Commit | Line | Data |
---|---|---|
c6d46f5f | 1 | /* Tags file maker to go with GNU Emacs |
f470f9bd KH |
2 | Copyright (C) 1984, 87, 88, 89, 93, 94, 95 |
3 | Free Software Foundation, Inc. and Ken Arnold | |
ea6cd314 | 4 | This file is not considered part of GNU Emacs. |
c6d46f5f | 5 | |
ea6cd314 | 6 | This program is free software; you can redistribute it and/or modify |
c6d46f5f | 7 | it under the terms of the GNU General Public License as published by |
ea6cd314 RS |
8 | the Free Software Foundation; either version 2 of the License, or |
9 | (at your option) any later version. | |
c6d46f5f | 10 | |
ea6cd314 | 11 | This program is distributed in the hope that it will be useful, |
c6d46f5f JB |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | GNU General Public License for more details. | |
15 | ||
16 | You should have received a copy of the GNU General Public License | |
ea6cd314 RS |
17 | along with this program; if not, write to the Free Software |
18 | Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ | |
c6d46f5f JB |
19 | |
20 | /* | |
21 | * Authors: | |
22 | * Ctags originally by Ken Arnold. | |
6dd5561c | 23 | * Fortran added by Jim Kleckner. |
c6d46f5f JB |
24 | * Ed Pelegri-Llopart added C typedefs. |
25 | * Gnu Emacs TAGS format and modifications by RMS? | |
26 | * Sam Kendall added C++. | |
b9755a12 FP |
27 | * Francesco Potorti` reorganised C and C++ based on work by Joe Wells. |
28 | #ifdef ETAGS_REGEXPS | |
29 | * Regexp tags by Tom Tromey. | |
30 | #endif | |
31d4b314 | 31 | * |
46c145db | 32 | * Francesco Potorti` (pot@cnuce.cnr.it) is the current maintainer. |
c6d46f5f JB |
33 | */ |
34 | ||
0ff74d81 | 35 | char pot_etags_version[] = "@(#) pot revision number is 11.30"; |
75bdbc6a FP |
36 | |
37 | #define TRUE 1 | |
38 | #define FALSE 0 | |
39 | #ifndef DEBUG | |
40 | # define DEBUG FALSE | |
41 | #endif | |
46c145db | 42 | |
c6880c90 RS |
43 | #ifdef MSDOS |
44 | #include <fcntl.h> | |
88f125fc | 45 | #include <sys/param.h> |
c6880c90 RS |
46 | #endif /* MSDOS */ |
47 | ||
c05b6df5 RS |
48 | #ifdef WINDOWSNT |
49 | #include <stdlib.h> | |
50 | #include <fcntl.h> | |
51 | #include <string.h> | |
52 | #define MAXPATHLEN _MAX_PATH | |
53 | #endif | |
54 | ||
72a339d7 | 55 | #ifdef HAVE_CONFIG_H |
b9755a12 | 56 | #include <config.h> |
aab1fdae FP |
57 | /* On some systems, Emacs defines static as nothing for the sake |
58 | of unexec. We don't want that here since we don't use unexec. */ | |
1ddff51c | 59 | #undef static |
1e134a5f RM |
60 | #endif |
61 | ||
918f9ad1 JB |
62 | #include <stdio.h> |
63 | #include <ctype.h> | |
dcc89e63 FP |
64 | #include <errno.h> |
65 | #ifndef errno | |
66 | extern int errno; | |
67 | #endif | |
918f9ad1 JB |
68 | #include <sys/types.h> |
69 | #include <sys/stat.h> | |
70 | ||
2b878b4c FP |
71 | #if !defined (S_ISREG) && defined (S_IFREG) |
72 | # define S_ISREG(m) (((m) & S_IFMT) == S_IFREG) | |
73 | #endif | |
74 | ||
b9755a12 FP |
75 | #include <getopt.h> |
76 | ||
77 | #ifdef ETAGS_REGEXPS | |
78 | #include <regex.h> | |
79 | #endif /* ETAGS_REGEXPS */ | |
918f9ad1 | 80 | |
32daa216 FP |
81 | /* Define CTAGS to make the program "ctags" compatible with the usual one. |
82 | Let it undefined to make the program "etags", which makes emacs-style | |
83 | tag tables and tags typedefs, #defines and struct/union/enum by default. */ | |
84 | #ifdef CTAGS | |
85 | # undef CTAGS | |
86 | # define CTAGS TRUE | |
87 | #else | |
88 | # define CTAGS FALSE | |
c6d46f5f JB |
89 | #endif |
90 | ||
91 | /* Exit codes for success and failure. */ | |
92 | #ifdef VMS | |
32daa216 FP |
93 | #define GOOD 1 |
94 | #define BAD 0 | |
c6d46f5f | 95 | #else |
32daa216 FP |
96 | #define GOOD 0 |
97 | #define BAD 1 | |
c6d46f5f JB |
98 | #endif |
99 | ||
55597f90 FP |
100 | /* C extensions. */ |
101 | #define C_PLPL 0x00001 /* C++ */ | |
102 | #define C_STAR 0x00003 /* C* */ | |
103 | #define YACC 0x10000 /* yacc file */ | |
c6d46f5f | 104 | |
aab1fdae FP |
105 | #define streq(s,t) (strcmp (s, t) == 0) |
106 | #define strneq(s,t,n) (strncmp (s, t, n) == 0) | |
c6d46f5f | 107 | |
79263656 FP |
108 | #define lowcase(c) ((c) | ' ') |
109 | ||
c6d46f5f JB |
110 | #define iswhite(arg) (_wht[arg]) /* T if char is white */ |
111 | #define begtoken(arg) (_btk[arg]) /* T if char can start token */ | |
112 | #define intoken(arg) (_itk[arg]) /* T if char can be in token */ | |
113 | #define endtoken(arg) (_etk[arg]) /* T if char ends tokens */ | |
c6d46f5f | 114 | |
b2db879b FP |
115 | #ifdef DOS_NT |
116 | # define absolutefn(fn) (fn[0] == '/' || (isalpha (fn[0]) && fn[1] == ':')) | |
117 | #else | |
118 | # define absolutefn(fn) (fn[0] == '/') | |
119 | #endif | |
120 | ||
121 | ||
55597f90 FP |
122 | /* |
123 | * xnew -- allocate storage | |
124 | * | |
125 | * SYNOPSIS: Type *xnew (int n, Type); | |
126 | */ | |
127 | #define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type))) | |
c6d46f5f | 128 | |
aab1fdae FP |
129 | typedef int logical; |
130 | ||
55597f90 | 131 | typedef struct nd_st |
c6d46f5f JB |
132 | { /* sorting structure */ |
133 | char *name; /* function or type name */ | |
134 | char *file; /* file name */ | |
135 | logical is_func; /* use pattern or line no */ | |
fe0b3356 | 136 | logical named; /* list name separately */ |
c6d46f5f JB |
137 | logical been_warned; /* set if noticed dup */ |
138 | int lno; /* line number tag is on */ | |
139 | long cno; /* character number line starts on */ | |
140 | char *pat; /* search pattern */ | |
141 | struct nd_st *left, *right; /* left and right sons */ | |
55597f90 | 142 | } NODE; |
c6d46f5f | 143 | |
55597f90 | 144 | extern char *getenv (); |
c6d46f5f JB |
145 | |
146 | char *concat (); | |
46c145db | 147 | char *savenstr (), *savestr (); |
b02c5fea FP |
148 | char *etags_strchr (), *etags_strrchr (); |
149 | char *etags_getcwd (); | |
46c145db | 150 | char *relative_filename (), *absolute_filename (), *absolute_dirname (); |
03cdafdf | 151 | long *xmalloc (), *xrealloc (); |
b9755a12 FP |
152 | |
153 | typedef void Lang_function (); | |
cdc1f6a7 | 154 | #if FALSE /* many compilers barf on this */ |
b9755a12 FP |
155 | Lang_function Asm_labels; |
156 | Lang_function default_C_entries; | |
157 | Lang_function C_entries; | |
158 | Lang_function Cplusplus_entries; | |
159 | Lang_function Cstar_entries; | |
160 | Lang_function Fortran_functions; | |
161 | Lang_function Yacc_entries; | |
162 | Lang_function Lisp_functions; | |
163 | Lang_function Pascal_functions; | |
164 | Lang_function Prolog_functions; | |
165 | Lang_function Scheme_functions; | |
166 | Lang_function TeX_functions; | |
167 | Lang_function just_read_file; | |
aab1fdae FP |
168 | #else /* so let's write it this way */ |
169 | void Asm_labels (); | |
aab1fdae | 170 | void C_entries (); |
79263656 FP |
171 | void default_C_entries (); |
172 | void plain_C_entries (); | |
aab1fdae FP |
173 | void Cplusplus_entries (); |
174 | void Cstar_entries (); | |
175 | void Fortran_functions (); | |
176 | void Yacc_entries (); | |
177 | void Lisp_functions (); | |
178 | void Pascal_functions (); | |
179 | void Prolog_functions (); | |
180 | void Scheme_functions (); | |
181 | void TeX_functions (); | |
182 | void just_read_file (); | |
183 | #endif | |
b9755a12 FP |
184 | |
185 | logical get_language (); | |
c6d46f5f | 186 | int total_size_of_entries (); |
c6d46f5f | 187 | long readline (); |
b9755a12 FP |
188 | long readline_internal (); |
189 | #ifdef ETAGS_REGEXPS | |
190 | void add_regex (); | |
191 | #endif | |
c6d46f5f JB |
192 | void add_node (); |
193 | void error (); | |
cdc1f6a7 | 194 | void fatal (), pfatal (); |
55597f90 | 195 | void find_entries (); |
c6d46f5f JB |
196 | void free_tree (); |
197 | void getit (); | |
c6d46f5f JB |
198 | void init (); |
199 | void initbuffer (); | |
c6d46f5f JB |
200 | void pfnote (); |
201 | void process_file (); | |
202 | void put_entries (); | |
203 | void takeprec (); | |
204 | ||
c6d46f5f | 205 | \f |
55597f90 | 206 | char searchar = '/'; /* use /.../ searches */ |
c6d46f5f | 207 | |
55597f90 FP |
208 | int lineno; /* line number of current line */ |
209 | long charno; /* current character number */ | |
c6d46f5f | 210 | |
55597f90 FP |
211 | long linecharno; /* charno of start of line; not used by C, |
212 | but by every other language. */ | |
c6d46f5f | 213 | |
55597f90 FP |
214 | char *curfile; /* current input file name */ |
215 | char *tagfile; /* output file */ | |
216 | char *progname; /* name this program was invoked with */ | |
217 | char *cwd; /* current working directory */ | |
218 | char *tagfiledir; /* directory of tagfile */ | |
c6d46f5f | 219 | |
55597f90 FP |
220 | FILE *tagf; /* ioptr for tags file */ |
221 | NODE *head; /* the head of the binary tree of tags */ | |
c6d46f5f | 222 | |
55597f90 FP |
223 | /* |
224 | * A `struct linebuffer' is a structure which holds a line of text. | |
225 | * `readline' reads a line from a stream into a linebuffer and works | |
226 | * regardless of the length of the line. | |
227 | */ | |
228 | struct linebuffer | |
229 | { | |
230 | long size; | |
231 | char *buffer; | |
232 | }; | |
c6d46f5f | 233 | |
55597f90 | 234 | struct linebuffer lb; /* the current line */ |
75bdbc6a | 235 | struct linebuffer token_name; /* used by C_entries as temporary area */ |
55597f90 FP |
236 | struct |
237 | { | |
238 | long linepos; | |
239 | struct linebuffer lb; /* used by C_entries instead of lb */ | |
240 | } lbs[2]; | |
c6d46f5f | 241 | |
55597f90 FP |
242 | /* boolean "functions" (see init) */ |
243 | logical _wht[0177], _etk[0177], _itk[0177], _btk[0177]; | |
244 | char | |
ee70dba5 FP |
245 | *white = " \f\t\n\013", /* white chars */ |
246 | *endtk = " \t\n\013\"'#()[]{}=-+%*/&|^~!<>;,.:?", /* token ending chars */ | |
13fde0cd | 247 | /* token starting chars */ |
b12756c8 | 248 | *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~", |
46c145db | 249 | /* valid in-token chars */ |
13fde0cd | 250 | *intk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789"; |
c6d46f5f | 251 | |
55597f90 FP |
252 | logical append_to_tagfile; /* -a: append to tags */ |
253 | /* The following three default to TRUE for etags, but to FALSE for ctags. */ | |
254 | logical typedefs; /* -t: create tags for typedefs */ | |
255 | logical typedefs_and_cplusplus; /* -T: create tags for typedefs, level */ | |
c6d46f5f | 256 | /* 0 struct/enum/union decls, and C++ */ |
32daa216 | 257 | /* member functions. */ |
55597f90 | 258 | logical constantypedefs; /* -d: create tags for C #define and enum */ |
32daa216 | 259 | /* constants. Enum consts not implemented. */ |
c6d46f5f | 260 | /* -D: opposite of -d. Default under ctags. */ |
55597f90 FP |
261 | logical update; /* -u: update tags */ |
262 | logical vgrind_style; /* -v: create vgrind style index output */ | |
263 | logical no_warnings; /* -w: suppress warnings */ | |
264 | logical cxref_style; /* -x: create cxref style output */ | |
265 | logical cplusplus; /* .[hc] means C++, not C */ | |
201f9f2b | 266 | logical noindentypedefs; /* -I: ignore indentation in C */ |
55597f90 | 267 | #define permit_duplicates TRUE /* allow duplicate tags */ |
c6d46f5f | 268 | |
7537186d FP |
269 | struct option longopts[] = |
270 | { | |
4746118a JB |
271 | { "append", no_argument, NULL, 'a' }, |
272 | { "backward-search", no_argument, NULL, 'B' }, | |
273 | { "c++", no_argument, NULL, 'C' }, | |
274 | { "cxref", no_argument, NULL, 'x' }, | |
275 | { "defines", no_argument, NULL, 'd' }, | |
7537186d | 276 | { "help", no_argument, NULL, 'h' }, |
4746118a | 277 | { "help", no_argument, NULL, 'H' }, |
201f9f2b | 278 | { "ignore-indentation", no_argument, NULL, 'I' }, |
4746118a | 279 | { "include", required_argument, NULL, 'i' }, |
b9755a12 | 280 | { "language", required_argument, NULL, 'l' }, |
4746118a | 281 | { "no-defines", no_argument, NULL, 'D' }, |
b9755a12 | 282 | { "no-regex", no_argument, NULL, 'R' }, |
4746118a JB |
283 | { "no-warn", no_argument, NULL, 'w' }, |
284 | { "output", required_argument, NULL, 'o' }, | |
b9755a12 | 285 | { "regex", required_argument, NULL, 'r' }, |
4746118a JB |
286 | { "typedefs", no_argument, NULL, 't' }, |
287 | { "typedefs-and-c++", no_argument, NULL, 'T' }, | |
288 | { "update", no_argument, NULL, 'u' }, | |
289 | { "version", no_argument, NULL, 'V' }, | |
290 | { "vgrind", no_argument, NULL, 'v' }, | |
291 | { 0 } | |
292 | }; | |
293 | ||
b9755a12 FP |
294 | #ifdef ETAGS_REGEXPS |
295 | /* Structure defining a regular expression. Elements are | |
296 | the compiled pattern, and the name string. */ | |
297 | struct pattern | |
298 | { | |
299 | struct re_pattern_buffer *pattern; | |
300 | struct re_registers regs; | |
301 | char *name_pattern; | |
302 | logical error_signaled; | |
303 | }; | |
304 | ||
305 | /* Number of regexps found. */ | |
306 | int num_patterns = 0; | |
307 | ||
308 | /* Array of all regexps. */ | |
309 | struct pattern *patterns = NULL; | |
310 | #endif /* ETAGS_REGEXPS */ | |
311 | ||
312 | /* Language stuff. */ | |
313 | struct lang_entry | |
314 | { | |
79263656 | 315 | char *suffix; |
b9755a12 FP |
316 | Lang_function *function; |
317 | }; | |
318 | ||
319 | /* Table of language names and corresponding functions. */ | |
320 | /* It is ok for a given function to be listed under more than one | |
321 | name. I just didn't. */ | |
322 | /* "auto" language reverts to default behavior. */ | |
323 | struct lang_entry lang_names[] = | |
324 | { | |
325 | { "asm", Asm_labels }, | |
326 | { "c", default_C_entries }, | |
327 | { "c++", Cplusplus_entries }, | |
328 | { "c*", Cstar_entries }, | |
329 | { "fortran", Fortran_functions }, | |
330 | { "lisp", Lisp_functions }, | |
331 | { "none", just_read_file }, | |
332 | { "pascal", Pascal_functions }, | |
333 | { "scheme" , Scheme_functions }, | |
334 | { "tex", TeX_functions }, | |
335 | { "auto", NULL }, | |
336 | { NULL, NULL } | |
337 | }; | |
338 | ||
79263656 FP |
339 | /* Table of file name suffixes and corresponding language functions. */ |
340 | struct lang_entry lang_suffixes[] = | |
b9755a12 FP |
341 | { |
342 | /* Assume that ".s" or ".a" is assembly code. -wolfgang. | |
343 | Or even ".sa". */ | |
344 | { "a", Asm_labels }, /* Unix assembler */ | |
345 | { "asm", Asm_labels }, /* Microcontroller assembly */ | |
346 | { "def", Asm_labels }, /* BSO/Tasking definition includes */ | |
347 | { "inc", Asm_labels }, /* Microcontroller include files */ | |
348 | { "ins", Asm_labels }, /* Microcontroller include files */ | |
349 | { "s", Asm_labels }, | |
350 | { "sa", Asm_labels }, /* Unix assembler */ | |
351 | { "src", Asm_labels }, /* BSO/Tasking C compiler output */ | |
352 | ||
353 | /* .aux, .bbl, .clo, .cls, .dtx or .tex implies LaTeX source code. */ | |
354 | { "aux", TeX_functions }, | |
355 | { "bbl", TeX_functions }, | |
356 | { "clo", TeX_functions }, | |
357 | { "cls", TeX_functions }, | |
358 | { "dtx", TeX_functions }, | |
359 | { "sty", TeX_functions }, | |
360 | { "tex", TeX_functions }, | |
361 | ||
362 | /* .l or .el or .lisp (or .cl or .clisp or ...) implies lisp source code */ | |
363 | { "cl", Lisp_functions }, | |
364 | { "clisp", Lisp_functions }, | |
365 | { "el", Lisp_functions }, | |
366 | { "l", Lisp_functions }, | |
367 | { "lisp", Lisp_functions }, | |
368 | { "lsp", Lisp_functions }, | |
369 | ||
370 | /* .scm or .sm or .scheme implies scheme source code */ | |
371 | { "SCM", Scheme_functions }, | |
372 | { "SM", Scheme_functions }, | |
373 | { "oak", Scheme_functions }, | |
374 | { "sch", Scheme_functions }, | |
375 | { "scheme", Scheme_functions }, | |
376 | { "scm", Scheme_functions }, | |
377 | { "sm", Scheme_functions }, | |
378 | { "t", Scheme_functions }, | |
379 | /* FIXME Can't do the `SCM' or `scm' prefix with a version number */ | |
380 | ||
79263656 FP |
381 | /* Note that .c and .h can be considered C++, if the --c++ flag was |
382 | given. That is why default_C_entries is called here. */ | |
b9755a12 FP |
383 | { "c", default_C_entries }, |
384 | { "h", default_C_entries }, | |
385 | ||
79263656 FP |
386 | /* .pc is a Pro*C file. */ |
387 | { "pc", plain_C_entries }, | |
388 | ||
2bd88040 FP |
389 | /* .C or .H or .c++ or .cc or .cpp or .cxx or .h++ or .hh or .hxx: |
390 | a C++ file */ | |
b9755a12 FP |
391 | { "C", Cplusplus_entries }, |
392 | { "H", Cplusplus_entries }, | |
2bd88040 | 393 | { "c++", Cplusplus_entries }, |
b9755a12 FP |
394 | { "cc", Cplusplus_entries }, |
395 | { "cpp", Cplusplus_entries }, | |
