Test and document --keep-directory-symlink
[tar.git] / src / transform.c
blobe450dd22ee91f2d1e99c8b4c2d9e00b401cd7af7
1 /* This file is part of GNU tar.
2 Copyright 2006-2008, 2013-2014, 2016-2017 Free Software Foundation,
3 Inc.
5 This program is free software; you can redistribute it and/or modify it
6 under the terms of the GNU General Public License as published by the
7 Free Software Foundation; either version 3, or (at your option) any later
8 version.
10 This program is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
13 Public License for more details.
15 You should have received a copy of the GNU General Public License along
16 with this program. If not, see <http://www.gnu.org/licenses/>. */
18 #include <system.h>
19 #include <regex.h>
20 #include "common.h"
22 enum transform_type
24 transform_first,
25 transform_global
28 enum replace_segm_type
30 segm_literal, /* Literal segment */
31 segm_backref, /* Back-reference segment */
32 segm_case_ctl /* Case control segment (GNU extension) */
35 enum case_ctl_type
37 ctl_stop, /* Stop case conversion */
38 ctl_upcase_next,/* Turn the next character to uppercase */
39 ctl_locase_next,/* Turn the next character to lowercase */
40 ctl_upcase, /* Turn the replacement to uppercase until ctl_stop */
41 ctl_locase /* Turn the replacement to lowercase until ctl_stop */
44 struct replace_segm
46 struct replace_segm *next;
47 enum replace_segm_type type;
48 union
50 struct
52 char *ptr;
53 size_t size;
54 } literal; /* type == segm_literal */
55 size_t ref; /* type == segm_backref */
56 enum case_ctl_type ctl; /* type == segm_case_ctl */
57 } v;
60 struct transform
62 struct transform *next;
63 enum transform_type transform_type;
64 int flags;
65 unsigned match_number;
66 regex_t regex;
67 /* Compiled replacement expression */
68 struct replace_segm *repl_head, *repl_tail;
69 size_t segm_count; /* Number of elements in the above list */
74 static int transform_flags = XFORM_ALL;
75 static struct transform *transform_head, *transform_tail;
77 static struct transform *
78 new_transform (void)
80 struct transform *p = xzalloc (sizeof *p);
81 if (transform_tail)
82 transform_tail->next = p;
83 else
84 transform_head = p;
85 transform_tail = p;
86 return p;
89 static struct replace_segm *
90 add_segment (struct transform *tf)
92 struct replace_segm *segm = xmalloc (sizeof *segm);
93 segm->next = NULL;
94 if (tf->repl_tail)
95 tf->repl_tail->next = segm;
96 else
97 tf->repl_head = segm;
98 tf->repl_tail = segm;
99 tf->segm_count++;
100 return segm;
103 static void
104 add_literal_segment (struct transform *tf, char *str, char *end)
106 size_t len = end - str;
107 if (len)
109 struct replace_segm *segm = add_segment (tf);
110 segm->type = segm_literal;
111 segm->v.literal.ptr = xmalloc (len + 1);
112 memcpy (segm->v.literal.ptr, str, len);
113 segm->v.literal.ptr[len] = 0;
114 segm->v.literal.size = len;
118 static void
119 add_char_segment (struct transform *tf, int chr)
121 struct replace_segm *segm = add_segment (tf);
122 segm->type = segm_literal;
123 segm->v.literal.ptr = xmalloc (2);
124 segm->v.literal.ptr[0] = chr;
125 segm->v.literal.ptr[1] = 0;
126 segm->v.literal.size = 1;
129 static void
130 add_backref_segment (struct transform *tf, size_t ref)
