * |
3789 | absolute_dirname (file, cwd) | |
3790 | char *file, *cwd; | |
3791 | { | |
3792 | char *slashp, *res; | |
3793 | char save; | |
3794 | ||
b02c5fea | 3795 | slashp = etags_strrchr (file, '/'); |
46c145db FP |
3796 | if (slashp == NULL) |
3797 | return cwd; | |
3798 | save = slashp[1]; | |
3799 | slashp[1] = '\0'; | |
3800 | res = absolute_filename (file, cwd); | |
3801 | slashp[1] = save; | |
3802 | ||
3803 | return res; | |
3804 | } | |
3805 | ||
c6d46f5f | 3806 | /* Like malloc but get fatal error if memory is exhausted. */ |
c6d46f5f JB |
3807 | char * |
3808 | xmalloc (size) | |
42680d3c | 3809 | unsigned int size; |
c6d46f5f | 3810 | { |
1a0d8c80 FP |
3811 | char *result = (char *) malloc (size); |
3812 | if (result == NULL) | |
c6d46f5f JB |
3813 | fatal ("virtual memory exhausted", 0); |
3814 | return result; | |
3815 | } | |
3816 | ||
3817 | char * | |
3818 | xrealloc (ptr, size) | |
3819 | char *ptr; | |
42680d3c | 3820 | unsigned int size; |
c6d46f5f | 3821 | { |
1a0d8c80 FP |
3822 | char *result = (char *) realloc (ptr, size); |
3823 | if (result == NULL) | |
c6d46f5f JB |
3824 | fatal ("virtual memory exhausted"); |
3825 | return result; | |
3826 | } |