396 | { "cxx", Cplusplus_entries }, | |
2bd88040 | 397 | { "h++", Cplusplus_entries }, |
b9755a12 FP |
398 | { "hh", Cplusplus_entries }, |
399 | { "hxx", Cplusplus_entries }, | |
400 | ||
401 | /* .y: a yacc file */ | |
402 | { "y", Yacc_entries }, | |
403 | ||
404 | /* .cs or .hs: a C* file */ | |
405 | { "cs", Cstar_entries }, | |
406 | { "hs", Cstar_entries }, | |
407 | ||
b2db879b | 408 | /* .F, .f and .for are FORTRAN. */ |
b9755a12 FP |
409 | { "F", Fortran_functions }, |
410 | { "f", Fortran_functions }, | |
411 | { "for", Fortran_functions }, | |
412 | ||
413 | /* .pl implies prolog source code */ | |
414 | { "pl", Prolog_functions }, | |
415 | ||
416 | /* .p or .pas: a Pascal file */ | |
417 | { "p", Pascal_functions }, | |
418 | { "pas", Pascal_functions }, | |
419 | ||
420 | { NULL, NULL } | |
421 | }; | |
422 | ||
423 | /* Non-NULL if language fixed. */ | |
424 | Lang_function *lang_func = NULL; | |
425 | ||
c6d46f5f | 426 | \f |
b9755a12 FP |
427 | void |
428 | print_language_names () | |
429 | { | |
430 | struct lang_entry *name, *ext; | |
431 | ||
432 | puts ("\nThese are the currently supported languages, along with the\n\ | |
79263656 FP |
433 | default file name suffixes:"); |
434 | for (name = lang_names; name->suffix; ++name) | |
b9755a12 | 435 | { |
79263656 FP |
436 | printf ("\t%s\t", name->suffix); |
437 | for (ext = lang_suffixes; ext->suffix; ++ext) | |
b9755a12 | 438 | if (name->function == ext->function) |
79263656 | 439 | printf (" .%s", ext->suffix); |
b9755a12 FP |
440 | puts (""); |
441 | } | |
79263656 FP |
442 | puts ("Where `auto' means use default language for files based on file\n\ |
443 | name suffix, and `none' means only do regexp processing on files.\n\ | |
444 | If no language is specified and no matching suffix is found,\n\ | |
b9755a12 FP |
445 | Fortran is tried first; if no tags are found, C is tried next."); |
446 | } | |
447 | ||
4746118a JB |
448 | void |
449 | print_version () | |
450 | { | |
32daa216 | 451 | #ifdef VERSION |
cc582743 | 452 | printf ("%s for Emacs version %s.\n", (CTAGS) ? "CTAGS" : "ETAGS", VERSION); |
32daa216 FP |
453 | #else |
454 | printf ("%s for Emacs version 19.\n", (CTAGS) ? "CTAGS" : "ETAGS"); | |
455 | #endif | |
4746118a | 456 | |
1a0d8c80 | 457 | exit (GOOD); |
4746118a JB |
458 | } |
459 | ||
460 | void | |
461 | print_help () | |
462 | { | |
463 | printf ("These are the options accepted by %s. You may use unambiguous\n\ | |
b9755a12 | 464 | abbreviations for the long option names. A - as file name means read\n\ |
1a0d8c80 | 465 | names from stdin.\n\n", progname); |
4746118a | 466 | |
52cc7c59 JB |
467 | puts ("-a, --append\n\ |
468 | Append tag entries to existing tags file."); | |
1a0d8c80 | 469 | |
32daa216 FP |
470 | if (CTAGS) |
471 | puts ("-B, --backward-search\n\ | |
1a0d8c80 | 472 | Write the search commands for the tag entries using '?', the\n\ |
3ad2882c | 473 | backward-search command instead of '/', the forward-search command."); |
1a0d8c80 | 474 | |
52cc7c59 | 475 | puts ("-C, --c++\n\ |
79263656 | 476 | Treat files whose name suffix defaults to C language as C++ files."); |
4746118a | 477 | |
32daa216 FP |
478 | if (CTAGS) |
479 | puts ("-d, --defines\n\ | |
ee70dba5 | 480 | Create tag entries for constant C #defines, too."); |
32daa216 FP |
481 | else |
482 | puts ("-D, --no-defines\n\ | |
ee70dba5 FP |
483 | Don't create tag entries for constant C #defines. This makes\n\ |
484 | the tags file smaller."); | |
4746118a | 485 | |
32daa216 | 486 | if (!CTAGS) |
b9755a12 FP |
487 | { |
488 | puts ("-i FILE, --include=FILE\n\ | |
1a0d8c80 FP |
489 | Include a note in tag file indicating that, when searching for\n\ |
490 | a tag, one should also consult the tags file FILE after\n\ | |
491 | checking the current file."); | |
b9755a12 FP |
492 | puts ("-l LANG, --language=LANG\n\ |
493 | Force the following files to be considered as written in the\n\ | |
494 | named language up to the next --language=LANG option."); | |
7537186d FP |
495 | } |
496 | ||
b9755a12 | 497 | #ifdef ETAGS_REGEXPS |
7537186d | 498 | puts ("-r /REGEXP/, --regex=/REGEXP/\n\ |
b9755a12 FP |
499 | Make a tag for each line matching pattern REGEXP in the\n\ |
500 | following files. REGEXP is anchored (as if preceded by ^).\n\ | |
501 | The form /REGEXP/NAME/ creates a named tag. For example Tcl\n\ | |
502 | named tags can be created with:\n\ | |
503 | --regex=/proc[ \\t]+\\([^ \\t]+\\)/\\1/."); | |
7537186d | 504 | puts ("-R, --no-regex\n\ |
b9755a12 FP |
505 | Don't create tags from regexps for the following files."); |
506 | #endif /* ETAGS_REGEXPS */ | |
1a0d8c80 FP |
507 | puts ("-o FILE, --output=FILE\n\ |
508 | Write the tags to FILE."); | |
201f9f2b | 509 | puts ("-I, --ignore-indentation\n\ |
4746118a JB |
510 | Don't rely on indentation quite as much as normal. Currently,\n\ |
511 | this means not to assume that a closing brace in the first\n\ | |
512 | column is the final brace of a function or structure\n\ | |
32daa216 | 513 | definition in C and C++."); |
4746118a | 514 | |
32daa216 FP |
515 | if (CTAGS) |
516 | { | |
517 | puts ("-t, --typedefs\n\ | |
518 | Generate tag entries for C typedefs."); | |
519 | puts ("-T, --typedefs-and-c++\n\ | |
520 | Generate tag entries for C typedefs, C struct/enum/union tags,\n\ | |
521 | and C++ member functions."); | |
32daa216 | 522 | puts ("-u, --update\n\ |
4746118a JB |
523 | Update the tag entries for the given files, leaving tag\n\ |
524 | entries for other files in place. Currently, this is\n\ | |
525 | implemented by deleting the existing entries for the given\n\ | |
526 | files and then rewriting the new entries at the end of the\n\ | |
527 | tags file. It is often faster to simply rebuild the entire\n\ | |
52cc7c59 | 528 | tag file than to use this."); |
32daa216 | 529 | puts ("-v, --vgrind\n\ |
4746118a JB |
530 | Generates an index of items intended for human consumption,\n\ |
531 | similar to the output of vgrind. The index is sorted, and\n\ | |
52cc7c59 | 532 | gives the page number of each item."); |
b9755a12 FP |
533 | puts ("-w, --no-warn\n\ |
534 | Suppress warning messages about entries defined in multiple\n\ | |
535 | files."); | |
32daa216 | 536 | puts ("-x, --cxref\n\ |
4746118a JB |
537 | Like --vgrind, but in the style of cxref, rather than vgrind.\n\ |
538 | The output uses line numbers instead of page numbers, but\n\ | |
539 | beyond that the differences are cosmetic; try both to see\n\ | |
52cc7c59 | 540 | which you like."); |
32daa216 | 541 | } |
4746118a JB |
542 | |
543 | puts ("-V, --version\n\ | |
544 | Print the version of the program.\n\ | |
7537186d | 545 | -h, --help\n\ |
4746118a JB |
546 | Print this help message."); |
547 | ||
b9755a12 FP |
548 | print_language_names (); |
549 | ||
1a0d8c80 | 550 | exit (GOOD); |
4746118a JB |
551 | } |
552 | ||
553 | \f | |
b9755a12 FP |
554 | enum argument_type |
555 | { | |
556 | at_language, | |
557 | at_regexp, | |
558 | at_filename | |
559 | }; | |
560 | ||
561 | /* This structure helps us allow mixing of --lang and filenames. */ | |
562 | typedef struct | |
563 | { | |
564 | enum argument_type arg_type; | |
565 | char *what; | |
566 | Lang_function *function; | |
567 | } ARGUMENT; | |
568 | ||
569 | #ifdef VMS /* VMS specific functions */ | |
570 | ||
571 | #define EOS '\0' | |
572 | ||
573 | /* This is a BUG! ANY arbitrary limit is a BUG! | |
574 | Won't someone please fix this? */ | |
575 | #define MAX_FILE_SPEC_LEN 255 | |
576 | typedef struct { | |
577 | short curlen; | |
578 | char body[MAX_FILE_SPEC_LEN + 1]; | |
579 | } vspec; | |
580 | ||
581 | /* | |
582 | v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names | |
583 | returning in each successive call the next filename matching the input | |
584 | spec. The function expects that each in_spec passed | |
585 | to it will be processed to completion; in particular, up to and | |
586 | including the call following that in which the last matching name | |
587 | is returned, the function ignores the value of in_spec, and will | |
588 | only start processing a new spec with the following call. | |
589 | If an error occurs, on return out_spec contains the value | |
590 | of in_spec when the error occurred. | |
591 | ||
592 | With each successive filename returned in out_spec, the | |
593 | function's return value is one. When there are no more matching | |
594 | names the function returns zero. If on the first call no file | |
595 | matches in_spec, or there is any other error, -1 is returned. | |
596 | */ | |
597 | ||
598 | #include <rmsdef.h> | |
599 | #include <descrip.h> | |
600 | #define OUTSIZE MAX_FILE_SPEC_LEN | |
601 | short | |
602 | fn_exp (out, in) | |
603 | vspec *out; | |
604 | char *in; | |
605 | { | |
606 | static long context = 0; | |
607 | static struct dsc$descriptor_s o; | |
608 | static struct dsc$descriptor_s i; | |
609 | static logical pass1 = TRUE; | |
610 | long status; | |
611 | short retval; | |
612 | ||
613 | if (pass1) | |
614 | { | |
615 | pass1 = FALSE; | |
616 | o.dsc$a_pointer = (char *) out; | |
617 | o.dsc$w_length = (short)OUTSIZE; | |
618 | i.dsc$a_pointer = in; | |
619 | i.dsc$w_length = (short)strlen(in); | |
620 | i.dsc$b_dtype = DSC$K_DTYPE_T; | |
621 | i.dsc$b_class = DSC$K_CLASS_S; | |
622 | o.dsc$b_dtype = DSC$K_DTYPE_VT; | |
623 | o.dsc$b_class = DSC$K_CLASS_VS; | |
624 | } | |
625 | if ((status = lib$find_file(&i, &o, &context, 0, 0)) == RMS$_NORMAL) | |
626 | { | |
627 | out->body[out->curlen] = EOS; | |
628 | return 1; | |
629 | } | |
630 | else if (status == RMS$_NMF) | |
631 | retval = 0; | |
632 | else | |
633 | { | |
634 | strcpy(out->body, in); | |
635 | retval = -1; | |
636 | } | |
637 | lib$find_file_end(&context); | |
638 | pass1 = TRUE; | |
639 | return retval; | |
640 | } | |
641 | ||
642 | /* | |
643 | v1.01 nmm 19-Aug-85 gfnames - return in successive calls the | |
644 | name of each file specified by the provided arg expanding wildcards. | |
645 | */ | |
646 | char * | |
647 | gfnames (arg, p_error) | |
648 | char *arg; | |
649 | logical *p_error; | |
650 | { | |
651 | static vspec filename = {MAX_FILE_SPEC_LEN, "\0"}; | |
652 | ||
653 | switch (fn_exp (&filename, arg)) | |
654 | { | |
655 | case 1: | |
656 | *p_error = FALSE; | |
657 | return filename.body; | |
658 | case 0: | |
659 | *p_error = FALSE; | |
660 | return NULL; | |
661 | default: | |
662 | *p_error = TRUE; | |
663 | return filename.body; | |
664 | } | |
665 | } | |
666 | ||
667 | #ifndef OLD /* Newer versions of VMS do provide `system'. */ | |
668 | system (cmd) | |
669 | char *cmd; | |
670 | { | |
671 | fprintf (stderr, "system() function not implemented under VMS\n"); | |
672 | } | |
673 | #endif | |
674 | ||
675 | #define VERSION_DELIM ';' | |
676 | char *massage_name (s) | |
677 | char *s; | |
678 | { | |
679 | char *start = s; | |
680 | ||
681 | for ( ; *s; s++) | |
682 | if (*s == VERSION_DELIM) | |
683 | { | |
684 | *s = EOS; | |
685 | break; | |
686 | } | |
687 | else | |
688 | *s = tolower(*s); | |
689 | return start; | |
690 | } | |
691 | #endif /* VMS */ | |
692 | ||
693 | \f | |
c6d46f5f JB |
694 | void |
695 | main (argc, argv) | |
696 | int argc; | |
697 | char *argv[]; | |
698 | { | |
c6d46f5f | 699 | int i; |
1e134a5f | 700 | unsigned int nincluded_files = 0; |
72a339d7 | 701 | char **included_files = xnew (argc, char *); |
c6d46f5f | 702 | char *this_file; |
b9755a12 FP |
703 | ARGUMENT *argbuffer; |
704 | int current_arg = 0, file_count = 0; | |
55597f90 | 705 | struct linebuffer filename_lb; |
c6d46f5f | 706 | #ifdef VMS |
b9755a12 | 707 | logical got_err; |
c6d46f5f | 708 | #endif |
46c145db | 709 | |
c05b6df5 | 710 | #ifdef DOS_NT |
42680d3c | 711 | _fmode = O_BINARY; /* all of files are treated as binary files */ |
c05b6df5 | 712 | #endif /* DOS_NT */ |
c6880c90 | 713 | |
c6d46f5f JB |
714 | progname = argv[0]; |
715 | ||
b9755a12 FP |
716 | /* Allocate enough no matter what happens. Overkill, but each one |
717 | is small. */ | |
718 | argbuffer = xnew (argc, ARGUMENT); | |
719 | ||
720 | #ifdef ETAGS_REGEXPS | |
721 | /* Set syntax for regular expression routines. */ | |
722 | re_set_syntax (RE_SYNTAX_EMACS); | |
723 | #endif /* ETAGS_REGEXPS */ | |
724 | ||
c6d46f5f JB |
725 | /* |
726 | * If etags, always find typedefs and structure tags. Why not? | |
727 | * Also default is to find macro constants. | |
728 | */ | |
32daa216 | 729 | if (!CTAGS) |
55597f90 | 730 | typedefs = typedefs_and_cplusplus = constantypedefs = TRUE; |
c6d46f5f | 731 | |
dcc89e63 | 732 | while (1) |
c6d46f5f | 733 | { |
b9755a12 | 734 | int opt = getopt_long (argc, argv, |
201f9f2b | 735 | "-aCdDf:Il:o:r:RStTi:BuvxwVhH", longopts, 0); |
4746118a JB |
736 | |
737 | if (opt == EOF) | |
738 | break; | |
739 | ||
740 | switch (opt) | |
c6d46f5f | 741 | { |
b02c5fea FP |
742 | case 0: |
743 | /* If getopt returns 0, then it has already processed a | |
4746118a JB |
744 | long-named option. We should do nothing. */ |
745 | break; | |
746 | ||
b9755a12 FP |
747 | case 1: |
748 | /* This means that a filename has been seen. Record it. */ | |
749 | argbuffer[current_arg].arg_type = at_filename; | |
750 | argbuffer[current_arg].what = optarg; | |
751 | ++current_arg; | |
752 | ++file_count; | |
753 | break; | |
754 | ||
4746118a JB |
755 | /* Common options. */ |
756 | case 'a': | |
55597f90 | 757 | append_to_tagfile = TRUE; |
4746118a JB |
758 | break; |
759 | case 'C': | |
55597f90 | 760 | cplusplus = TRUE; |
4746118a JB |
761 | break; |
762 | case 'd': | |
55597f90 | 763 | constantypedefs = TRUE; |
4746118a JB |
764 | break; |
765 | case 'D': | |
55597f90 | 766 | constantypedefs = FALSE; |
4746118a | 767 | break; |
32daa216 | 768 | case 'f': /* for compatibility with old makefiles */ |
4746118a | 769 | case 'o': |
6dd5561c | 770 | if (tagfile) |
c6d46f5f | 771 | { |
201f9f2b FP |
772 | fprintf (stderr, "%s: -%c option may only be given once.\n", |
773 | progname, opt); | |
c6d46f5f JB |
774 | goto usage; |
775 | } | |
6dd5561c | 776 | tagfile = optarg; |
4746118a | 777 | break; |
201f9f2b FP |
778 | case 'I': |
779 | case 'S': /* for backward compatibility */ | |
780 | noindentypedefs = TRUE; | |
781 | break; | |
b9755a12 FP |
782 | case 'l': |
783 | if (!get_language (optarg, &argbuffer[current_arg].function)) | |
784 | { | |
785 | fprintf (stderr, "%s: language \"%s\" not recognized.\n", | |
786 | progname, optarg); | |
787 | goto usage; | |
788 | } | |
789 | argbuffer[current_arg].arg_type = at_language; | |
790 | ++current_arg; | |
791 | break; | |
792 | #ifdef ETAGS_REGEXPS | |
793 | case 'r': | |
794 | argbuffer[current_arg].arg_type = at_regexp; | |
795 | argbuffer[current_arg].what = optarg; | |
796 | ++current_arg; | |
797 | break; | |
798 | case 'R': | |
799 | argbuffer[current_arg].arg_type = at_regexp; | |
800 | argbuffer[current_arg].what = NULL; | |
801 | ++current_arg; | |
802 | break; | |
803 | #endif /* ETAGS_REGEXPS */ | |
4746118a JB |
804 | case 'V': |
805 | print_version (); | |
806 | break; | |
7537186d | 807 | case 'h': |
4746118a JB |
808 | case 'H': |
809 | print_help (); | |
810 | break; | |
0c1fd2e3 | 811 | case 't': |
55597f90 | 812 | typedefs = TRUE; |
0c1fd2e3 FP |
813 | break; |
814 | case 'T': | |
55597f90 | 815 | typedefs = typedefs_and_cplusplus = TRUE; |
0c1fd2e3 | 816 | break; |
b02c5fea | 817 | #if (!CTAGS) |
4746118a JB |
818 | /* Etags options */ |
819 | case 'i': | |
4746118a JB |
820 | included_files[nincluded_files++] = optarg; |
821 | break; | |
b02c5fea | 822 | #else /* CTAGS */ |
4746118a JB |
823 | /* Ctags options. */ |
824 | case 'B': | |
825 | searchar = '?'; | |
4746118a | 826 | break; |
4746118a | 827 | case 'u': |
55597f90 | 828 | update = TRUE; |
4746118a JB |
829 | break; |
830 | case 'v': | |
55597f90 | 831 | vgrind_style = TRUE; |
4746118a JB |
832 | /*FALLTHRU*/ |
833 | case 'x': | |
55597f90 | 834 | cxref_style = TRUE; |
4746118a JB |
835 | break; |
836 | case 'w': | |
55597f90 | 837 | no_warnings = TRUE; |
4746118a | 838 | break; |
b02c5fea | 839 | #endif /* CTAGS */ |
4746118a JB |
840 | default: |
841 | goto usage; | |
c6d46f5f | 842 | } |
c6d46f5f JB |
843 | } |
844 | ||
b9755a12 FP |
845 | for (; optind < argc; ++optind) |
846 | { | |
847 | argbuffer[current_arg].arg_type = at_filename; | |
848 | argbuffer[current_arg].what = argv[optind]; | |
849 | ++current_arg; | |
850 | ++file_count; | |
851 | } | |
852 | ||
853 | if (nincluded_files == 0 && file_count == 0) | |
c6d46f5f | 854 | { |
4746118a JB |
855 | fprintf (stderr, "%s: No input files specified.\n", progname); |
856 | ||
c6d46f5f | 857 | usage: |
201f9f2b FP |
858 | fprintf (stderr, "\tTry `%s --help' for a complete list of options.\n", |
859 | progname); | |
c6d46f5f JB |
860 | exit (BAD); |
861 | } | |
862 | ||
6dd5561c | 863 | if (tagfile == NULL) |
c6d46f5f | 864 | { |
6dd5561c | 865 | tagfile = CTAGS ? "tags" : "TAGS"; |
c6d46f5f | 866 | } |
b02c5fea | 867 | cwd = etags_getcwd (); /* the current working directory */ |
46c145db | 868 | strcat (cwd, "/"); |
6dd5561c | 869 | if (streq (tagfile, "-")) |
46c145db | 870 | { |
6dd5561c | 871 | tagfiledir = cwd; |
46c145db FP |
872 | } |
873 | else | |
874 | { | |
6dd5561c | 875 | tagfiledir = absolute_dirname (tagfile, cwd); |
46c145db | 876 | } |
c6d46f5f | 877 | |
b9755a12 | 878 | init (); /* set up boolean "functions" */ |
c6d46f5f JB |
879 | |
880 | initbuffer (&lb); | |
75bdbc6a | 881 | initbuffer (&token_name); |
13fde0cd RS |
882 | initbuffer (&lbs[0].lb); |
883 | initbuffer (&lbs[1].lb); | |
c6d46f5f | 884 | initbuffer (&filename_lb); |
b9755a12 | 885 | |
32daa216 | 886 | if (!CTAGS) |
c6d46f5f | 887 | { |
6dd5561c FP |
888 | if (streq (tagfile, "-")) |
889 | tagf = stdout; | |
c6d46f5f | 890 | else |
6dd5561c FP |
891 | tagf = fopen (tagfile, append_to_tagfile ? "a" : "w"); |
892 | if (tagf == NULL) | |
cdc1f6a7 | 893 | pfatal (tagfile); |
c6d46f5f JB |
894 | } |
895 | ||
b9755a12 FP |
896 | /* |
897 | * Loop through files finding functions. | |
898 | */ | |
899 | for (i = 0; i < current_arg; ++i) | |
c6d46f5f | 900 | { |
b9755a12 | 901 | switch (argbuffer[i].arg_type) |
c6d46f5f | 902 | { |
b9755a12 FP |
903 | case at_language: |
904 | lang_func = argbuffer[i].function; | |
905 | break; | |
906 | #ifdef ETAGS_REGEXPS | |
907 | case at_regexp: | |
908 | add_regex (argbuffer[i].what); | |
909 | break; | |
c6d46f5f | 910 | #endif |
b9755a12 FP |
911 | case at_filename: |
912 | #ifdef VMS | |
913 | while ((this_file = gfnames (argbuffer[i].what, &got_err)) != NULL) | |
914 | { | |
915 | if (got_err) | |
916 | { | |
917 | error ("Can't find file %s\n", this_file); | |
918 | argc--, argv++; | |
919 | } | |
920 | else | |
921 | { | |
922 | this_file = massage_name (this_file); | |
923 | } | |
c6d46f5f | 924 | #else |
b9755a12 | 925 | this_file = argbuffer[i].what; |
c6d46f5f | 926 | #endif |
b9755a12 FP |
927 | /* Input file named "-" means read file names from stdin |
928 | and use them. */ | |
929 | if (streq (this_file, "-")) | |
9cb0aa73 FP |
930 | while (readline_internal (&filename_lb, stdin) > 0) |
931 | process_file (filename_lb.buffer); | |
b9755a12 FP |
932 | else |
933 | process_file (this_file); | |
934 | #ifdef VMS | |
c6d46f5f | 935 | } |
b9755a12 FP |
936 | #endif |
937 | break; | |
c6d46f5f | 938 | } |
46c145db | 939 | } |
9cb0aa73 | 940 | |
32daa216 | 941 | if (!CTAGS) |
c6d46f5f | 942 | { |
1e134a5f | 943 | while (nincluded_files-- > 0) |
6dd5561c | 944 | fprintf (tagf, "\f\n%s,include\n", *included_files++); |
1e134a5f | 945 | |
55597f90 | 946 | fclose (tagf); |
1a0d8c80 | 947 | exit (GOOD); |
c6d46f5f JB |
948 | } |
949 | ||
55597f90 FP |
950 | /* If CTAGS, we are here. process_file did not write the tags yet, |
951 | because we want them ordered. Let's do it now. */ | |
c6d46f5f JB |
952 | if (cxref_style) |
953 | { | |
55597f90 FP |
954 | tagf = fopen (tagfile, append_to_tagfile ? "a" : "w"); |
955 | if (tagf == NULL) | |
956 | pfatal (tagfile); | |
c6d46f5f JB |
957 | put_entries (head); |
958 | exit (GOOD); | |
959 | } | |
55597f90 | 960 | |
4746118a | 961 | if (update) |
c6d46f5f | 962 | { |
55597f90 | 963 | char cmd[BUFSIZ]; |
b9755a12 | 964 | for (i = 0; i < current_arg; ++i) |
c6d46f5f | 965 | { |
55597f90 | 966 | if (argbuffer[i].arg_type != at_filename) |
b9755a12 | 967 | continue; |
c6d46f5f JB |
968 | sprintf (cmd, |
969 | "mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS", | |
b9755a12 | 970 | tagfile, argbuffer[i].what, tagfile); |
55597f90 FP |
971 | if (system (cmd) != GOOD) |
972 | fatal ("failed to execute shell command"); | |
c6d46f5f | 973 | } |
55597f90 | 974 | append_to_tagfile = TRUE; |
c6d46f5f | 975 | } |
55597f90 | 976 | |
6dd5561c FP |
977 | tagf = fopen (tagfile, append_to_tagfile ? "a" : "w"); |
978 | if (tagf == NULL) | |
55597f90 | 979 | pfatal (tagfile); |
c6d46f5f | 980 | put_entries (head); |
55597f90 FP |
981 | fclose (tagf); |
982 | ||
c6d46f5f JB |
983 | if (update) |
984 | { | |
55597f90 | 985 | char cmd[BUFSIZ]; |
6dd5561c | 986 | sprintf (cmd, "sort %s -o %s", tagfile, tagfile); |
55597f90 | 987 | exit (system (cmd)); |
c6d46f5f JB |
988 | } |
989 | exit (GOOD); | |
990 | } | |
991 | ||
992 | ||
b9755a12 FP |
993 | /* |
994 | * Set the language, given the name. | |
995 | */ | |
996 | logical | |
997 | get_language (language, func) | |
998 | char *language; | |
999 | Lang_function **func; | |
1000 | { | |
1001 | struct lang_entry *lang; | |
1002 | ||
79263656 | 1003 | for (lang = lang_names; lang->suffix; ++lang) |
b9755a12 | 1004 | { |
79263656 | 1005 | if (streq (language, lang->suffix)) |
b9755a12 FP |
1006 | { |
1007 | *func = lang->function; | |
1008 | return TRUE; | |
1009 | } | |
1010 | } | |
1011 | ||
1012 | return FALSE; | |
1013 | } | |
1014 | ||
1015 | ||
c6d46f5f JB |
1016 | /* |
1017 | * This routine is called on each file argument. | |
1018 | */ | |
1019 | void | |
1020 | process_file (file) | |
1021 | char *file; | |
1022 | { | |
1023 | struct stat stat_buf; | |
55597f90 | 1024 | FILE *inf; |
c6d46f5f | 1025 | |
42680d3c | 1026 | if (stat (file, &stat_buf) == 0 && !S_ISREG (stat_buf.st_mode)) |
c6d46f5f JB |
1027 | { |
1028 | fprintf (stderr, "Skipping %s: it is not a regular file.\n", file); | |
1029 | return; | |
1030 | } | |
6dd5561c | 1031 | if (streq (file, tagfile) && !streq (tagfile, "-")) |
c6d46f5f JB |
1032 | { |
1033 | fprintf (stderr, "Skipping inclusion of %s in self.\n", file); | |
1034 | return; | |
1035 | } | |
55597f90 FP |
1036 | inf = fopen (file, "r"); |
1037 | if (inf == NULL) | |
42680d3c | 1038 | { |
55597f90 | 1039 | perror (file); |
42680d3c FP |
1040 | return; |
1041 | } | |
55597f90 FP |
1042 | |
1043 | find_entries (file, inf); | |
1044 | ||
32daa216 | 1045 | if (!CTAGS) |
c6d46f5f | 1046 | { |
46c145db FP |
1047 | char *filename; |
1048 | ||
b2db879b | 1049 | if (absolutefn (file)) |
46c145db FP |
1050 | { |
1051 | /* file is an absolute filename. Canonicalise it. */ | |
1052 | filename = absolute_filename (file, cwd); | |
1053 | } | |
1054 | else | |
1055 | { | |
1056 | /* file is a filename relative to cwd. Make it relative | |
1057 | to the directory of the tags file. */ | |
6dd5561c | 1058 | filename = relative_filename (file, tagfiledir); |
46c145db | 1059 | } |
6dd5561c | 1060 | fprintf (tagf, "\f\n%s,%d\n", filename, total_size_of_entries (head)); |
a8d9bd4b | 1061 | free (filename); |
c6d46f5f JB |
1062 | put_entries (head); |
1063 | free_tree (head); | |
1064 | head = NULL; | |
1065 | } | |
1066 | } | |
1067 | ||
1068 | /* | |
eb8c3be9 | 1069 | * This routine sets up the boolean pseudo-functions which work |
99e0a2e0 | 1070 | * by setting boolean flags dependent upon the corresponding character |
c6d46f5f JB |
1071 | * Every char which is NOT in that string is not a white char. Therefore, |
1072 | * all of the array "_wht" is set to FALSE, and then the elements | |
1073 | * subscripted by the chars in "white" are set to TRUE. Thus "_wht" | |
1074 | * of a char is TRUE if it is the string "white", else FALSE. | |
1075 | */ | |
1076 | void | |
1077 | init () | |
1078 | { | |
13fde0cd RS |
1079 | register char *sp; |
1080 | register int i; | |
c6d46f5f JB |
1081 | |
1082 | for (i = 0; i < 0177; i++) | |
13fde0cd | 1083 | _wht[i] = _etk[i] = _itk[i] = _btk[i] = FALSE; |
c6d46f5f JB |
1084 | for (sp = white; *sp; sp++) |
1085 | _wht[*sp] = TRUE; | |
1086 | for (sp = endtk; *sp; sp++) | |
1087 | _etk[*sp] = TRUE; | |
1088 | for (sp = intk; *sp; sp++) | |
1089 | _itk[*sp] = TRUE; | |
1090 | for (sp = begtk; *sp; sp++) | |
1091 | _btk[*sp] = TRUE; | |
c6d46f5f JB |
1092 | _wht[0] = _wht['\n']; |
1093 | _etk[0] = _etk['\n']; | |
1094 | _btk[0] = _btk['\n']; | |
1095 | _itk[0] = _itk['\n']; | |
c6d46f5f JB |
1096 | } |
1097 | ||
1098 | /* | |
1099 | * This routine opens the specified file and calls the function | |
1100 | * which finds the function and type definitions. | |
1101 | */ | |
55597f90 FP |
1102 | void |
1103 | find_entries (file, inf) | |
c6d46f5f | 1104 | char *file; |
55597f90 | 1105 | FILE *inf; |
c6d46f5f | 1106 | { |
b9755a12 FP |
1107 | char *cp; |
1108 | struct lang_entry *lang; | |
1109 | NODE *old_last_node; | |
1110 | extern NODE *last_node; | |
c6d46f5f | 1111 | |
c6d46f5f | 1112 | curfile = savestr (file); |
b02c5fea | 1113 | cp = etags_strrchr (file, '.'); |
c6d46f5f | 1114 | |
b9755a12 FP |
1115 | /* If user specified a language, use it. */ |
1116 | if (lang_func != NULL) | |
13fde0cd | 1117 | { |
b9755a12 FP |
1118 | lang_func (inf); |
1119 | fclose (inf); | |
55597f90 | 1120 | return; |
13fde0cd | 1121 | } |
b9755a12 FP |
1122 | |
1123 | if (cp) | |
c6d46f5f | 1124 | { |
b9755a12 | 1125 | ++cp; |
79263656 | 1126 | for (lang = lang_suffixes; lang->suffix; ++lang) |
b9755a12 | 1127 | { |
79263656 | 1128 | if (streq (cp, lang->suffix)) |
b9755a12 FP |
1129 | { |
1130 | lang->function (inf); | |
1131 | fclose (inf); | |
55597f90 | 1132 | return; |
b9755a12 FP |
1133 | } |
1134 | } | |
c6d46f5f | 1135 | } |
c6d46f5f | 1136 | |
b9755a12 FP |
1137 | /* Try Fortran. */ |
1138 | old_last_node = last_node; | |
1139 | Fortran_functions (inf); | |
c6d46f5f | 1140 | |
b9755a12 FP |
1141 | /* No Fortran entries found. Try C. */ |
1142 | if (old_last_node == last_node) | |
b2db879b FP |
1143 | { |
1144 | rewind (inf); | |
1145 | default_C_entries (inf); | |
1146 | } | |
b9755a12 | 1147 | fclose (inf); |
c6d46f5f JB |
1148 | } |
1149 | \f | |
1150 | /* Record a tag. */ | |
c6d46f5f | 1151 | void |
fe0b3356 | 1152 | pfnote (name, is_func, named, linestart, linelen, lno, cno) |
c6d46f5f | 1153 | char *name; /* tag name */ |
55597f90 FP |
1154 | logical is_func; /* tag is a function */ |
1155 | logical named; /* tag different from text of definition */ | |
1156 | char *linestart; /* start of the line where tag is */ | |
1157 | int linelen; /* length of the line where tag is */ | |
1158 | int lno; /* line number */ | |
1159 | long cno; /* character number */ | |
1160 | { | |
1161 | register NODE *np = xnew (1, NODE); | |
c6d46f5f | 1162 | register char *fp; |
c6d46f5f | 1163 | |
c6d46f5f | 1164 | /* If ctags mode, change name "main" to M<thisfilename>. */ |
32daa216 | 1165 | if (CTAGS && !cxref_style && streq (name, "main")) |
c6d46f5f | 1166 | { |
b02c5fea | 1167 | fp = etags_strrchr (curfile, '/'); |
55597f90 FP |
1168 | np->name = concat ("M", fp == 0 ? curfile : fp + 1, ""); |
1169 | fp = etags_strrchr (np->name, '.'); | |
c6d46f5f | 1170 | if (fp && fp[1] != '\0' && fp[2] == '\0') |
55597f90 FP |
1171 | fp[0] = 0; |
1172 | np->named = TRUE; | |
1173 | } | |
1174 | else | |
1175 | { | |
1176 | np->name = name; | |
1177 | np->named = named; | |
c6d46f5f | 1178 | } |
4b533b5b | 1179 | np->been_warned = FALSE; |
c6d46f5f JB |
1180 | np->file = curfile; |
1181 | np->is_func = is_func; | |
c6d46f5f | 1182 | np->lno = lno; |
aab1fdae FP |
1183 | /* Our char numbers are 0-base, because of C language tradition? |
1184 | ctags compatibility? old versions compatibility? I don't know. | |
1185 | Anyway, since emacs's are 1-base we espect etags.el to take care | |
1186 | of the difference. If we wanted to have 1-based numbers, we would | |
1187 | uncomment the +1 below. */ | |
1188 | np->cno = cno /* + 1 */ ; | |
55597f90 FP |
1189 | np->left = np->right = NULL; |
1190 | np->pat = savenstr (linestart, ((CTAGS && !cxref_style) ? 50 : linelen)); | |
c6d46f5f JB |
1191 | |
1192 | add_node (np, &head); | |
1193 | } | |
1194 | ||
1195 | /* | |
1196 | * free_tree () | |
1197 | * recurse on left children, iterate on right children. | |
1198 | */ | |
1199 | void | |
1200 | free_tree (node) | |
1201 | register NODE *node; | |
1202 | { | |
1203 | while (node) | |
1204 | { | |
1205 | register NODE *node_right = node->right; | |
1206 | free_tree (node->left); | |
55597f90 FP |
1207 | if (node->named) |
1208 | free (node->name); | |
c6d46f5f JB |
1209 | free (node->pat); |
1210 | free ((char *) node); | |
1211 | node = node_right; | |
1212 | } | |
1213 | } | |
1214 | ||
1215 | /* | |
1216 | * add_node () | |
1217 | * Adds a node to the tree of nodes. In etags mode, we don't keep | |
1218 | * it sorted; we just keep a linear list. In ctags mode, maintain | |
1219 | * an ordered tree, with no attempt at balancing. | |
1220 | * | |
1221 | * add_node is the only function allowed to add nodes, so it can | |
1222 | * maintain state. | |
1223 | */ | |
6dd5561c | 1224 | NODE *last_node = NULL; |
c6d46f5f JB |
1225 | void |
1226 | add_node (node, cur_node_p) | |
1227 | NODE *node, **cur_node_p; | |
1228 | { | |
1229 | register int dif; | |
1230 | register NODE *cur_node = *cur_node_p; | |
c6d46f5f JB |
1231 | |
1232 | if (cur_node == NULL) | |
1233 | { | |
1234 | *cur_node_p = node; | |
1235 | last_node = node; | |
1236 | return; | |
1237 | } | |
1238 | ||
32daa216 | 1239 | if (!CTAGS) |
c6d46f5f JB |
1240 | { |
1241 | /* Etags Mode */ | |
1a0d8c80 FP |
1242 | if (last_node == NULL) |
1243 | fatal ("internal error in add_node", 0); | |
c6d46f5f JB |
1244 | last_node->right = node; |
1245 | last_node = node; | |
1246 | } | |
1247 | else | |
1248 | { | |
1249 | /* Ctags Mode */ | |
1250 | dif = strcmp (node->name, cur_node->name); | |
1251 | ||
1252 | /* | |
1253 | * If this tag name matches an existing one, then | |
1254 | * do not add the node, but maybe print a warning. | |
1255 | */ | |
1256 | if (!dif) | |
1257 | { | |
1258 | if (node->file == cur_node->file) | |
1259 | { | |
1260 | if (!no_warnings) | |
1261 | { | |
1262 | fprintf (stderr, "Duplicate entry in file %s, line %d: %s\n", | |
1263 | node->file, lineno, node->name); | |
1264 | fprintf (stderr, "Second entry ignored\n"); | |
1265 | } | |
1266 | return; | |
1267 | } | |
1268 | if (!cur_node->been_warned && !no_warnings) | |
1269 | { | |
1270 | fprintf (stderr, | |
1271 | "Duplicate entry in files %s and %s: %s (Warning only)\n", | |
1272 | node->file, cur_node->file, node->name); | |
1273 | } | |
1274 | cur_node->been_warned = TRUE; | |
1275 | return; | |
1276 | } | |
1277 | ||
1278 | /* Maybe refuse to add duplicate nodes. */ | |
1279 | if (!permit_duplicates) | |
1280 | { | |
1a0d8c80 FP |
1281 | if (streq (node->name, cur_node->name) |
1282 | && streq (node->file, cur_node->file)) | |
c6d46f5f JB |
1283 | return; |
1284 | } | |
1285 | ||
1286 | /* Actually add the node */ | |
1287 | add_node (node, dif < 0 ? &cur_node->left : &cur_node->right); | |
1288 | } | |
1289 | } | |
1290 | \f | |
1291 | void | |
1292 | put_entries (node) | |
13fde0cd | 1293 | register NODE *node; |
c6d46f5f | 1294 | { |
13fde0cd | 1295 | register char *sp; |
c6d46f5f JB |
1296 | |
1297 | if (node == NULL) | |
1298 | return; | |
1299 | ||
1300 | /* Output subentries that precede this one */ | |
1301 | put_entries (node->left); | |
1302 | ||
1303 | /* Output this entry */ | |
1304 | ||
32daa216 | 1305 | if (!CTAGS) |
c6d46f5f | 1306 | { |
fe0b3356 | 1307 | if (node->named) |
c6d46f5f | 1308 | { |
6dd5561c | 1309 | fprintf (tagf, "%s\177%s\001%d,%d\n", |
cc6d6e58 RM |
1310 | node->pat, node->name, |
1311 | node->lno, node->cno); | |
c6d46f5f JB |
1312 | } |
1313 | else | |
1314 | { | |
6dd5561c | 1315 | fprintf (tagf, "%s\177%d,%d\n", |
cc6d6e58 RM |
1316 | node->pat, |
1317 | node->lno, node->cno); | |
c6d46f5f JB |
1318 | } |
1319 | } | |
1320 | else if (!cxref_style) | |
1321 | { | |
6dd5561c | 1322 | fprintf (tagf, "%s\t%s\t", |
c6d46f5f JB |
1323 | node->name, node->file); |
1324 | ||
1325 | if (node->is_func) | |
1326 | { /* a function */ | |
6dd5561c FP |
1327 | putc (searchar, tagf); |
1328 | putc ('^', tagf); | |
c6d46f5f JB |
1329 | |
1330 | for (sp = node->pat; *sp; sp++) | |
1331 | { | |
1332 | if (*sp == '\\' || *sp == searchar) | |
6dd5561c FP |
1333 | putc ('\\', tagf); |
1334 | putc (*sp, tagf); | |
c6d46f5f | 1335 | } |
6dd5561c | 1336 | putc (searchar, tagf); |
c6d46f5f JB |
1337 | } |
1338 | else | |
1339 | { /* a typedef; text pattern inadequate */ | |
6dd5561c | 1340 | fprintf (tagf, "%d", node->lno); |
c6d46f5f | 1341 | } |
6dd5561c | 1342 | putc ('\n', tagf); |
c6d46f5f JB |
1343 | } |
1344 | else if (vgrind_style) | |
1345 | fprintf (stdout, "%s %s %d\n", | |
1346 | node->name, node->file, (node->lno + 63) / 64); | |
1347 | else | |
daa37602 | 1348 | fprintf (stdout, "%-16s %3d %-16s %s\n", |
c6d46f5f JB |
1349 | node->name, node->lno, node->file, node->pat); |
1350 | ||
1351 | /* Output subentries that follow this one */ | |
1352 | put_entries (node->right); | |
1353 | } | |
1354 | ||
1355 | /* Length of a number's decimal representation. */ | |
1356 | int | |
1357 | number_len (num) | |
1358 | long num; | |
1359 | { | |
1360 | int len = 0; | |
1361 | if (!num) | |
1362 | return 1; | |
1363 | for (; num; num /= 10) | |
1364 | ++len; | |
1365 | return len; | |
1366 | } | |
1367 | ||
1368 | /* | |
1369 | * Return total number of characters that put_entries will output for | |
32daa216 FP |
1370 | * the nodes in the subtree of the specified node. Works only if |
1371 | * we are not ctags, but called only in that case. This count | |
1372 | * is irrelevant with the new tags.el, but is still supplied for | |
1373 | * backward compatibility. | |
c6d46f5f JB |
1374 | */ |
1375 | int | |
1376 | total_size_of_entries (node) | |
13fde0cd | 1377 | register NODE *node; |
c6d46f5f | 1378 | { |
13fde0cd | 1379 | register int total; |
c6d46f5f JB |
1380 | |
1381 | if (node == NULL) | |
1382 | return 0; | |
1383 | ||
1384 | total = 0; | |
1385 | for (; node; node = node->right) | |
1386 | { | |
1387 | /* Count left subentries. */ | |
1388 | total += total_size_of_entries (node->left); | |
1389 | ||
1390 | /* Count this entry */ | |
1391 | total += strlen (node->pat) + 1; | |
1392 | total += number_len ((long) node->lno) + 1 + number_len (node->cno) + 1; | |
fe0b3356 | 1393 | if (node->named) |
c6d46f5f JB |
1394 | total += 1 + strlen (node->name); /* \001name */ |
1395 | } | |
1396 | ||
1397 | return total; | |
1398 | } | |
1399 | \f | |
1400 | /* | |
1401 | * The C symbol tables. | |
1402 | */ | |
55597f90 FP |
1403 | enum sym_type |
1404 | { | |
1405 | st_none, st_C_struct, st_C_enum, st_C_define, st_C_typedef, st_C_typespec | |
1406 | }; | |
c6d46f5f | 1407 | |
42680d3c FP |
1408 | /* Feed stuff between (but not including) %[ and %] lines to: |
1409 | gperf -c -k1,3 -o -p -r -t | |
1410 | %[ | |
1411 | struct C_stab_entry { char *name; int c_ext; enum sym_type type; } | |
1412 | %% | |
1413 | class, C_PLPL, st_C_struct | |
1414 | domain, C_STAR, st_C_struct | |
1415 | union, 0, st_C_struct | |
1416 | struct, 0, st_C_struct | |
1417 | enum, 0, st_C_enum | |
1418 | typedef, 0, st_C_typedef | |
1419 | define, 0, st_C_define | |
1420 | long, 0, st_C_typespec | |
1421 | short, 0, st_C_typespec | |
1422 | int, 0, st_C_typespec | |
1423 | char, 0, st_C_typespec | |
1424 | float, 0, st_C_typespec | |
1425 | double, 0, st_C_typespec | |
1426 | signed, 0, st_C_typespec | |
1427 | unsigned, 0, st_C_typespec | |
1428 | auto, 0, st_C_typespec | |
1429 | void, 0, st_C_typespec | |
1430 | extern, 0, st_C_typespec | |
1431 | static, 0, st_C_typespec | |
1432 | const, 0, st_C_typespec | |
1433 | volatile, 0, st_C_typespec | |
1434 | %] | |
1435 | and replace lines between %< and %> with its output. */ | |
1436 | /*%<*/ | |
1437 | /* C code produced by gperf version 1.8.1 (K&R C version) */ | |
1438 | /* Command-line: gperf -c -k1,3 -o -p -r -t */ | |
1439 | ||
1440 | ||
1441 | struct C_stab_entry { char *name; int c_ext; enum sym_type type; }; | |
1442 | ||
1443 | #define MIN_WORD_LENGTH 3 | |
1444 | #define MAX_WORD_LENGTH 8 | |
1445 | #define MIN_HASH_VALUE 10 | |
1446 | #define MAX_HASH_VALUE 62 | |
c6d46f5f | 1447 | /* |
42680d3c FP |
1448 | 21 keywords |
1449 | 53 is the maximum key range | |
1450 | */ | |
1451 | ||
1452 | static int | |
1453 | hash (str, len) | |
1454 | register char *str; | |
1455 | register int len; | |
1456 | { | |
1457 | static unsigned char hash_table[] = | |
1458 | { | |
1459 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1460 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1461 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1462 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1463 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1464 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1465 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1466 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1467 | 62, 62, 62, 62, 62, 62, 62, 62, 62, 62, | |
1468 | 62, 62, 62, 62, 62, 62, 62, 2, 62, 7, | |
1469 | 6, 9, 15, 30, 62, 24, 62, 62, 1, 24, | |
1470 | 7, 27, 13, 62, 19, 26, 18, 27, 1, 62, | |
1471 | 62, 62, 62, 62, 62, 62, 62, 62, | |
1472 | }; | |
1473 | return len + hash_table[str[2]] + hash_table[str[0]]; | |
1474 | } | |
c6d46f5f | 1475 | |
42680d3c FP |
1476 | struct C_stab_entry * |
1477 | in_word_set (str, len) | |
1478 | register char *str; | |
1479 | register int len; | |
1480 | { | |
1481 | ||
1482 | static struct C_stab_entry wordlist[] = | |
1483 | { | |
1484 | {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, | |
1485 | {"",}, | |
1486 | {"volatile", 0, st_C_typespec}, | |
1487 | {"",}, | |
1488 | {"long", 0, st_C_typespec}, | |
1489 | {"char", 0, st_C_typespec}, | |
1490 | {"class", C_PLPL, st_C_struct}, | |
1491 | {"",}, {"",}, {"",}, {"",}, | |
1492 | {"const", 0, st_C_typespec}, | |
1493 | {"",}, {"",}, {"",}, {"",}, | |
1494 | {"auto", 0, st_C_typespec}, | |
1495 | {"",}, {"",}, | |
1496 | {"define", 0, st_C_define}, | |
1497 | {"",}, | |
1498 | {"void", 0, st_C_typespec}, | |
1499 | {"",}, {"",}, {"",}, | |
1500 | {"extern", 0, st_C_typespec}, | |
1501 | {"static", 0, st_C_typespec}, | |
1502 | {"",}, | |
1503 | {"domain", C_STAR, st_C_struct}, | |
1504 | {"",}, | |
1505 | {"typedef", 0, st_C_typedef}, | |
1506 | {"double", 0, st_C_typespec}, | |
1507 | {"enum", 0, st_C_enum}, | |
1508 | {"",}, {"",}, {"",}, {"",}, | |
1509 | {"int", 0, st_C_typespec}, | |
1510 | {"",}, | |
1511 | {"float", 0, st_C_typespec}, | |
1512 | {"",}, {"",}, {"",}, | |
1513 | {"struct", 0, st_C_struct}, | |
1514 | {"",}, {"",}, {"",}, {"",}, | |
1515 | {"union", 0, st_C_struct}, | |
1516 | {"",}, | |
1517 | {"short", 0, st_C_typespec}, | |
1518 | {"",}, {"",}, | |
1519 | {"unsigned", 0, st_C_typespec}, | |
1520 | {"signed", 0, st_C_typespec}, | |
1521 | }; | |
1522 | ||
1523 | if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH) | |
1524 | { | |
1525 | register int key = hash (str, len); | |
1526 | ||
1527 | if (key <= MAX_HASH_VALUE && key >= MIN_HASH_VALUE) | |
1528 | { | |
1529 | register char *s = wordlist[key].name; | |
1530 | ||
1a0d8c80 | 1531 | if (*s == *str && strneq (str + 1, s + 1, len - 1)) |
42680d3c FP |
1532 | return &wordlist[key]; |
1533 | } | |
1534 | } | |
1535 | return 0; | |
c6d46f5f | 1536 | } |
42680d3c | 1537 | /*%>*/ |
c6d46f5f | 1538 | |
42680d3c FP |
1539 | enum sym_type |
1540 | C_symtype(str, len, c_ext) | |
1541 | char *str; | |
1542 | int len; | |
c6d46f5f JB |
1543 | int c_ext; |
1544 | { | |
42680d3c | 1545 | register struct C_stab_entry *se = in_word_set(str, len); |
c6d46f5f | 1546 | |
42680d3c FP |
1547 | if (se == NULL || (se->c_ext && !(c_ext & se->c_ext))) |
1548 | return st_none; | |
1549 | return se->type; | |
c6d46f5f JB |
1550 | } |
1551 | \f | |
13fde0cd | 1552 | /* |
13fde0cd RS |
1553 | * C functions are recognized using a simple finite automaton. |
1554 | * funcdef is its state variable. | |
1555 | */ | |
1556 | typedef enum | |
1557 | { | |
31d4b314 FP |
1558 | fnone, /* nothing seen */ |
1559 | ftagseen, /* function-like tag seen */ | |
b12756c8 | 1560 | fstartlist, /* just after open parenthesis */ |
31d4b314 FP |
1561 | finlist, /* in parameter list */ |
1562 | flistseen, /* after parameter list */ | |
46e4cb76 | 1563 | fignore /* before open brace */ |
13fde0cd RS |
1564 | } FUNCST; |
1565 | FUNCST funcdef; | |
1566 | ||
1567 | ||
46c145db FP |
1568 | /* |
1569 | * typedefs are recognized using a simple finite automaton. | |
13fde0cd RS |
1570 | * typeddef is its state variable. |
1571 | */ | |
1572 | typedef enum | |
1573 | { | |
31d4b314 FP |
1574 | tnone, /* nothing seen */ |
1575 | ttypedseen, /* typedef keyword seen */ | |
1576 | tinbody, /* inside typedef body */ | |
46c145db FP |
1577 | tend, /* just before typedef tag */ |
1578 | tignore /* junk after typedef tag */ | |
13fde0cd RS |
1579 | } TYPEDST; |
1580 | TYPEDST typdef; | |
1581 | ||
1582 | ||
46c145db FP |
1583 | /* |
1584 | * struct-like structures (enum, struct and union) are recognized | |
1585 | * using another simple finite automaton. `structdef' is its state | |
1586 | * variable. | |
13fde0cd RS |
1587 | */ |
1588 | typedef enum | |
1589 | { | |
1590 | snone, /* nothing seen yet */ | |
1591 | skeyseen, /* struct-like keyword seen */ | |
1592 | stagseen, /* struct-like tag seen */ | |
1593 | scolonseen, /* colon seen after struct-like tag */ | |
46e4cb76 | 1594 | sinbody /* in struct body: recognize member func defs*/ |
13fde0cd RS |
1595 | } STRUCTST; |
1596 | STRUCTST structdef; | |
46c145db | 1597 | |
13fde0cd RS |
1598 | /* |
1599 | * When structdef is stagseen, scolonseen, or sinbody, structtag is the | |
42680d3c FP |
1600 | * struct tag, and structtype is the type of the preceding struct-like |
1601 | * keyword. | |
13fde0cd | 1602 | */ |
55597f90 | 1603 | char *structtag = "<uninited>"; |
42680d3c | 1604 | enum sym_type structtype; |
13fde0cd RS |
1605 | |
1606 | /* | |
1607 | * Yet another little state machine to deal with preprocessor lines. | |
1608 | */ | |
1609 | typedef enum | |
1610 | { | |
1611 | dnone, /* nothing seen */ | |
1612 | dsharpseen, /* '#' seen as first char on line */ | |
1613 | ddefineseen, /* '#' and 'define' seen */ | |
46e4cb76 | 1614 | dignorerest /* ignore rest of line */ |
13fde0cd RS |
1615 | } DEFINEST; |
1616 | DEFINEST definedef; | |
1617 | ||
1618 | /* | |
1619 | * Set this to TRUE, and the next token considered is called a function. | |
cdc1f6a7 | 1620 | * Used only for GNU emacs's function-defining macros. |
13fde0cd RS |
1621 | */ |
1622 | logical next_token_is_func; | |
1623 | ||
1624 | /* | |
1625 | * TRUE in the rules part of a yacc file, FALSE outside (parse as C). | |
1626 | */ | |
1627 | logical yacc_rules; | |
1628 | ||
6dd5561c FP |
1629 | /* |
1630 | * consider_token () | |
1631 | * checks to see if the current token is at the start of a | |
1632 | * function, or corresponds to a typedef, or is a struct/union/enum | |
1633 | * tag. | |
1634 | * | |
1635 | * *IS_FUNC gets TRUE iff the token is a function or macro with args. | |
1636 | * C_EXT is which language we are looking at. | |
1637 | * | |
1638 | * In the future we will need some way to adjust where the end of | |
1639 | * the token is; for instance, implementing the C++ keyword | |
1640 | * `operator' properly will adjust the end of the token to be after | |
1641 | * whatever follows `operator'. | |
1642 | * | |
1643 | * Globals | |
1644 | * funcdef IN OUT | |
1645 | * structdef IN OUT | |
1646 | * definedef IN OUT | |
1647 | * typdef IN OUT | |
1648 | * next_token_is_func IN OUT | |
1649 | */ | |
1650 | ||
1651 | logical | |
55597f90 FP |
1652 | consider_token (str, len, c, c_ext, cblev, is_func) |
1653 | register char *str; /* IN: token pointer */ | |
1654 | register int len; /* IN: token length */ | |
6dd5561c | 1655 | register char c; /* IN: first char after the token */ |
6dd5561c FP |
1656 | int c_ext; /* IN: C extensions mask */ |
1657 | int cblev; /* IN: curly brace level */ | |
715b6f8c | 1658 | logical *is_func; /* OUT: function found */ |
6dd5561c | 1659 | { |
55597f90 | 1660 | enum sym_type toktype = C_symtype (str, len, c_ext); |
6dd5561c FP |
1661 | |
1662 | /* | |
1663 | * Advance the definedef state machine. | |
1664 | */ | |
1665 | switch (definedef) | |
1666 | { | |
1667 | case dnone: | |
1668 | /* We're not on a preprocessor line. */ | |
1669 | break; | |
1670 | case dsharpseen: | |
1671 | if (toktype == st_C_define) | |
1672 | { | |
1673 | definedef = ddefineseen; | |
1674 | } | |
1675 | else | |
1676 | { | |
1677 | definedef = dignorerest; | |
1678 | } | |
b9755a12 | 1679 | return FALSE; |
6dd5561c FP |
1680 | case ddefineseen: |
1681 | /* | |
ee70dba5 FP |
1682 | * Make a tag for any macro, unless it is a constant |
1683 | * and constantypedefs is FALSE. | |
6dd5561c FP |
1684 | */ |
1685 | definedef = dignorerest; | |
1686 | *is_func = (c == '('); | |
1687 | if (!*is_func && !constantypedefs) | |
b9755a12 | 1688 | return FALSE; |
6dd5561c | 1689 | else |
b9755a12 | 1690 | return TRUE; |
6dd5561c | 1691 | case dignorerest: |
b9755a12 | 1692 | return FALSE; |
6dd5561c FP |
1693 | default: |
1694 | error ("internal error: definedef value.", 0); | |
1695 | } | |
1696 | ||
1697 | /* | |
1698 | * Now typedefs | |
1699 | */ | |
1700 | switch (typdef) | |
1701 | { | |
1702 | case tnone: | |
1703 | if (toktype == st_C_typedef) | |
1704 | { | |
1705 | if (typedefs) | |
1706 | typdef = ttypedseen; | |
1707 | funcdef = fnone; | |
b9755a12 | 1708 | return FALSE; |
6dd5561c FP |
1709 | } |
1710 | break; | |
1711 | case ttypedseen: | |
1712 | switch (toktype) | |
1713 | { | |
1714 | case st_none: | |
1715 | case st_C_typespec: | |
1716 | typdef = tend; | |
1717 | break; | |
1718 | case st_C_struct: | |
1719 | case st_C_enum: | |
1720 | break; | |
1721 | } | |
1722 | /* Do not return here, so the structdef stuff has a chance. */ | |
1723 | break; | |
1724 | case tend: | |
1725 | switch (toktype) | |
1726 | { | |
1727 | case st_C_typespec: | |
1728 | case st_C_struct: | |
1729 | case st_C_enum: | |
b9755a12 | 1730 | return FALSE; |
6dd5561c | 1731 | } |
b9755a12 | 1732 | return TRUE; |
6dd5561c FP |
1733 | } |
1734 | ||
1735 | /* | |
1736 | * This structdef business is currently only invoked when cblev==0. | |
1737 | * It should be recursively invoked whatever the curly brace level, | |
1738 | * and a stack of states kept, to allow for definitions of structs | |
1739 | * within structs. | |
1740 | * | |
1741 | * This structdef business is NOT invoked when we are ctags and the | |
1742 | * file is plain C. This is because a struct tag may have the same | |
1743 | * name as another tag, and this loses with ctags. | |
1744 | * | |
1745 | * This if statement deals with the typdef state machine as | |
1746 | * follows: if typdef==ttypedseen and token is struct/union/class/enum, | |
b9755a12 | 1747 | * return FALSE. All the other code here is for the structdef |
6dd5561c FP |
1748 | * state machine. |
1749 | */ | |
1750 | switch (toktype) | |
1751 | { | |
1752 | case st_C_struct: | |
1753 | case st_C_enum: | |
1754 | if (typdef == ttypedseen | |
1755 | || (typedefs_and_cplusplus && cblev == 0 && structdef == snone)) | |
1756 | { | |
1757 | structdef = skeyseen; | |
1758 | structtype = toktype; | |
1759 | } | |
b9755a12 | 1760 | return FALSE; |
6dd5561c FP |
1761 | } |
1762 | if (structdef == skeyseen) | |
1763 | { | |
55597f90 FP |
1764 | /* Save the tag for struct/union/class, for functions that may be |
1765 | defined inside. */ | |
6dd5561c | 1766 | if (structtype == st_C_struct) |
55597f90 | 1767 | structtag = savenstr (str, len); |
6dd5561c | 1768 | else |
55597f90 | 1769 | structtag = "<enum>"; |
6dd5561c | 1770 | structdef = stagseen; |
b9755a12 | 1771 | return TRUE; |
6dd5561c FP |
1772 | } |
1773 | ||
1774 | /* Avoid entering funcdef stuff if typdef is going on. */ | |
1775 | if (typdef != tnone) | |
1776 | { | |
1777 | definedef = dnone; | |
b9755a12 | 1778 | return FALSE; |
6dd5561c FP |
1779 | } |
1780 | ||
715b6f8c | 1781 | /* Detect GNU macros. */ |
6dd5561c | 1782 | if (definedef == dnone) |
79263656 | 1783 | if (strneq (str, "DEFUN", len) /* Used in emacs */ |
715b6f8c FP |
1784 | #if FALSE |
1785 | These are defined inside C functions, so currently they | |
1786 | are not met anyway. | |
79263656 | 1787 | || strneq (str, "EXFUN", len) /* Used in glibc */ |
55597f90 | 1788 | || strneq (str, "DEFVAR_", 7) /* Used in emacs */ |
715b6f8c | 1789 | #endif |
79263656 FP |
1790 | || strneq (str, "SYSCALL", len) /* Used in glibc (mach) */ |
1791 | || strneq (str, "ENTRY", len) /* Used in glibc */ | |
1792 | || strneq (str, "PSEUDO", len)) /* Used in glibc */ | |
715b6f8c FP |
1793 | |
1794 | { | |
1795 | next_token_is_func = TRUE; | |
b9755a12 | 1796 | return FALSE; |
715b6f8c | 1797 | } |
6dd5561c FP |
1798 | if (next_token_is_func) |
1799 | { | |
1800 | next_token_is_func = FALSE; | |
715b6f8c FP |
1801 | funcdef = fignore; |
1802 | *is_func = TRUE; | |
b9755a12 | 1803 | return TRUE; |
6dd5561c FP |
1804 | } |
1805 | ||
1806 | /* A function? */ | |
1807 | switch (toktype) | |
1808 | { | |
1809 | case st_C_typespec: | |
1810 | if (funcdef != finlist && funcdef != fignore) | |
1811 | funcdef = fnone; /* should be useless */ | |
b9755a12 | 1812 | return FALSE; |
6dd5561c FP |
1813 | default: |
1814 | if (funcdef == fnone) | |
1815 | { | |
1816 | funcdef = ftagseen; | |
1817 | *is_func = TRUE; | |
b9755a12 | 1818 | return TRUE; |
6dd5561c FP |
1819 | } |
1820 | } | |
1821 | ||
b9755a12 | 1822 | return FALSE; |
6dd5561c FP |
1823 | } |
1824 | ||
c6d46f5f JB |
1825 | /* |
1826 | * C_entries () | |
13fde0cd RS |
1827 | * This routine finds functions, typedefs, #define's and |
1828 | * struct/union/enum definitions in C syntax and adds them | |
c6d46f5f JB |
1829 | * to the list. |
1830 | */ | |
55597f90 FP |
1831 | typedef struct |
1832 | { | |
75bdbc6a | 1833 | logical valid; |
55597f90 FP |
1834 | char *str; |
1835 | logical named; | |
1836 | int linelen; | |
1837 | int lineno; | |
2bd88040 FP |
1838 | long linepos; |
1839 | char *buffer; | |
55597f90 FP |
1840 | } TOKEN; |
1841 | ||
1842 | #define current_lb_is_new (newndx == curndx) | |
1843 | #define switch_line_buffers() (curndx = 1 - curndx) | |
c6d46f5f | 1844 | |
13fde0cd RS |
1845 | #define curlb (lbs[curndx].lb) |
1846 | #define othlb (lbs[1-curndx].lb) | |
1847 | #define newlb (lbs[newndx].lb) | |
1848 | #define curlinepos (lbs[curndx].linepos) | |
1849 | #define othlinepos (lbs[1-curndx].linepos) | |
1850 | #define newlinepos (lbs[newndx].linepos) | |
1851 | ||
c6d46f5f | 1852 | #define CNL_SAVE_DEFINEDEF \ |
13fde0cd | 1853 | do { \ |
55597f90 | 1854 | curlinepos = charno; \ |
c6d46f5f | 1855 | lineno++; \ |
13fde0cd RS |
1856 | charno += readline (&curlb, inf); \ |
1857 | lp = curlb.buffer; \ | |
1858 | quotednl = FALSE; \ | |
1859 | newndx = curndx; \ | |
b9755a12 | 1860 | } while (0) |
c6d46f5f JB |
1861 | |
1862 | #define CNL \ | |
13fde0cd | 1863 | do { \ |
c6d46f5f | 1864 | CNL_SAVE_DEFINEDEF; \ |
75bdbc6a | 1865 | if (savetok.valid) \ |
55597f90 FP |
1866 | { \ |
1867 | tok = savetok; \ | |
75bdbc6a | 1868 | savetok.valid = FALSE; \ |
55597f90 | 1869 | } \ |
c6d46f5f | 1870 | definedef = dnone; \ |
b9755a12 | 1871 | } while (0) |
13fde0cd | 1872 | |
75bdbc6a FP |
1873 | #define make_tag(isfun) do \ |
1874 | { \ | |
1875 | if (tok.valid) \ | |
1876 | pfnote (savestr (token_name.buffer), isfun, tok.named, \ | |
1877 | tok.buffer, tok.linelen, tok.lineno, tok.linepos); \ | |
1878 | else if (DEBUG) abort (); \ | |
1879 | tok.valid = FALSE; \ | |
1880 | } while (0) | |
c6d46f5f JB |
1881 | |
1882 | void | |
6dd5561c | 1883 | C_entries (c_ext, inf) |
b9755a12 FP |
1884 | int c_ext; /* extension of C */ |
1885 | FILE *inf; /* input file */ | |
c6d46f5f | 1886 | { |
13fde0cd | 1887 | register char c; /* latest char read; '\0' for end of line */ |
c6d46f5f | 1888 | register char *lp; /* pointer one beyond the character `c' */ |
13fde0cd | 1889 | int curndx, newndx; /* indices for current and new lb */ |
55597f90 FP |
1890 | TOKEN tok; /* latest token read */ |
1891 | register int tokoff; /* offset in line of start of current token */ | |
1892 | register int toklen; /* length of current token */ | |
591fa824 | 1893 | int cblev; /* current curly brace level */ |
b12756c8 | 1894 | int parlev; /* current parenthesis level */ |
13fde0cd RS |
1895 | logical incomm, inquote, inchar, quotednl, midtoken; |
1896 | logical cplpl; | |
55597f90 | 1897 | TOKEN savetok; /* token saved during preprocessor handling */ |
c6d46f5f | 1898 | |
75bdbc6a | 1899 | |
13fde0cd | 1900 | curndx = newndx = 0; |
c6d46f5f JB |
1901 | lineno = 0; |
1902 | charno = 0; | |
13fde0cd | 1903 | lp = curlb.buffer; |
c6d46f5f JB |
1904 | *lp = 0; |
1905 | ||
46c145db | 1906 | definedef = dnone; funcdef = fnone; typdef = tnone; structdef = snone; |
75bdbc6a | 1907 | next_token_is_func = yacc_rules = FALSE; |
13fde0cd | 1908 | midtoken = inquote = inchar = incomm = quotednl = FALSE; |
75bdbc6a | 1909 | tok.valid = savetok.valid = FALSE; |
591fa824 | 1910 | cblev = 0; |
b12756c8 | 1911 | parlev = 0; |
13fde0cd | 1912 | cplpl = c_ext & C_PLPL; |
c6d46f5f | 1913 | |
c6d46f5f JB |
1914 | while (!feof (inf)) |
1915 | { | |
1916 | c = *lp++; | |
c6d46f5f JB |
1917 | if (c == '\\') |
1918 | { | |
4746118a JB |
1919 | /* If we're at the end of the line, the next character is a |
1920 | '\0'; don't skip it, because it's the thing that tells us | |
1921 | to read the next line. */ | |
13fde0cd | 1922 | if (*lp == '\0') |
99e0a2e0 | 1923 | { |
13fde0cd | 1924 | quotednl = TRUE; |
99e0a2e0 RS |
1925 | continue; |
1926 | } | |
1e134a5f | 1927 | lp++; |
c6d46f5f JB |
1928 | c = ' '; |
1929 | } | |
1930 | else if (incomm) | |
1931 | { | |
13fde0cd | 1932 | switch (c) |
c6d46f5f | 1933 | { |
13fde0cd RS |
1934 | case '*': |
1935 | if (*lp == '/') | |
1936 | { | |
1937 | c = *lp++; | |
1938 | incomm = FALSE; | |
1939 | } | |
1940 | break; | |
1941 | case '\0': | |
1942 | /* Newlines inside comments do not end macro definitions in | |
1943 | traditional cpp. */ | |
1944 | CNL_SAVE_DEFINEDEF; | |
1945 | break; | |
c6d46f5f | 1946 | } |
13fde0cd | 1947 | continue; |
c6d46f5f JB |
1948 | } |
1949 | else if (inquote) | |
1950 | { | |
13fde0cd RS |
1951 | switch (c) |
1952 | { | |
1953 | case '"': | |
1954 | inquote = FALSE; | |
1955 | break; | |
1956 | case '\0': | |
42680d3c | 1957 | /* Newlines inside strings do not end macro definitions |
13fde0cd RS |
1958 | in traditional cpp, even though compilers don't |
1959 | usually accept them. */ | |
1960 | CNL_SAVE_DEFINEDEF; | |
1961 | break; | |
1962 | } | |
1963 | continue; | |
c6d46f5f JB |
1964 | } |
1965 | else if (inchar) | |
1966 | { | |
42680d3c FP |
1967 | switch (c) |
1968 | { | |
1969 | case '\0': | |
1970 | /* Hmmm, something went wrong. */ | |
1971 | CNL; | |
1972 | /* FALLTHRU */ | |
1973 | case '\'': | |
46c145db | 1974 | inchar = FALSE; |
42680d3c FP |
1975 | break; |
1976 | } | |
c6d46f5f JB |
1977 | continue; |
1978 | } | |
13fde0cd | 1979 | else |
c6d46f5f JB |
1980 | switch (c) |
1981 | { | |
1982 | case '"': | |
1983 | inquote = TRUE; | |
b12756c8 FP |
1984 | if (funcdef != finlist && funcdef != fignore) |
1985 | funcdef = fnone; | |
c6d46f5f JB |
1986 | continue; |
1987 | case '\'': | |
1988 | inchar = TRUE; | |
b12756c8 FP |
1989 | if (funcdef != finlist && funcdef != fignore) |
1990 | funcdef = fnone; | |
c6d46f5f JB |
1991 | continue; |
1992 | case '/': | |
1993 | if (*lp == '*') | |
1994 | { | |
1995 | lp++; | |
1996 | incomm = TRUE; | |
13fde0cd | 1997 | continue; |
c6d46f5f | 1998 | } |
13fde0cd | 1999 | else if (cplpl && *lp == '/') |
c6d46f5f | 2000 | { |
daa37602 JB |
2001 | c = 0; |
2002 | break; | |
c6d46f5f | 2003 | } |
b12756c8 FP |
2004 | else |
2005 | break; | |
13fde0cd RS |
2006 | case '%': |
2007 | if ((c_ext & YACC) && *lp == '%') | |
2008 | { | |
2009 | /* entering or exiting rules section in yacc file */ | |
2010 | lp++; | |
2011 | definedef = dnone; funcdef = fnone; | |
46c145db | 2012 | typdef = tnone; structdef = snone; |
13fde0cd RS |
2013 | next_token_is_func = FALSE; |
2014 | midtoken = inquote = inchar = incomm = quotednl = FALSE; | |
591fa824 | 2015 | cblev = 0; |
13fde0cd RS |
2016 | yacc_rules = !yacc_rules; |
2017 | continue; | |
591fa824 | 2018 | } |
b12756c8 FP |
2019 | else |
2020 | break; | |
c6d46f5f | 2021 | case '#': |
ee70dba5 FP |
2022 | if (definedef == dnone) |
2023 | { | |
2024 | char *cp; | |
2025 | logical cpptoken = TRUE; | |
2026 | ||
2027 | /* Look back on this line. If all blanks, or nonblanks | |
2028 | followed by an end of comment, this is a preprocessor | |
2029 | token. */ | |
2030 | for (cp = newlb.buffer; cp < lp-1; cp++) | |
2031 | if (!iswhite (*cp)) | |
2032 | { | |
2033 | if (*cp == '*' && *(cp+1) == '/') | |
2034 | { | |
2035 | cp++; | |
2036 | cpptoken = TRUE; | |
2037 | } | |
2038 | else | |
2039 | cpptoken = FALSE; | |
2040 | } | |
2041 | if (cpptoken) | |
2042 | definedef = dsharpseen; | |
2043 | } /* if (definedef == dnone) */ | |
2044 | ||
c6d46f5f | 2045 | continue; |
13fde0cd | 2046 | } /* switch (c) */ |
c6d46f5f | 2047 | |
c6d46f5f | 2048 | |
591fa824 | 2049 | /* Consider token only if some complicated conditions are satisfied. */ |
ee70dba5 FP |
2050 | if ((definedef != dnone |
2051 | || (cblev == 0 && structdef != scolonseen) | |
591fa824 | 2052 | || (cblev == 1 && cplpl && structdef == sinbody)) |
46c145db | 2053 | && typdef != tignore |
13fde0cd | 2054 | && definedef != dignorerest |
ee70dba5 | 2055 | && funcdef != finlist) |
c6d46f5f JB |
2056 | { |
2057 | if (midtoken) | |
2058 | { | |
2059 | if (endtoken (c)) | |
2060 | { | |
b12756c8 | 2061 | if (cplpl && c == ':' && *lp == ':' && begtoken(*(lp + 1))) |
c6d46f5f JB |
2062 | { |
2063 | /* | |
ee70dba5 FP |
2064 | * This handles :: in the middle, but not at the |
2065 | * beginning of an identifier. | |
c6d46f5f JB |
2066 | */ |
2067 | lp += 2; | |
2068 | toklen += 3; | |
2069 | } | |
2070 | else | |
2071 | { | |
fe0b3356 | 2072 | logical is_func = FALSE; |
c6d46f5f | 2073 | |
13fde0cd | 2074 | if (yacc_rules |
55597f90 FP |
2075 | || consider_token (newlb.buffer + tokoff, toklen, |
2076 | c, c_ext, cblev, &is_func)) | |
c6d46f5f | 2077 | { |
99e0a2e0 | 2078 | if (structdef == sinbody |
fe0b3356 FP |
2079 | && definedef == dnone |
2080 | && is_func) | |
2081 | /* function defined in C++ class body */ | |
2082 | { | |
2bd88040 | 2083 | int strsize = strlen(structtag) + 2 + toklen + 1; |
75bdbc6a | 2084 | while (token_name.size < strsize) |
2bd88040 | 2085 | { |
75bdbc6a | 2086 | token_name.size *= 2; |
03cdafdf RS |
2087 | token_name.buffer |
2088 | = (char *) xrealloc (token_name.buffer, | |
2089 | token_name.size); | |
2bd88040 | 2090 | } |
75bdbc6a FP |
2091 | strcpy (token_name.buffer, structtag); |
2092 | strcat (token_name.buffer, "::"); | |
2093 | strncat (token_name.buffer, | |
2bd88040 | 2094 | newlb.buffer+tokoff, toklen); |
ee70dba5 | 2095 | tok.named = TRUE; |
c6d46f5f JB |
2096 | } |
2097 | else | |
2098 | { | |
75bdbc6a | 2099 | while (token_name.size < toklen + 1) |
2bd88040 | 2100 | { |
75bdbc6a | 2101 | token_name.size *= 2; |
03cdafdf RS |
2102 | token_name.buffer |
2103 | = (char *) xrealloc (token_name.buffer, | |
2104 | token_name.size); | |
2bd88040 | 2105 | } |
75bdbc6a | 2106 | strncpy (token_name.buffer, |
2bd88040 | 2107 | newlb.buffer+tokoff, toklen); |
75bdbc6a | 2108 | token_name.buffer[toklen] = '\0'; |
55597f90 FP |
2109 | if (structdef == stagseen |
2110 | || typdef == tend | |
2111 | || (is_func | |
2112 | && definedef == dignorerest)) /* macro */ | |
2113 | tok.named = TRUE; | |
2114 | else | |
2115 | tok.named = FALSE; | |
c6d46f5f | 2116 | } |
55597f90 FP |
2117 | tok.lineno = lineno; |
2118 | tok.linelen = tokoff + toklen + 1; | |
2bd88040 FP |
2119 | tok.buffer = newlb.buffer; |
2120 | tok.linepos = newlinepos; | |
75bdbc6a | 2121 | tok.valid = TRUE; |
fe0b3356 | 2122 | |
b12756c8 FP |
2123 | if (definedef == dnone |
2124 | && (funcdef == ftagseen | |
2125 | || structdef == stagseen | |
2126 | || typdef == tend)) | |
13fde0cd | 2127 | { |
55597f90 FP |
2128 | if (current_lb_is_new) |
2129 | switch_line_buffers (); | |
13fde0cd RS |
2130 | } |
2131 | else | |
2bd88040 | 2132 | make_tag (is_func); |
c6d46f5f JB |
2133 | } |
2134 | midtoken = FALSE; | |
2135 | } | |
13fde0cd | 2136 | } /* if (endtoken (c)) */ |
c6d46f5f | 2137 | else if (intoken (c)) |
13fde0cd RS |
2138 | { |
2139 | toklen++; | |
2140 | continue; | |
2141 | } | |
2142 | } /* if (midtoken) */ | |
c6d46f5f JB |
2143 | else if (begtoken (c)) |
2144 | { | |
b12756c8 | 2145 | switch (definedef) |
13fde0cd | 2146 | { |
b12756c8 FP |
2147 | case dnone: |
2148 | switch (funcdef) | |
2149 | { | |
2150 | case fstartlist: | |
2151 | funcdef = finlist; | |
2152 | continue; | |
2153 | case flistseen: | |
2bd88040 | 2154 | make_tag (TRUE); |
b12756c8 FP |
2155 | funcdef = fignore; |
2156 | break; | |
2157 | case ftagseen: | |
2158 | funcdef = fnone; | |
2159 | break; | |
2160 | } | |
2161 | if (structdef == stagseen) | |
2162 | structdef = snone; | |
13fde0cd | 2163 | break; |
b12756c8 | 2164 | case dsharpseen: |
4b533b5b | 2165 | savetok = tok; |
13fde0cd | 2166 | } |
13fde0cd RS |
2167 | if (!yacc_rules || lp == newlb.buffer + 1) |
2168 | { | |
2169 | tokoff = lp - 1 - newlb.buffer; | |
2170 | toklen = 1; | |
2171 | midtoken = TRUE; | |
2172 | } | |
2173 | continue; | |
4b533b5b | 2174 | } /* if (begtoken) */ |
13fde0cd RS |
2175 | } /* if must look at token */ |
2176 | ||
2177 | ||
2178 | /* Detect end of line, colon, comma, semicolon and various braces | |
b12756c8 | 2179 | after having handled a token.*/ |
13fde0cd | 2180 | switch (c) |
1e134a5f | 2181 | { |
13fde0cd | 2182 | case ':': |
b12756c8 FP |
2183 | if (definedef != dnone) |
2184 | break; | |
13fde0cd RS |
2185 | if (structdef == stagseen) |
2186 | structdef = scolonseen; | |
b12756c8 FP |
2187 | else |
2188 | switch (funcdef) | |
2189 | { | |
2190 | case ftagseen: | |
2191 | if (yacc_rules) | |
2192 | { | |
2bd88040 | 2193 | make_tag (FALSE); |
b12756c8 FP |
2194 | funcdef = fignore; |
2195 | } | |
2196 | break; | |
2197 | case fstartlist: | |
2198 | funcdef = fnone; | |
2199 | break; | |
2200 | } | |
13fde0cd RS |
2201 | break; |
2202 | case ';': | |
b12756c8 FP |
2203 | if (definedef != dnone) |
2204 | break; | |
46c145db FP |
2205 | if (cblev == 0) |
2206 | switch (typdef) | |
2207 | { | |
2208 | case tend: | |
2bd88040 | 2209 | make_tag (FALSE); |
46c145db FP |
2210 | /* FALLTHRU */ |
2211 | default: | |
2212 | typdef = tnone; | |
2213 | } | |
31d4b314 FP |
2214 | if (funcdef != fignore) |
2215 | funcdef = fnone; | |
46c145db FP |
2216 | if (structdef == stagseen) |
2217 | structdef = snone; | |
2218 | break; | |
13fde0cd | 2219 | case ',': |
46c145db FP |
2220 | if (definedef != dnone) |
2221 | break; | |
2222 | if (funcdef != finlist && funcdef != fignore) | |
2223 | funcdef = fnone; | |
2224 | if (structdef == stagseen) | |
2225 | structdef = snone; | |
2226 | break; | |
13fde0cd | 2227 | case '[': |
b12756c8 FP |
2228 | if (definedef != dnone) |
2229 | break; | |
46c145db FP |
2230 | if (cblev == 0 && typdef == tend) |
2231 | { | |
2232 | typdef = tignore; | |
2bd88040 | 2233 | make_tag (FALSE); |
46c145db FP |
2234 | break; |
2235 | } | |
31d4b314 | 2236 | if (funcdef != finlist && funcdef != fignore) |
13fde0cd RS |
2237 | funcdef = fnone; |
2238 | if (structdef == stagseen) | |
2239 | structdef = snone; | |
2240 | break; | |
2241 | case '(': | |
b12756c8 FP |
2242 | if (definedef != dnone) |
2243 | break; | |
13fde0cd | 2244 | switch (funcdef) |
57e83cfe | 2245 | { |
ee70dba5 FP |
2246 | case fnone: |
2247 | switch (typdef) | |
2248 | { | |
2249 | case ttypedseen: | |
2250 | case tend: | |
2251 | /* Make sure that the next char is not a '*'. | |
2252 | This handles constructs like: | |
2253 | typedef void OperatorFun (int fun); */ | |
2254 | if (*lp != '*') | |
2255 | { | |
2256 | typdef = tignore; | |
2bd88040 | 2257 | make_tag (FALSE); |
ee70dba5 FP |
2258 | } |
2259 | break; | |
2260 | } /* switch (typdef) */ | |
2261 | break; | |
13fde0cd | 2262 | case ftagseen: |
b12756c8 | 2263 | funcdef = fstartlist; |
13fde0cd | 2264 | break; |
13fde0cd | 2265 | case flistseen: |
b12756c8 | 2266 | funcdef = finlist; |
13fde0cd | 2267 | break; |
57e83cfe | 2268 | } |
b12756c8 | 2269 | parlev++; |
13fde0cd RS |
2270 | break; |
2271 | case ')': | |
b12756c8 FP |
2272 | if (definedef != dnone) |
2273 | break; | |
2274 | if (--parlev == 0) | |
2275 | { | |
2276 | switch (funcdef) | |
2277 | { | |
2278 | case fstartlist: | |
2279 | case finlist: | |
2280 | funcdef = flistseen; | |
2281 | break; | |
2282 | } | |
46c145db FP |
2283 | if (cblev == 0 && typdef == tend) |
2284 | { | |
2285 | typdef = tignore; | |
2bd88040 | 2286 | make_tag (FALSE); |
46c145db | 2287 | } |
b12756c8 FP |
2288 | } |
2289 | else if (parlev < 0) /* can happen due to ill-conceived #if's. */ | |
2290 | parlev = 0; | |
13fde0cd RS |
2291 | break; |
2292 | case '{': | |
b12756c8 FP |
2293 | if (definedef != dnone) |
2294 | break; | |
13fde0cd RS |
2295 | if (typdef == ttypedseen) |
2296 | typdef = tinbody; | |
2297 | switch (structdef) | |
2298 | { | |
2299 | case skeyseen: /* unnamed struct */ | |
55597f90 | 2300 | structtag = "_anonymous_"; |
13fde0cd RS |
2301 | structdef = sinbody; |
2302 | break; | |
2303 | case stagseen: | |
2304 | case scolonseen: /* named struct */ | |
2305 | structdef = sinbody; | |
2bd88040 | 2306 | make_tag (FALSE); |
13fde0cd RS |
2307 | break; |
2308 | } | |
31d4b314 FP |
2309 | switch (funcdef) |
2310 | { | |
2311 | case flistseen: | |
2bd88040 | 2312 | make_tag (TRUE); |
31d4b314 FP |
2313 | /* FALLTHRU */ |
2314 | case fignore: | |
2315 | funcdef = fnone; | |
46c145db FP |
2316 | break; |
2317 | case fnone: | |
715b6f8c | 2318 | /* Neutralize `extern "C" {' grot and look inside structs. */ |
46c145db | 2319 | if (cblev == 0 && structdef == snone && typdef == tnone) |
715b6f8c | 2320 | cblev = -1; |
31d4b314 | 2321 | } |
591fa824 | 2322 | cblev++; |
31d4b314 | 2323 | break; |
13fde0cd | 2324 | case '*': |
b12756c8 FP |
2325 | if (definedef != dnone) |
2326 | break; | |
2327 | if (funcdef == fstartlist) | |
2328 | funcdef = fnone; /* avoid tagging `foo' in `foo (*bar()) ()' */ | |
13fde0cd RS |
2329 | break; |
2330 | case '}': | |
b12756c8 FP |
2331 | if (definedef != dnone) |
2332 | break; | |
13fde0cd | 2333 | if (!noindentypedefs && lp == newlb.buffer + 1) |
b12756c8 FP |
2334 | { |
2335 | cblev = 0; /* reset curly brace level if first column */ | |
2336 | parlev = 0; /* also reset paren level, just in case... */ | |
2337 | } | |
591fa824 RS |
2338 | else if (cblev > 0) |
2339 | cblev--; | |
2340 | if (cblev == 0) | |
13fde0cd RS |
2341 | { |
2342 | if (typdef == tinbody) | |
2343 | typdef = tend; | |
55597f90 FP |
2344 | #if FALSE /* too risky */ |
2345 | if (structdef == sinbody) | |
2346 | free (structtag); | |
2347 | #endif | |
9cb0aa73 | 2348 | |
13fde0cd | 2349 | structdef = snone; |
55597f90 | 2350 | structtag = "<error>"; |
13fde0cd RS |
2351 | } |
2352 | break; | |
b12756c8 | 2353 | case '=': |
42680d3c FP |
2354 | case '#': case '+': case '-': case '~': case '&': case '%': case '/': |
2355 | case '|': case '^': case '!': case '<': case '>': case '.': case '?': | |
b12756c8 FP |
2356 | if (definedef != dnone) |
2357 | break; | |
2358 | /* These surely cannot follow a function tag. */ | |
2359 | if (funcdef != finlist && funcdef != fignore) | |
2360 | funcdef = fnone; | |
2361 | break; | |
13fde0cd RS |
2362 | case '\0': |
2363 | /* If a macro spans multiple lines don't reset its state. */ | |
2364 | if (quotednl) | |
2365 | CNL_SAVE_DEFINEDEF; | |
2366 | else | |
2367 | CNL; | |
2368 | break; | |
2369 | } /* switch (c) */ | |
2370 | ||
2371 | } /* while not eof */ | |
c6d46f5f | 2372 | } |
b9755a12 FP |
2373 | |
2374 | /* | |
2375 | * Process either a C++ file or a C file depending on the setting | |
2376 | * of a global flag. | |
2377 | */ | |
2378 | void | |
2379 | default_C_entries (inf) | |
2380 | FILE *inf; | |
2381 | { | |
2382 | C_entries (cplusplus ? C_PLPL : 0, inf); | |
2383 | } | |
2384 | ||
79263656 FP |
2385 | /* Always do plain ANSI C. */ |
2386 | void | |
2387 | plain_C_entries (inf) | |
2388 | FILE *inf; | |
2389 | { | |
2390 | C_entries (0, inf); | |
2391 | } | |
2392 | ||
b9755a12 FP |
2393 | /* Always do C++. */ |
2394 | void | |
2395 | Cplusplus_entries (inf) | |
2396 | FILE *inf; | |
2397 | { | |
2398 | C_entries (C_PLPL, inf); | |
2399 | } | |
2400 | ||
2401 | /* Always do C*. */ | |
2402 | void | |
2403 | Cstar_entries (inf) | |
2404 | FILE *inf; | |
2405 | { | |
2406 | C_entries (C_STAR, inf); | |
2407 | } | |
2408 | ||
2409 | /* Always do Yacc. */ | |
2410 | void | |
2411 | Yacc_entries (inf) | |
2412 | FILE *inf; | |
2413 | { | |
2414 | C_entries (YACC, inf); | |
2415 | } | |
6dd5561c FP |
2416 | \f |
2417 | /* Fortran parsing */ | |
c6d46f5f | 2418 | |
6dd5561c | 2419 | char *dbp; |
c6d46f5f JB |
2420 | |
2421 | logical | |
6dd5561c FP |
2422 | tail (cp) |
2423 | char *cp; | |
c6d46f5f | 2424 | { |
6dd5561c | 2425 | register int len = 0; |
c6d46f5f | 2426 | |
79263656 | 2427 | while (*cp && lowcase(*cp) == lowcase(dbp[len])) |
6dd5561c | 2428 | cp++, len++; |
79263656 | 2429 | if (*cp == 0 && !intoken(dbp[len])) |
c6d46f5f | 2430 | { |
6dd5561c | 2431 | dbp += len; |
b9755a12 | 2432 | return TRUE; |
c6d46f5f | 2433 | } |
b9755a12 | 2434 | return FALSE; |
6dd5561c | 2435 | } |
13fde0cd | 2436 | |
6dd5561c FP |
2437 | void |
2438 | takeprec () | |
2439 | { | |
2440 | while (isspace (*dbp)) | |
2441 | dbp++; | |
2442 | if (*dbp != '*') | |
2443 | return; | |
2444 | dbp++; | |
2445 | while (isspace (*dbp)) | |
2446 | dbp++; | |
79263656 FP |
2447 | if (strneq (dbp, "(*)", 3)) |
2448 | { | |
2449 | dbp += 3; | |
2450 | return; | |
2451 | } | |
6dd5561c | 2452 | if (!isdigit (*dbp)) |
c6d46f5f | 2453 | { |
6dd5561c FP |
2454 | --dbp; /* force failure */ |
2455 | return; | |
c6d46f5f | 2456 | } |
6dd5561c FP |
2457 | do |
2458 | dbp++; | |
2459 | while (isdigit (*dbp)); | |
2460 | } | |
13fde0cd | 2461 | |
6dd5561c FP |
2462 | void |
2463 | getit (inf) | |
2464 | FILE *inf; | |
2465 | { | |
2466 | register char *cp; | |
13fde0cd | 2467 | |
6dd5561c FP |
2468 | while (isspace (*dbp)) |
2469 | dbp++; | |
2470 | if (*dbp == '\0') | |
c6d46f5f | 2471 | { |
6dd5561c FP |
2472 | lineno++; |
2473 | linecharno = charno; | |
2474 | charno += readline (&lb, inf); | |
2475 | dbp = lb.buffer; | |
2476 | if (dbp[5] != '&') | |
2477 | return; | |
2478 | dbp += 6; | |
2479 | while (isspace (*dbp)) | |
2480 | dbp++; | |
c6d46f5f | 2481 | } |
6dd5561c FP |
2482 | if (!isalpha (*dbp) |
2483 | && *dbp != '_' | |
2484 | && *dbp != '$') | |
2485 | return; | |
2486 | for (cp = dbp + 1; | |
2487 | (*cp | |
2488 | && (isalpha (*cp) || isdigit (*cp) || (*cp == '_') || (*cp == '$'))); | |
2489 | cp++) | |
2490 | continue; | |
55597f90 | 2491 | pfnote (savenstr (dbp, cp-dbp), TRUE, FALSE, lb.buffer, |
6dd5561c | 2492 | cp - lb.buffer + 1, lineno, linecharno); |
c6d46f5f | 2493 | } |
c6d46f5f | 2494 | |
b9755a12 | 2495 | void |
6dd5561c FP |
2496 | Fortran_functions (inf) |
2497 | FILE *inf; | |
c6d46f5f JB |
2498 | { |
2499 | lineno = 0; | |
2500 | charno = 0; | |
c6d46f5f | 2501 | |
6dd5561c | 2502 | while (!feof (inf)) |
c6d46f5f JB |
2503 | { |
2504 | lineno++; | |
2505 | linecharno = charno; | |
6dd5561c | 2506 | charno += readline (&lb, inf); |
c6d46f5f JB |
2507 | dbp = lb.buffer; |
2508 | if (*dbp == '%') | |
2509 | dbp++; /* Ratfor escape to fortran */ | |
2510 | while (isspace (*dbp)) | |
2511 | dbp++; | |
2512 | if (*dbp == 0) | |
2513 | continue; | |
79263656 | 2514 | switch (lowcase (*dbp)) |
c6d46f5f JB |
2515 | { |
2516 | case 'i': | |
2517 | if (tail ("integer")) | |
2518 | takeprec (); | |
2519 | break; | |
2520 | case 'r': | |
2521 | if (tail ("real")) | |
2522 | takeprec (); | |
2523 | break; | |
2524 | case 'l': | |
2525 | if (tail ("logical")) | |
2526 | takeprec (); | |
2527 | break; | |
2528 | case 'c': | |
2529 | if (tail ("complex") || tail ("character")) | |
2530 | takeprec (); | |
2531 | break; | |
2532 | case 'd': | |
2533 | if (tail ("double")) | |
2534 | { | |
2535 | while (isspace (*dbp)) | |
2536 | dbp++; | |
2537 | if (*dbp == 0) | |
2538 | continue; | |
2539 | if (tail ("precision")) | |
2540 | break; | |
2541 | continue; | |
2542 | } | |
2543 | break; | |
2544 | } | |
2545 | while (isspace (*dbp)) | |
2546 | dbp++; | |
2547 | if (*dbp == 0) | |
2548 | continue; | |
79263656 | 2549 | switch (lowcase (*dbp)) |
c6d46f5f JB |
2550 | { |
2551 | case 'f': | |
2552 | if (tail ("function")) | |
6dd5561c | 2553 | getit (inf); |
c6d46f5f JB |
2554 | continue; |
2555 | case 's': | |
2556 | if (tail ("subroutine")) | |
6dd5561c | 2557 | getit (inf); |
c6d46f5f | 2558 | continue; |
8a6c8bcf RS |
2559 | case 'e': |
2560 | if (tail ("entry")) | |
6dd5561c | 2561 | getit (inf); |
8a6c8bcf | 2562 | continue; |
c6d46f5f JB |
2563 | case 'p': |
2564 | if (tail ("program")) | |
2565 | { | |
6dd5561c | 2566 | getit (inf); |
c6d46f5f JB |
2567 | continue; |
2568 | } | |
2569 | if (tail ("procedure")) | |
6dd5561c | 2570 | getit (inf); |
c6d46f5f JB |
2571 | continue; |
2572 | } | |
2573 | } | |
c6d46f5f | 2574 | } |
6dd5561c FP |
2575 | \f |
2576 | /* | |
2577 | * Bob Weiner, Motorola Inc., 4/3/94 | |
2578 | * Unix and microcontroller assembly tag handling | |
2579 | * look for '^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]' | |
2580 | */ | |
c6d46f5f | 2581 | void |
6dd5561c FP |
2582 | Asm_labels (inf) |
2583 | FILE *inf; | |
c6d46f5f JB |
2584 | { |
2585 | register char *cp; | |
c6d46f5f JB |
2586 | |
2587 | lineno = 0; | |
2588 | charno = 0; | |
c6d46f5f | 2589 | |
6dd5561c | 2590 | while (!feof (inf)) |
c6d46f5f JB |
2591 | { |
2592 | lineno++; | |
2593 | linecharno = charno; | |
6dd5561c FP |
2594 | charno += readline (&lb, inf); |
2595 | cp = lb.buffer; | |
2596 | ||
2597 | /* If first char is alphabetic or one of [_.$], test for colon | |
2598 | following identifier. */ | |
2599 | if (isalpha (*cp) || *cp == '_' || *cp == '.' || *cp == '$') | |
2600 | { | |
2601 | /* Read past label. */ | |
2602 | cp++; | |
2603 | while (isalnum (*cp) || *cp == '_' || *cp == '.' || *cp == '$') | |
2604 | cp++; | |
2605 | if (*cp == ':' || isspace (*cp)) | |
2606 | { | |
2607 | /* Found end of label, so copy it and add it to the table. */ | |
55597f90 FP |
2608 | pfnote (savenstr (lb.buffer, cp-lb.buffer), TRUE, FALSE, |
2609 | lb.buffer, cp - lb.buffer + 1, lineno, linecharno); | |
6dd5561c FP |
2610 | } |
2611 | } | |
c6d46f5f JB |
2612 | } |
2613 | } | |
2614 | \f | |
2615 | /* Added by Mosur Mohan, 4/22/88 */ | |
2616 | /* Pascal parsing */ | |
2617 | ||
2618 | #define GET_NEW_LINE \ | |
2619 | { \ | |
2620 | linecharno = charno; lineno++; \ | |
2621 | charno += 1 + readline (&lb, inf); \ | |
2622 | dbp = lb.buffer; \ | |
2623 | } | |
2624 | ||
aab1fdae FP |
2625 | /* |
2626 | * Locates tags for procedures & functions. Doesn't do any type- or | |
2627 | * var-definitions. It does look for the keyword "extern" or | |
2628 | * "forward" immediately following the procedure statement; if found, | |
2629 | * the tag is skipped. | |
c6d46f5f | 2630 | */ |
c6d46f5f | 2631 | void |
6dd5561c FP |
2632 | Pascal_functions (inf) |
2633 | FILE *inf; | |
c6d46f5f JB |
2634 | { |
2635 | struct linebuffer tline; /* mostly copied from C_entries */ | |
2636 | long save_lcno; | |
2637 | int save_lineno; | |
2638 | char c, *cp; | |
55597f90 | 2639 | char *nambuf; |
c6d46f5f JB |
2640 | |
2641 | logical /* each of these flags is TRUE iff: */ | |
b9755a12 | 2642 | incomment, /* point is inside a comment */ |
c6d46f5f JB |
2643 | inquote, /* point is inside '..' string */ |
2644 | get_tagname, /* point is after PROCEDURE/FUNCTION */ | |
2645 | /* keyword, so next item = potential tag */ | |
2646 | found_tag, /* point is after a potential tag */ | |
2647 | inparms, /* point is within parameter-list */ | |
2648 | verify_tag; /* point has passed the parm-list, so the */ | |
2649 | /* next token will determine whether */ | |
2650 | /* this is a FORWARD/EXTERN to be */ | |
2651 | /* ignored, or whether it is a real tag */ | |
2652 | ||
2653 | lineno = 0; | |
2654 | charno = 0; | |
2655 | dbp = lb.buffer; | |
2656 | *dbp = 0; | |
2657 | initbuffer (&tline); | |
2658 | ||
b9755a12 | 2659 | incomment = inquote = FALSE; |
c6d46f5f JB |
2660 | found_tag = FALSE; /* have a proc name; check if extern */ |
2661 | get_tagname = FALSE; /* have found "procedure" keyword */ | |
2662 | inparms = FALSE; /* found '(' after "proc" */ | |
2663 | verify_tag = FALSE; /* check if "extern" is ahead */ | |
2664 | ||
2665 | /* long main loop to get next char */ | |
6dd5561c | 2666 | while (!feof (inf)) |
c6d46f5f JB |
2667 | { |
2668 | c = *dbp++; | |
55597f90 | 2669 | if (c == '\0') /* if end of line */ |
c6d46f5f JB |
2670 | { |
2671 | GET_NEW_LINE; | |
55597f90 | 2672 | if (*dbp == '\0') |
c6d46f5f JB |
2673 | continue; |
2674 | if (!((found_tag && verify_tag) || | |
2675 | get_tagname)) | |
2676 | c = *dbp++; /* only if don't need *dbp pointing */ | |
55597f90 FP |
2677 | /* to the beginning of the name of */ |
2678 | /* the procedure or function */ | |
c6d46f5f | 2679 | } |
b9755a12 | 2680 | if (incomment) |
c6d46f5f | 2681 | { |
b9755a12 FP |
2682 | if (c == '}') /* within { - } comments */ |
2683 | incomment = FALSE; | |
2684 | else if (c == '*' && dbp[1] == ')') /* within (* - *) comments */ | |
c6d46f5f | 2685 | { |
b9755a12 FP |
2686 | dbp++; |
2687 | incomment = FALSE; | |
c6d46f5f JB |
2688 | } |
2689 | continue; | |
2690 | } | |
2691 | else if (inquote) | |
2692 | { | |
2693 | if (c == '\'') | |
2694 | inquote = FALSE; | |
2695 | continue; | |
2696 | } | |
55597f90 | 2697 | else |
c6d46f5f JB |
2698 | switch (c) |
2699 | { | |
2700 | case '\'': | |
2701 | inquote = TRUE; /* found first quote */ | |
2702 | continue; | |
2703 | case '{': /* found open-{-comment */ | |
b9755a12 | 2704 | incomment = TRUE; |
c6d46f5f JB |
2705 | continue; |
2706 | case '(': | |
2707 | if (*dbp == '*') /* found open-(*-comment */ | |
2708 | { | |
b9755a12 | 2709 | incomment = TRUE; |
c6d46f5f JB |
2710 | dbp++; |
2711 | } | |
2712 | else if (found_tag) /* found '(' after tag, i.e., parm-list */ | |
2713 | inparms = TRUE; | |
2714 | continue; | |
2715 | case ')': /* end of parms list */ | |
2716 | if (inparms) | |
2717 | inparms = FALSE; | |
2718 | continue; | |
2719 | case ';': | |
2720 | if ((found_tag) && (!inparms)) /* end of proc or fn stmt */ | |
2721 | { | |
2722 | verify_tag = TRUE; | |
2723 | break; | |
2724 | } | |
2725 | continue; | |
2726 | } | |
2727 | if ((found_tag) && (verify_tag) && (*dbp != ' ')) | |
2728 | { | |
2729 | /* check if this is an "extern" declaration */ | |
2730 | if (*dbp == 0) | |
2731 | continue; | |
2732 | if ((*dbp == 'e') || (*dbp == 'E')) | |
2733 | { | |
2734 | if (tail ("extern")) /* superfluous, really! */ | |
2735 | { | |
2736 | found_tag = FALSE; | |
2737 | verify_tag = FALSE; | |
2738 | } | |
2739 | } | |
2740 | else if ((*dbp == 'f') || (*dbp == 'F')) | |
2741 | { | |
2742 | if (tail ("forward")) /* check for forward reference */ | |
2743 | { | |
2744 | found_tag = FALSE; | |
2745 | verify_tag = FALSE; | |
2746 | } | |
2747 | } | |
46c145db | 2748 | if ((found_tag) && (verify_tag)) /* not external proc, so make tag */ |
c6d46f5f JB |
2749 | { |
2750 | found_tag = FALSE; | |
2751 | verify_tag = FALSE; | |
55597f90 FP |
2752 | pfnote (nambuf, TRUE, FALSE, tline.buffer, |
2753 | cp - tline.buffer + 1, save_lineno, save_lcno); | |
c6d46f5f JB |
2754 | continue; |
2755 | } | |
2756 | } | |
2757 | if (get_tagname) /* grab name of proc or fn */ | |
2758 | { | |
2759 | if (*dbp == 0) | |
2760 | continue; | |
2761 | ||
2762 | /* save all values for later tagging */ | |
2763 | tline.size = lb.size; | |
2764 | strcpy (tline.buffer, lb.buffer); | |
2765 | save_lineno = lineno; | |
2766 | save_lcno = linecharno; | |
2767 | ||
2768 | /* grab block name */ | |
2769 | for (cp = dbp + 1; *cp && (!endtoken (*cp)); cp++) | |
2770 | continue; | |
55597f90 | 2771 | nambuf = savenstr (dbp, cp-dbp); |
c6d46f5f JB |
2772 | dbp = cp; /* restore dbp to e-o-token */ |
2773 | get_tagname = FALSE; | |
2774 | found_tag = TRUE; | |
2775 | continue; | |
2776 | ||
2777 | /* and proceed to check for "extern" */ | |
2778 | } | |
55597f90 | 2779 | else if (!incomment && !inquote && !found_tag) |
c6d46f5f JB |
2780 | { |
2781 | /* check for proc/fn keywords */ | |
79263656 | 2782 | switch (lowcase (c)) |
c6d46f5f JB |
2783 | { |
2784 | case 'p': | |
2785 | if (tail ("rocedure")) /* c = 'p', dbp has advanced */ | |
2786 | get_tagname = TRUE; | |
2787 | continue; | |
2788 | case 'f': | |
2789 | if (tail ("unction")) | |
2790 | get_tagname = TRUE; | |
2791 | continue; | |
2792 | } | |
2793 | } | |
6dd5561c | 2794 | } /* while not eof */ |
c6d46f5f JB |
2795 | } |
2796 | \f | |
2797 | /* | |
2798 | * lisp tag functions | |
55597f90 | 2799 | * look for (def or (DEF, quote or QUOTE |
c6d46f5f | 2800 | */ |
c6d46f5f | 2801 | int |
55597f90 FP |
2802 | L_isdef (strp) |
2803 | register char *strp; | |
c6d46f5f | 2804 | { |
55597f90 FP |
2805 | return ((strp[1] == 'd' || strp[1] == 'D') |
2806 | && (strp[2] == 'e' || strp[2] == 'E') | |
2807 | && (strp[3] == 'f' || strp[3] == 'F')); | |
31d4b314 FP |
2808 | } |
2809 | ||
2810 | int | |
55597f90 FP |
2811 | L_isquote (strp) |
2812 | register char *strp; | |
2813 | { | |
2814 | return ((*(++strp) == 'q' || *strp == 'Q') | |
2815 | && (*(++strp) == 'u' || *strp == 'U') | |
2816 | && (*(++strp) == 'o' || *strp == 'O') | |
2817 | && (*(++strp) == 't' || *strp == 'T') | |
2818 | && (*(++strp) == 'e' || *strp == 'E') | |
2819 | && isspace(*(++strp))); | |
c6d46f5f JB |
2820 | } |
2821 | ||
2822 | void | |
2823 | L_getit () | |
2824 | { | |
2825 | register char *cp; | |
c6d46f5f | 2826 | |
31d4b314 FP |
2827 | if (*dbp == '\'') /* Skip prefix quote */ |
2828 | dbp++; | |
2829 | else if (*dbp == '(' && L_isquote (dbp)) /* Skip "(quote " */ | |
2830 | { | |
2831 | dbp += 7; | |
2832 | while (isspace(*dbp)) | |
2833 | dbp++; | |
2834 | } | |
55597f90 FP |
2835 | for (cp = dbp /*+1*/; |
2836 | *cp && *cp != '(' && *cp != ' ' && *cp != ')'; | |
2837 | cp++) | |
c6d46f5f | 2838 | continue; |
31d4b314 FP |
2839 | if (cp == dbp) |
2840 | return; | |
2841 | ||
55597f90 | 2842 | pfnote (savenstr (dbp, cp-dbp), TRUE, FALSE, lb.buffer, |
591fa824 | 2843 | cp - lb.buffer + 1, lineno, linecharno); |
c6d46f5f | 2844 | } |
6dd5561c FP |
2845 | |
2846 | void | |
2847 | Lisp_functions (inf) | |
2848 | FILE *inf; | |
2849 | { | |
2850 | lineno = 0; | |
2851 | charno = 0; | |
6dd5561c FP |
2852 | |
2853 | while (!feof (inf)) | |
2854 | { | |
2855 | lineno++; | |
2856 | linecharno = charno; | |
2857 | charno += readline (&lb, inf); | |
2858 | dbp = lb.buffer; | |
2859 | if (dbp[0] == '(') | |
2860 | { | |
2861 | if (L_isdef (dbp)) | |
2862 | { | |
2863 | while (!isspace (*dbp)) | |
2864 | dbp++; | |
2865 | while (isspace (*dbp)) | |
2866 | dbp++; | |
2867 | L_getit (); | |
2868 | } | |
2869 | else | |
2870 | { | |
2871 | /* Check for (foo::defmumble name-defined ... */ | |
2872 | do | |
2873 | dbp++; | |
2874 | while (*dbp && !isspace (*dbp) | |
2875 | && *dbp != ':' && *dbp != '(' && *dbp != ')'); | |
2876 | if (*dbp == ':') | |
2877 | { | |
2878 | do | |
2879 | dbp++; | |
2880 | while (*dbp == ':'); | |
2881 | ||
2882 | if (L_isdef (dbp - 1)) | |
2883 | { | |
2884 | while (!isspace (*dbp)) | |
2885 | dbp++; | |
2886 | while (isspace (*dbp)) | |
2887 | dbp++; | |
2888 | L_getit (); | |
2889 | } | |
2890 | } | |
2891 | } | |
2892 | } | |
2893 | } | |
2894 | } | |
c6d46f5f JB |
2895 | \f |
2896 | /* | |
2897 | * Scheme tag functions | |
2898 | * look for (def... xyzzy | |
2899 | * look for (def... (xyzzy | |
2900 | * look for (def ... ((...(xyzzy .... | |
2901 | * look for (set! xyzzy | |
2902 | */ | |
2903 | ||
6dd5561c | 2904 | void get_scheme (); |
c6d46f5f JB |
2905 | |
2906 | void | |
6dd5561c FP |
2907 | Scheme_functions (inf) |
2908 | FILE *inf; | |
c6d46f5f JB |
2909 | { |
2910 | lineno = 0; | |
2911 | charno = 0; | |
c6d46f5f | 2912 | |
6dd5561c | 2913 | while (!feof (inf)) |
c6d46f5f JB |
2914 | { |
2915 | lineno++; | |
2916 | linecharno = charno; | |
6dd5561c | 2917 | charno += readline (&lb, inf); |
c6d46f5f JB |
2918 | dbp = lb.buffer; |
2919 | if (dbp[0] == '(' && | |
2920 | (dbp[1] == 'D' || dbp[1] == 'd') && | |
2921 | (dbp[2] == 'E' || dbp[2] == 'e') && | |
2922 | (dbp[3] == 'F' || dbp[3] == 'f')) | |
2923 | { | |
2924 | while (!isspace (*dbp)) | |
2925 | dbp++; | |
2926 | /* Skip over open parens and white space */ | |
2927 | while (*dbp && (isspace (*dbp) || *dbp == '(')) | |
2928 | dbp++; | |
2929 | get_scheme (); | |
2930 | } | |
2931 | if (dbp[0] == '(' && | |
2932 | (dbp[1] == 'S' || dbp[1] == 's') && | |
2933 | (dbp[2] == 'E' || dbp[2] == 'e') && | |
2934 | (dbp[3] == 'T' || dbp[3] == 't') && | |
2935 | (dbp[4] == '!' || dbp[4] == '!') && | |
2936 | (isspace (dbp[5]))) | |
2937 | { | |
2938 | while (!isspace (*dbp)) | |
2939 | dbp++; | |
2940 | /* Skip over white space */ | |
2941 | while (isspace (*dbp)) | |
2942 | dbp++; | |
2943 | get_scheme (); | |
2944 | } | |
2945 | } | |
2946 | } | |
2947 | ||
6dd5561c | 2948 | void |
c6d46f5f JB |
2949 | get_scheme () |
2950 | { | |
2951 | register char *cp; | |
c6d46f5f JB |
2952 | |
2953 | if (*dbp == 0) | |
2954 | return; | |
2955 | /* Go till you get to white space or a syntactic break */ | |
55597f90 FP |
2956 | for (cp = dbp + 1; |
2957 | *cp && *cp != '(' && *cp != ')' && !isspace (*cp); | |
2958 | cp++) | |
c6d46f5f | 2959 | continue; |
55597f90 FP |
2960 | pfnote (savenstr (dbp, cp-dbp), TRUE, FALSE, |
2961 | lb.buffer, cp - lb.buffer + 1, lineno, linecharno); | |
c6d46f5f JB |
2962 | } |
2963 | \f | |
2964 | /* Find tags in TeX and LaTeX input files. */ | |
2965 | ||
2966 | /* TEX_toktab is a table of TeX control sequences that define tags. | |
2967 | Each TEX_tabent records one such control sequence. | |
2968 | CONVERT THIS TO USE THE Stab TYPE!! */ | |
c6d46f5f JB |
2969 | struct TEX_tabent |
2970 | { | |
2971 | char *name; | |
2972 | int len; | |
2973 | }; | |
2974 | ||
2975 | struct TEX_tabent *TEX_toktab = NULL; /* Table with tag tokens */ | |
2976 | ||
2977 | /* Default set of control sequences to put into TEX_toktab. | |
2978 | The value of environment var TEXTAGS is prepended to this. */ | |
2979 | ||
6dd5561c FP |
2980 | char *TEX_defenv = "\ |
2981 | :chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem:typeout"; | |
c6d46f5f JB |
2982 | |
2983 | void TEX_mode (); | |
2984 | struct TEX_tabent *TEX_decode_env (); | |
2985 | void TEX_getit (); | |
2986 | int TEX_Token (); | |
2987 | ||
6dd5561c FP |
2988 | char TEX_esc = '\\'; |
2989 | char TEX_opgrp = '{'; | |
2990 | char TEX_clgrp = '}'; | |
c6d46f5f JB |
2991 | |
2992 | /* | |
2993 | * TeX/LaTeX scanning loop. | |
2994 | */ | |
c6d46f5f | 2995 | void |
6dd5561c FP |
2996 | TeX_functions (inf) |
2997 | FILE *inf; | |
c6d46f5f JB |
2998 | { |
2999 | char *lasthit; | |
3000 | ||
3001 | lineno = 0; | |
3002 | charno = 0; | |
c6d46f5f JB |
3003 | |
3004 | /* Select either \ or ! as escape character. */ | |
6dd5561c | 3005 | TEX_mode (inf); |
c6d46f5f JB |
3006 | |
3007 | /* Initialize token table once from environment. */ | |
3008 | if (!TEX_toktab) | |
3009 | TEX_toktab = TEX_decode_env ("TEXTAGS", TEX_defenv); | |
3010 | ||
6dd5561c | 3011 | while (!feof (inf)) |
d2729198 | 3012 | { /* Scan each line in file */ |
c6d46f5f JB |
3013 | lineno++; |
3014 | linecharno = charno; | |
6dd5561c | 3015 | charno += readline (&lb, inf); |
c6d46f5f JB |
3016 | dbp = lb.buffer; |
3017 | lasthit = dbp; | |
b02c5fea | 3018 | while (dbp = etags_strchr (dbp, TEX_esc)) /* Look at each esc in line */ |
8a6c8bcf RS |
3019 | { |
3020 | register int i; | |
c6d46f5f | 3021 | |
8a6c8bcf RS |
3022 | if (!*(++dbp)) |
3023 | break; | |
3024 | linecharno += dbp - lasthit; | |
c6d46f5f | 3025 | lasthit = dbp; |
8a6c8bcf RS |
3026 | i = TEX_Token (lasthit); |
3027 | if (0 <= i) | |
c6d46f5f | 3028 | { |
8a6c8bcf | 3029 | TEX_getit (lasthit, TEX_toktab[i].len); |
d2729198 | 3030 | break; /* We only save a line once */ |
c6d46f5f JB |
3031 | } |
3032 | } | |
3033 | } | |
3034 | } | |
3035 | ||
3036 | #define TEX_LESC '\\' | |
3037 | #define TEX_SESC '!' | |
3038 | #define TEX_cmt '%' | |
3039 | ||
aab1fdae FP |
3040 | /* Figure out whether TeX's escapechar is '\\' or '!' and set grouping |
3041 | chars accordingly. */ | |
c6d46f5f | 3042 | void |
6dd5561c FP |
3043 | TEX_mode (inf) |
3044 | FILE *inf; | |
c6d46f5f JB |
3045 | { |
3046 | int c; | |
3047 | ||
6dd5561c | 3048 | while ((c = getc (inf)) != EOF) |
c6d46f5f JB |
3049 | { |
3050 | /* Skip to next line if we hit the TeX comment char. */ | |
3051 | if (c == TEX_cmt) | |
3052 | while (c != '\n') | |
6dd5561c | 3053 | c = getc (inf); |
c6d46f5f JB |
3054 | else if (c == TEX_LESC || c == TEX_SESC ) |
3055 | break; | |
3056 | } | |
3057 | ||
3058 | if (c == TEX_LESC) | |
3059 | { | |
3060 | TEX_esc = TEX_LESC; | |
3061 | TEX_opgrp = '{'; | |
3062 | TEX_clgrp = '}'; | |
3063 | } | |
3064 | else | |
3065 | { | |
3066 | TEX_esc = TEX_SESC; | |
3067 | TEX_opgrp = '<'; | |
3068 | TEX_clgrp = '>'; | |
3069 | } | |
6dd5561c | 3070 | rewind (inf); |
c6d46f5f JB |
3071 | } |
3072 | ||
aab1fdae FP |
3073 | /* Read environment and prepend it to the default string. |
3074 | Build token table. */ | |
c6d46f5f JB |
3075 | struct TEX_tabent * |
3076 | TEX_decode_env (evarname, defenv) | |
3077 | char *evarname; | |
3078 | char *defenv; | |
3079 | { | |
3080 | register char *env, *p; | |
c6d46f5f JB |
3081 | |
3082 | struct TEX_tabent *tab; | |
3083 | int size, i; | |
3084 | ||
3085 | /* Append default string to environment. */ | |
3086 | env = getenv (evarname); | |
3087 | if (!env) | |
3088 | env = defenv; | |
3089 | else | |
3090 | env = concat (env, defenv, ""); | |
3091 | ||
3092 | /* Allocate a token table */ | |
3093 | for (size = 1, p = env; p;) | |
b02c5fea | 3094 | if ((p = etags_strchr (p, ':')) && *(++p)) |
c6d46f5f | 3095 | size++; |
8a6c8bcf RS |
3096 | /* Add 1 to leave room for null terminator. */ |
3097 | tab = xnew (size + 1, struct TEX_tabent); | |
c6d46f5f JB |
3098 | |
3099 | /* Unpack environment string into token table. Be careful about */ | |
3100 | /* zero-length strings (leading ':', "::" and trailing ':') */ | |
3101 | for (i = 0; *env;) | |
3102 | { | |
b02c5fea | 3103 | p = etags_strchr (env, ':'); |
c6d46f5f JB |
3104 | if (!p) /* End of environment string. */ |
3105 | p = env + strlen (env); | |
3106 | if (p - env > 0) | |
3107 | { /* Only non-zero strings. */ | |
3108 | tab[i].name = savenstr (env, p - env); | |
3109 | tab[i].len = strlen (tab[i].name); | |
3110 | i++; | |
3111 | } | |
3112 | if (*p) | |
3113 | env = p + 1; | |
3114 | else | |
3115 | { | |
3116 | tab[i].name = NULL; /* Mark end of table. */ | |
3117 | tab[i].len = 0; | |
3118 | break; | |
3119 | } | |
3120 | } | |
3121 | return tab; | |
3122 | } | |
3123 | ||
3124 | /* Record a tag defined by a TeX command of length LEN and starting at NAME. | |
3125 | The name being defined actually starts at (NAME + LEN + 1). | |
3126 | But we seem to include the TeX command in the tag name. */ | |
c6d46f5f JB |
3127 | void |
3128 | TEX_getit (name, len) | |
3129 | char *name; | |
3130 | int len; | |
3131 | { | |
3132 | char *p = name + len; | |
c6d46f5f JB |
3133 | |
3134 | if (*name == 0) | |
3135 | return; | |
3136 | ||
3137 | /* Let tag name extend to next group close (or end of line) */ | |
3138 | while (*p && *p != TEX_clgrp) | |
3139 | p++; | |
55597f90 FP |
3140 | pfnote (savenstr (name, p-name), TRUE, FALSE, lb.buffer, |
3141 | strlen (lb.buffer), lineno, linecharno); | |
c6d46f5f JB |
3142 | } |
3143 | ||
3144 | /* If the text at CP matches one of the tag-defining TeX command names, | |
b02c5fea | 3145 | return the pointer to the first occurrence of that command in TEX_toktab. |
aab1fdae FP |
3146 | Otherwise return -1. |
3147 | Keep the capital `T' in `Token' for dumb truncating compilers | |
c6d46f5f JB |
3148 | (this distinguishes it from `TEX_toktab' */ |
3149 | int | |
3150 | TEX_Token (cp) | |
3151 | char *cp; | |
3152 | { | |
3153 | int i; | |
3154 | ||
3155 | for (i = 0; TEX_toktab[i].len > 0; i++) | |
1a0d8c80 | 3156 | if (strneq (TEX_toktab[i].name, cp, TEX_toktab[i].len)) |
c6d46f5f JB |
3157 | return i; |
3158 | return -1; | |
3159 | } | |
3160 | \f | |
3161 | /* Support for Prolog. */ | |
3162 | ||
aab1fdae | 3163 | /* Whole head (not only functor, but also arguments) |
c6d46f5f | 3164 | is gotten in compound term. */ |
c6d46f5f | 3165 | void |
b9755a12 | 3166 | prolog_getit (s) |
c6d46f5f | 3167 | char *s; |
c6d46f5f | 3168 | { |
55597f90 | 3169 | char *save_s; |
c6d46f5f JB |
3170 | int insquote, npar; |
3171 | ||
3172 | save_s = s; | |
3173 | insquote = FALSE; | |
3174 | npar = 0; | |
3175 | while (1) | |
3176 | { | |
55597f90 | 3177 | if (s[0] == '\0') /* syntax error. */ |
c6d46f5f | 3178 | return; |
55597f90 | 3179 | else if (insquote && s[0] == '\'' && s[1] == '\'') |
c6d46f5f | 3180 | s += 2; |
55597f90 | 3181 | else if (s[0] == '\'') |
c6d46f5f JB |
3182 | { |
3183 | insquote = !insquote; | |
3184 | s++; | |
3185 | } | |
55597f90 | 3186 | else if (!insquote && s[0] == '(') |
c6d46f5f JB |
3187 | { |
3188 | npar++; | |
3189 | s++; | |
3190 | } | |
55597f90 | 3191 | else if (!insquote && s[0] == ')') |
c6d46f5f JB |
3192 | { |
3193 | npar--; | |
3194 | s++; | |
3195 | if (npar == 0) | |
3196 | break; | |
3197 | else if (npar < 0) /* syntax error. */ | |
3198 | return; | |
3199 | } | |
55597f90 FP |
3200 | else if (!insquote && s[0] == '.' |
3201 | && (isspace (s[1]) || s[1] == '\0')) | |
c6d46f5f JB |
3202 | { /* fullstop. */ |
3203 | if (npar != 0) /* syntax error. */ | |
3204 | return; | |
3205 | s++; | |
3206 | break; | |
3207 | } | |
3208 | else | |
3209 | s++; | |
3210 | } | |
55597f90 FP |
3211 | pfnote (savenstr (save_s, s-save_s), TRUE, FALSE, |
3212 | save_s, s-save_s, lineno, linecharno); | |
c6d46f5f JB |
3213 | } |
3214 | ||
3215 | /* It is assumed that prolog predicate starts from column 0. */ | |
c6d46f5f | 3216 | void |
6dd5561c FP |
3217 | Prolog_functions (inf) |
3218 | FILE *inf; | |
c6d46f5f JB |
3219 | { |
3220 | void skip_comment (), prolog_getit (); | |
3221 | ||
3222 | lineno = linecharno = charno = 0; | |
6dd5561c | 3223 | while (!feof (inf)) |
c6d46f5f JB |
3224 | { |
3225 | lineno++; | |
3226 | linecharno += charno; | |
6dd5561c | 3227 | charno = readline (&lb, inf) + 1; /* 1 for newline. */ |
c6d46f5f JB |
3228 | dbp = lb.buffer; |
3229 | if (isspace (dbp[0])) /* not predicate header. */ | |
3230 | continue; | |
3231 | else if (dbp[0] == '%') /* comment. */ | |
3232 | continue; | |
3233 | else if (dbp[0] == '/' && dbp[1] == '*') /* comment. */ | |
6dd5561c | 3234 | skip_comment (&lb, inf, &lineno, &linecharno); |
c6d46f5f | 3235 | else /* found. */ |
b9755a12 | 3236 | prolog_getit (dbp); |
c6d46f5f JB |
3237 | } |
3238 | } | |
3239 | ||
3240 | void | |
6dd5561c | 3241 | skip_comment (plb, inf, plineno, plinecharno) |
c6d46f5f | 3242 | struct linebuffer *plb; |
6dd5561c | 3243 | FILE *inf; |
c6d46f5f JB |
3244 | int *plineno; /* result */ |
3245 | long *plinecharno; /* result */ | |
3246 | { | |
b9755a12 FP |
3247 | char *cp; |
3248 | ||
3249 | do | |
c6d46f5f | 3250 | { |
b9755a12 FP |
3251 | for (cp = plb->buffer; *cp != '\0'; cp++) |
3252 | if (cp[0] == '*' && cp[1] == '/') | |
3253 | return; | |
c6d46f5f | 3254 | (*plineno)++; |
b9755a12 FP |
3255 | *plinecharno += readline (plb, inf) + 1; /* 1 for newline. */ |
3256 | } | |
3257 | while (!feof(inf)); | |
c6d46f5f | 3258 | } |
b9755a12 FP |
3259 | \f |
3260 | #ifdef ETAGS_REGEXPS | |
3261 | /* Take a string like "/blah/" and turn it into "blah", making sure | |
3262 | that the first and last characters are the same, and handling | |
3263 | quoted separator characters. Actually, stops on the occurence of | |
3264 | an unquoted separator. Also turns "\t" into a Tab character. | |
3265 | Returns pointer to terminating separator. Works in place. Null | |
3266 | terminates name string. */ | |
3267 | char * | |
3268 | scan_separators (name) | |
3269 | char *name; | |
3270 | { | |
3271 | char sep = name[0]; | |
3272 | char *copyto = name; | |
3273 | logical quoted = FALSE; | |
3274 | ||
3275 | for (++name; *name != '\0'; ++name) | |
3276 | { | |
3277 | if (quoted) | |
3278 | { | |
3279 | if (*name == 't') | |
3280 | *copyto++ = '\t'; | |
3281 | else if (*name == sep) | |
3282 | *copyto++ = sep; | |
3283 | else | |
3284 | { | |
3285 | /* Something else is quoted, so preserve the quote. */ | |
3286 | *copyto++ = '\\'; | |
3287 | *copyto++ = *name; | |
3288 | } | |
3289 | quoted = FALSE; | |
3290 | } | |
3291 | else if (*name == '\\') | |
3292 | quoted = TRUE; | |
3293 | else if (*name == sep) | |
3294 | break; | |
3295 | else | |
3296 | *copyto++ = *name; | |
3297 | } | |
c6d46f5f | 3298 | |
b9755a12 FP |
3299 | /* Terminate copied string. */ |
3300 | *copyto = '\0'; | |
3301 | return name; | |
3302 | } | |
c6d46f5f | 3303 | |
b9755a12 FP |
3304 | /* Turn a name, which is an ed-style (but Emacs syntax) regular |
3305 | expression, into a real regular expression by compiling it. */ | |
3306 | void | |
3307 | add_regex (regexp_pattern) | |
3308 | char *regexp_pattern; | |
c6d46f5f | 3309 | { |
b9755a12 FP |
3310 | char *name; |
3311 | const char *err; | |
3312 | struct re_pattern_buffer *patbuf; | |
c6d46f5f | 3313 | |
b9755a12 FP |
3314 | if (regexp_pattern == NULL) |
3315 | { | |
3316 | /* Remove existing regexps. */ | |
3317 | num_patterns = 0; | |
3318 | patterns = NULL; | |
3319 | return; | |
3320 | } | |
c6d46f5f | 3321 | |
b9755a12 FP |
3322 | if (regexp_pattern[0] == '\0') |
3323 | { | |
3324 | error ("missing regexp", 0); | |
3325 | return; | |
3326 | } | |
3327 | if (regexp_pattern[strlen(regexp_pattern)-1] != regexp_pattern[0]) | |
3328 | { | |
3329 | error ("%s: unterminated regexp", regexp_pattern); | |
3330 | return; | |
3331 | } | |
3332 | name = scan_separators (regexp_pattern); | |
3333 | if (regexp_pattern[0] == '\0') | |
3334 | { | |
3335 | error ("null regexp", 0); | |
3336 | return; | |
3337 | } | |
3338 | (void) scan_separators (name); | |
3339 | ||
3340 | patbuf = xnew (1, struct re_pattern_buffer); | |
3341 | patbuf->translate = NULL; | |
3342 | patbuf->fastmap = NULL; | |
3343 | patbuf->buffer = NULL; | |
3344 | patbuf->allocated = 0; | |
3345 | ||
3346 | err = re_compile_pattern (regexp_pattern, strlen (regexp_pattern), patbuf); | |
3347 | if (err != NULL) | |
3348 | { | |
3349 | error ("%s while compiling pattern", err); | |
3350 | return; | |
3351 | } | |
3352 | ||
3353 | num_patterns += 1; | |
3354 | if (num_patterns == 1) | |
3355 | patterns = xnew (1, struct pattern); | |
c6d46f5f | 3356 | else |
b9755a12 FP |
3357 | patterns = ((struct pattern *) |
3358 | xrealloc (patterns, | |
3359 | (num_patterns * sizeof (struct pattern)))); | |
3360 | patterns[num_patterns - 1].pattern = patbuf; | |
3361 | patterns[num_patterns - 1].name_pattern = savestr (name); | |
3362 | patterns[num_patterns - 1].error_signaled = FALSE; | |
3363 | } | |
3364 | ||
3365 | /* | |
3366 | * Do the subtitutions indicated by the regular expression and | |
3367 | * arguments. | |
3368 | */ | |
3369 | char * | |
3370 | substitute (in, out, regs) | |
3371 | char *in, *out; | |
3372 | struct re_registers *regs; | |
3373 | { | |
3374 | char *result = NULL, *t; | |
3375 | int size = 0; | |
3376 | ||
3377 | /* Pass 1: figure out how much size to allocate. */ | |
3378 | for (t = out; *t; ++t) | |
3379 | { | |
3380 | if (*t == '\\') | |
3381 | { | |
3382 | ++t; | |
3383 | if (!*t) | |
3384 | { | |
3385 | fprintf (stderr, "%s: pattern subtitution ends prematurely\n", | |
3386 | progname); | |
3387 | return NULL; | |
3388 | } | |
3389 | if (isdigit (*t)) | |
3390 | { | |
3391 | int dig = *t - '0'; | |
3392 | size += regs->end[dig] - regs->start[dig]; | |
3393 | } | |
3394 | } | |
3395 | } | |
3396 | ||
3397 | /* Allocate space and do the substitutions. */ | |
3398 | result = xnew (size + 1, char); | |
3399 | size = 0; | |
3400 | for (; *out; ++out) | |
3401 | { | |
3402 | if (*out == '\\') | |
3403 | { | |
3404 | ++out; | |
3405 | if (isdigit (*out)) | |
3406 | { | |
3407 | /* Using "dig2" satisfies my debugger. Bleah. */ | |
3408 | int dig2 = *out - '0'; | |
3409 | strncpy (result + size, in + regs->start[dig2], | |
3410 | regs->end[dig2] - regs->start[dig2]); | |
3411 | size += regs->end[dig2] - regs->start[dig2]; | |
3412 | } | |
3413 | else | |
3414 | { | |
3415 | switch (*out) | |
3416 | { | |
3417 | case '\t': | |
3418 | result[size++] = '\t'; | |
3419 | break; | |
3420 | case '\\': | |
3421 | *out = '\\'; | |
3422 | break; | |
3423 | default: | |
3424 | result[size++] = *out; | |
3425 | break; | |
3426 | } | |
3427 | } | |
3428 | } | |
3429 | else | |
3430 | result[size++] = *out; | |
3431 | } | |
3432 | result[size] = '\0'; | |
3433 | ||
3434 | return result; | |
c6d46f5f JB |
3435 | } |
3436 | \f | |
b9755a12 | 3437 | #endif /* ETAGS_REGEXPS */ |
c6d46f5f | 3438 | /* Initialize a linebuffer for use */ |
c6d46f5f JB |
3439 | void |
3440 | initbuffer (linebuffer) | |
3441 | struct linebuffer *linebuffer; | |
3442 | { | |
3443 | linebuffer->size = 200; | |
3444 | linebuffer->buffer = xnew (200, char); | |
3445 | } | |
3446 | ||
3447 | /* | |
3448 | * Read a line of text from `stream' into `linebuffer'. | |
3449 | * Return the number of characters read from `stream', | |
3450 | * which is the length of the line including the newline, if any. | |
3451 | */ | |
3452 | long | |
b9755a12 | 3453 | readline_internal (linebuffer, stream) |
c6d46f5f JB |
3454 | struct linebuffer *linebuffer; |
3455 | register FILE *stream; | |
3456 | { | |
3457 | char *buffer = linebuffer->buffer; | |
3458 | register char *p = linebuffer->buffer; | |
3459 | register char *pend; | |
aab1fdae | 3460 | int chars_deleted; |
c6d46f5f | 3461 | |
eb8c3be9 | 3462 | pend = p + linebuffer->size; /* Separate to avoid 386/IX compiler bug. */ |
c6d46f5f JB |
3463 | |
3464 | while (1) | |
3465 | { | |
3466 | register int c = getc (stream); | |
3467 | if (p == pend) | |
3468 | { | |
3469 | linebuffer->size *= 2; | |
3470 | buffer = (char *) xrealloc (buffer, linebuffer->size); | |
3471 | p += buffer - linebuffer->buffer; | |
3472 | pend = buffer + linebuffer->size; | |
3473 | linebuffer->buffer = buffer; | |
3474 | } | |
aab1fdae | 3475 | if (c == EOF) |
c6d46f5f | 3476 | { |
aab1fdae FP |
3477 | chars_deleted = 0; |
3478 | break; | |
3479 | } | |
3480 | if (c == '\n') | |
3481 | { | |
a8d9bd4b | 3482 | if (p > buffer && p[-1] == '\r') |
aab1fdae FP |
3483 | { |
3484 | *--p = '\0'; | |
3485 | chars_deleted = 2; | |
3486 | } | |
3487 | else | |
3488 | { | |
3489 | *p = '\0'; | |
3490 | chars_deleted = 1; | |
3491 | } | |
c6d46f5f JB |
3492 | break; |
3493 | } | |
3494 | *p++ = c; | |
3495 | } | |
3496 | ||
aab1fdae | 3497 | return p - buffer + chars_deleted; |
c6d46f5f | 3498 | } |
b9755a12 FP |
3499 | |
3500 | /* | |
3501 | * Like readline_internal, above, but try to match the input | |
3502 | * line against any existing regular expressions. | |
3503 | */ | |
3504 | long | |
3505 | readline (linebuffer, stream) | |
3506 | struct linebuffer *linebuffer; | |
3507 | FILE *stream; | |
3508 | { | |
3509 | /* Read new line. */ | |
3510 | int i; | |
3511 | long result = readline_internal (linebuffer, stream); | |
3512 | ||
3513 | #ifdef ETAGS_REGEXPS | |
3514 | /* Match against all listed patterns. */ | |
3515 | for (i = 0; i < num_patterns; ++i) | |
3516 | { | |
3517 | int match = re_match (patterns[i].pattern, linebuffer->buffer, | |
3518 | (int)result, 0, &patterns[i].regs); | |
3519 | switch (match) | |
3520 | { | |
3521 | case -2: | |
3522 | /* Some error. */ | |
3523 | if (!patterns[i].error_signaled) | |
3524 | { | |
3525 | error ("error while matching pattern %d", i); | |
3526 | patterns[i].error_signaled = TRUE; | |
3527 | } | |
3528 | break; | |
3529 | case -1: | |
3530 | /* No match. */ | |
3531 | break; | |
3532 | default: | |
3533 | /* Match occurred. Construct a tag. */ | |
3534 | if (patterns[i].name_pattern[0] != '\0') | |
3535 | { | |
3536 | /* Make a named tag. */ | |
3537 | char *name = substitute (linebuffer->buffer, | |
3538 | patterns[i].name_pattern, | |
3539 | &patterns[i].regs); | |
3540 | if (name != NULL) | |
3541 | pfnote (name, TRUE, TRUE, linebuffer->buffer, | |
3542 | match, lineno, linecharno); | |
3543 | } | |
3544 | else | |
3545 | { | |
3546 | /* Make an unnamed tag. */ | |
3547 | pfnote (NULL, TRUE, FALSE, linebuffer->buffer, | |
3548 | match, lineno, linecharno); | |
3549 | } | |
3550 | break; | |
3551 | } | |
3552 | } | |
3553 | #endif /* ETAGS_REGEXPS */ | |
3554 | ||
3555 | return result; | |
3556 | } | |
3557 | ||
3558 | /* | |
3559 | * Read a file, but do no processing. This is used to do regexp | |
3560 | * matching on files that have no language defined. | |
3561 | */ | |
3562 | void | |
3563 | just_read_file (inf) | |
3564 | FILE *inf; | |
3565 | { | |
3566 | while (!feof (inf)) | |
3567 | { | |
3568 | ++lineno; | |
3569 | linecharno = charno; | |
3570 | charno += readline (&lb, inf) + 1; | |
3571 | } | |
3572 | } | |
3573 | ||
c6d46f5f | 3574 | \f |
55597f90 FP |
3575 | /* |
3576 | * Return a pointer to a space of size strlen(cp)+1 allocated | |
3577 | * with xnew where the string CP has been copied. | |
3578 | */ | |
c6d46f5f JB |
3579 | char * |
3580 | savestr (cp) | |
3581 | char *cp; | |
3582 | { | |
3583 | return savenstr (cp, strlen (cp)); | |
3584 | } | |
3585 | ||
55597f90 FP |
3586 | /* |
3587 | * Return a pointer to a space of size LEN+1 allocated with xnew where | |
3588 | * the string CP has been copied for at most the first LEN characters. | |
3589 | */ | |
c6d46f5f JB |
3590 | char * |
3591 | savenstr (cp, len) | |
3592 | char *cp; | |
3593 | int len; | |
3594 | { | |
3595 | register char *dp; | |
3596 | ||
3597 | dp = xnew (len + 1, char); | |
1a0d8c80 | 3598 | strncpy (dp, cp, len); |
c6d46f5f JB |
3599 | dp[len] = '\0'; |
3600 | return dp; | |
3601 | } | |
3602 | ||
c6d46f5f JB |
3603 | /* |
3604 | * Return the ptr in sp at which the character c last | |
3605 | * appears; NULL if not found | |
3606 | * | |
b02c5fea | 3607 | * Identical to System V strrchr, included for portability. |
c6d46f5f | 3608 | */ |
c6d46f5f | 3609 | char * |
b02c5fea | 3610 | etags_strrchr (sp, c) |
c6d46f5f JB |
3611 | register char *sp, c; |
3612 | { | |
3613 | register char *r; | |
3614 | ||
3615 | r = NULL; | |
3616 | do | |
3617 | { | |
3618 | if (*sp == c) | |
3619 | r = sp; | |
3620 | } while (*sp++); | |
b9755a12 | 3621 | return r; |
c6d46f5f JB |
3622 | } |
3623 | ||
9d7ad1b3 | 3624 | |
c6d46f5f JB |
3625 | /* |
3626 | * Return the ptr in sp at which the character c first | |
3627 | * appears; NULL if not found | |
3628 | * | |
b02c5fea | 3629 | * Identical to System V strchr, included for portability. |
c6d46f5f | 3630 | */ |
c6d46f5f | 3631 | char * |
b02c5fea | 3632 | etags_strchr (sp, c) |
c6d46f5f JB |
3633 | register char *sp, c; |
3634 | { | |
3635 | do | |
3636 | { | |
3637 | if (*sp == c) | |
b9755a12 FP |
3638 | return sp; |
3639 | } while (*sp++); | |
3640 | return NULL; | |
c6d46f5f JB |
3641 | } |
3642 | ||
c6d46f5f | 3643 | /* Print error message and exit. */ |
c6d46f5f JB |
3644 | void |
3645 | fatal (s1, s2) | |
3646 | char *s1, *s2; | |
3647 | { | |
3648 | error (s1, s2); | |
1a0d8c80 | 3649 | exit (BAD); |
c6d46f5f JB |
3650 | } |
3651 | ||
cdc1f6a7 FP |
3652 | void |
3653 | pfatal (s1) | |
3654 | char *s1; | |
3655 | { | |
3656 | perror (s1); | |
3657 | exit (BAD); | |
3658 | } | |
3659 | ||
c6d46f5f | 3660 | /* Print error message. `s1' is printf control string, `s2' is arg for it. */ |
c6d46f5f JB |
3661 | void |
3662 | error (s1, s2) | |
3663 | char *s1, *s2; | |
3664 | { | |
3665 | fprintf (stderr, "%s: ", progname); | |
3666 | fprintf (stderr, s1, s2); | |
3667 | fprintf (stderr, "\n"); | |
3668 | } | |
3669 | ||
46c145db FP |
3670 | /* Return a newly-allocated string whose contents |
3671 | concatenate those of s1, s2, s3. */ | |
c6d46f5f JB |
3672 | char * |
3673 | concat (s1, s2, s3) | |
3674 | char *s1, *s2, *s3; | |
3675 | { | |
3676 | int len1 = strlen (s1), len2 = strlen (s2), len3 = strlen (s3); | |
3677 | char *result = xnew (len1 + len2 + len3 + 1, char); | |
3678 | ||
1a0d8c80 FP |
3679 | strcpy (result, s1); |
3680 | strcpy (result + len1, s2); | |
3681 | strcpy (result + len1 + len2, s3); | |
46c145db | 3682 | result[len1 + len2 + len3] = '\0'; |
c6d46f5f JB |
3683 | |
3684 | return result; | |
3685 | } | |
b02c5fea | 3686 | \f |
cdc1f6a7 FP |
3687 | /* Does the same work as the system V getcwd, but does not need to |
3688 | guess buffer size in advance. */ | |
88f125fc RS |
3689 | char * |
3690 | etags_getcwd () | |
915e30c8 | 3691 | { |
0ff74d81 | 3692 | #ifdef DOS_NT |
cdc1f6a7 FP |
3693 | char *p, path[MAXPATHLEN + 1]; /* Fixed size is safe on MSDOS. */ |
3694 | ||
3695 | getwd (path); | |
3696 | p = path; | |
88f125fc RS |
3697 | while (*p) |
3698 | if (*p == '\\') | |
3699 | *p++ = '/'; | |
3700 | else | |
3701 | *p++ = tolower (*p); | |
cdc1f6a7 FP |
3702 | |
3703 | return strdup (path); | |
0ff74d81 FP |
3704 | #else /* not DOS_NT */ |
3705 | #if HAVE_GETCWD | |
cdc1f6a7 FP |
3706 | int bufsize = 200; |
3707 | char *path = xnew (bufsize, char); | |
c6d46f5f | 3708 | |
cdc1f6a7 | 3709 | while (getcwd (path, bufsize) == NULL) |
b02c5fea | 3710 | { |
dcc89e63 | 3711 | if (errno != ERANGE) |
0f394065 | 3712 | pfatal ("getcwd"); |
5e9c8296 | 3713 | bufsize *= 2; |
cdc1f6a7 | 3714 | path = xnew (bufsize, char); |
5e9c8296 | 3715 | } |
b02c5fea | 3716 | |
cdc1f6a7 | 3717 | return path; |
cdc1f6a7 | 3718 | #else /* not DOS_NT and not HAVE_GETCWD */ |
cdc1f6a7 FP |
3719 | struct linebuffer path; |
3720 | FILE *pipe; | |
b02c5fea | 3721 | |
cdc1f6a7 | 3722 | initbuffer (&path); |
915e30c8 | 3723 | pipe = (FILE *) popen ("pwd 2>/dev/null", "r"); |
cdc1f6a7 | 3724 | if (pipe == NULL || readline_internal (&path, pipe) == 0) |
915e30c8 | 3725 | pfatal ("pwd"); |
cdc1f6a7 | 3726 | pclose (pipe); |
b02c5fea | 3727 | |
cdc1f6a7 | 3728 | return path.buffer; |
0ff74d81 FP |
3729 | #endif /* not HAVE_GETCWD */ |
3730 | #endif /* not DOS_NT */ | |
b02c5fea FP |
3731 | } |
3732 | ||
3733 | /* Return a newly allocated string containing the filename | |
3734 | of FILE relative to the absolute directory DIR (which | |
3735 | should end with a slash). */ | |
46c145db FP |
3736 | char * |
3737 | relative_filename (file, dir) | |
3738 | char *file, *dir; | |
3739 | { | |
3740 | char *fp, *dp, *res; | |
3741 | ||
3742 | /* Find the common root of file and dir. */ | |
3743 | fp = absolute_filename (file, cwd); | |
3744 | dp = dir; | |
3745 | while (*fp++ == *dp++) | |
3746 | continue; | |
3747 | do | |
3748 | { | |
3749 | fp--; | |
3750 | dp--; | |
3751 | } | |
3752 | while (*fp != '/'); | |
3753 | ||
3754 | /* Build a sequence of "../" strings for the resulting relative filename. */ | |
b02c5fea | 3755 | for (dp = etags_strchr (dp + 1, '/'), res = ""; |
46c145db | 3756 | dp != NULL; |
b02c5fea | 3757 | dp = etags_strchr (dp + 1, '/')) |
46c145db FP |
3758 | { |
3759 | res = concat (res, "../", ""); | |
3760 | } | |
3761 | ||
3762 | /* Add the filename relative to the common root of file and dir. */ | |
3763 | res = concat (res, fp + 1, ""); | |
3764 | ||
3765 | return res; /* temporary stub */ | |
3766 | } | |
3767 | ||
3768 | /* Return a newly allocated string containing the | |
b02c5fea FP |
3769 | absolute filename of FILE given CWD (which should |
3770 | end with a slash). */ | |
46c145db FP |
3771 | char * |
3772 | absolute_filename (file, cwd) | |
3773 | char *file, *cwd; | |
3774 | { | |
3775 | char *slashp, *cp, *res; | |
3776 | ||
b2db879b | 3777 | if (absolutefn (file)) |
46c145db FP |
3778 | res = concat (file, "", ""); |
3779 | else | |
3780 | res = concat (cwd, file, ""); | |
3781 | ||
3782 | /* Delete the "/dirname/.." and "/." substrings. */ | |
b02c5fea | 3783 | slashp = etags_strchr (res, '/'); |
46c145db FP |
3784 | while (slashp != NULL && slashp[0] != '\0') |
3785 | { | |
3786 | if (slashp[1] == '.') | |
3787 | { | |
3788 | if (slashp[2] == '.' | |
3789 | && (slashp[3] == '/' || slashp[3] == '\0')) | |
3790 | { | |
3791 | cp = slashp; | |
3792 | do | |
3793 | cp--; | |
3794 | while (cp >= res && *cp != '/'); | |
3795 | if (*cp == '/') | |
3796 | { | |
3797 | strcpy (cp, slashp + 3); | |
3798 | } | |
3799 | else /* else (cp == res) */ | |
3800 | { | |
1875d994 | 3801 | if (slashp[3] != '\0') |
46c145db FP |
3802 | strcpy (cp, slashp + 4); |
3803 | else | |
3804 | return "."; | |
3805 | } | |
3806 | slashp = cp; | |
e9b2b94c | 3807 | continue; |
46c145db FP |
3808 | } |
3809 | else if (slashp[2] == '/' || slashp[2] == '\0') | |
3810 | { | |
3811 | strcpy (slashp, slashp + 2); | |
e9b2b94c | 3812 | continue; |
46c145db FP |
3813 | } |
3814 | } | |
e9b2b94c FP |
3815 | |
3816 | slashp = etags_strchr (slashp + 1, '/'); | |
46c145db FP |
3817 | } |
3818 | ||
3819 | return res; | |
3820 | } | |
3821 | ||
b02c5fea FP |
3822 | /* Return a newly allocated string containing the absolute |
3823 | filename of dir where FILE resides given CWD (which should | |
3824 | end with a slash). */ | |
46c145db FP |
3825 | char * |
3826 | absolute_dirname (file, cwd) | |
3827 | char *file, *cwd; | |
3828 | { | |
3829 | char *slashp, *res; | |
3830 | char save; | |
3831 | ||
b02c5fea | 3832 | slashp = etags_strrchr (file, '/'); |
46c145db FP |
3833 | if (slashp == NULL) |
3834 | return cwd; | |
3835 | save = slashp[1]; | |
3836 | slashp[1] = '\0'; | |
3837 | res = absolute_filename (file, cwd); | |
3838 | slashp[1] = save; | |
3839 | ||
3840 | return res; | |
3841 | } | |
3842 | ||
c6d46f5f | 3843 | /* Like malloc but get fatal error if memory is exhausted. */ |
03cdafdf | 3844 | long * |
c6d46f5f | 3845 | xmalloc (size) |
42680d3c | 3846 | unsigned int size; |
c6d46f5f | 3847 | { |
03cdafdf | 3848 | long *result = (long *) malloc (size); |
1a0d8c80 | 3849 | if (result == NULL) |
c6d46f5f JB |
3850 | fatal ("virtual memory exhausted", 0); |
3851 | return result; | |
3852 | } | |
3853 | ||
03cdafdf | 3854 | long * |
c6d46f5f JB |
3855 | xrealloc (ptr, size) |
3856 | char *ptr; | |
42680d3c | 3857 | unsigned int size; |
c6d46f5f | 3858 | { |
03cdafdf | 3859 | long *result = (long *) realloc (ptr, size); |
1a0d8c80 | 3860 | if (result == NULL) |
c6d46f5f JB |
3861 | fatal ("virtual memory exhausted"); |
3862 | return result; | |
3863 | } |