132 struct replace_segm *segm = add_segment (tf);
133 segm->type = segm_backref;
134 segm->v.ref = ref;
137 static int
138 parse_xform_flags (int *pflags, int c)
140 switch (c)
142 case 'r':
143 *pflags |= XFORM_REGFILE;
144 break;
146 case 'R':
147 *pflags &= ~XFORM_REGFILE;
148 break;
150 case 'h':
151 *pflags |= XFORM_LINK;
152 break;
154 case 'H':
155 *pflags &= ~XFORM_LINK;
156 break;
158 case 's':
159 *pflags |= XFORM_SYMLINK;
160 break;
162 case 'S':
163 *pflags &= ~XFORM_SYMLINK;
164 break;
166 default:
167 return 1;
169 return 0;
172 static void
173 add_case_ctl_segment (struct transform *tf, enum case_ctl_type ctl)
175 struct replace_segm *segm = add_segment (tf);
176 segm->type = segm_case_ctl;
177 segm->v.ctl = ctl;
180 static const char *
181 parse_transform_expr (const char *expr)
183 int delim;
184 int i, j, rc;
185 char *str, *beg, *cur;
186 const char *p;
187 int cflags = 0;
188 struct transform *tf = new_transform ();
190 if (expr[0] != 's')
192 if (strncmp (expr, "flags=", 6) == 0)
194 transform_flags = 0;
195 for (expr += 6; *expr; expr++)
197 if (*expr == ';')
199 expr++;
200 break;
202 if (parse_xform_flags (&transform_flags, *expr))
203 USAGE_ERROR ((0, 0, _("Unknown transform flag: %c"),
204 *expr));
206 return expr;
208 USAGE_ERROR ((0, 0, _("Invalid transform expression")));
211 delim = expr[1];
213 /* Scan regular expression */
214 for (i = 2; expr[i] && expr[i] != delim; i++)
215 if (expr[i] == '\\' && expr[i+1])
216 i++;
218 if (expr[i] != delim)
219 USAGE_ERROR ((0, 0, _("Invalid transform expression")));
221 /* Scan replacement expression */
222 for (j = i + 1; expr[j] && expr[j] != delim; j++)
223 if (expr[j] == '\\' && expr[j+1])
224 j++;
226 if (expr[j] != delim)
227 USAGE_ERROR ((0, 0, _("Invalid transform expression")));
229 /* Check flags */
230 tf->transform_type = transform_first;
231 tf->flags = transform_flags;
232 for (p = expr + j + 1; *p && *p != ';'; p++)
233 switch (*p)
235 case 'g':
236 tf->transform_type = transform_global;
237 break;
239 case 'i':
240 cflags |= REG_ICASE;
241 break;
243 case 'x':
244 cflags |= REG_EXTENDED;
245 break;
247 case '0': case '1': case '2': case '3': case '4':
248 case '5': case '6': case '7': case '8': case '9':
249 tf->match_number = strtoul (p, (char**) &p, 0);
250 p--;
251 break;
253 default:
254 if (parse_xform_flags (&tf->flags, *p))
255 USAGE_ERROR ((0, 0, _("Unknown flag in transform expression: %c"),
256 *p));
259 if (*p == ';')
260 p++;
262 /* Extract and compile regex */
263 str = xmalloc (i - 1);
264 memcpy (str, expr + 2, i - 2);
265 str[i - 2] = 0;
267 rc = regcomp (&tf->regex, str, cflags);
269 if (rc)
271 char errbuf[512];
272 regerror (rc, &tf->regex, errbuf, sizeof (errbuf));
273 USAGE_ERROR ((0, 0, _("Invalid transform expression: %s"), errbuf));
276 if (str[0] == '^' || str[strlen (str) - 1] == '$')
277 tf->transform_type = transform_first;
279 free (str);
281 /* Extract and compile replacement expr */
282 i++;
283 str = xmalloc (j - i + 1);
284 memcpy (str, expr + i, j - i);
285 str[j - i] = 0;
287 for (cur = beg = str; *cur;)
289 if (*cur == '\\')
291 size_t n;
293 add_literal_segment (tf, beg, cur);
294 switch (*++cur)
296 case '0': case '1': case '2': case '3': case '4':
297 case '5': case '6': case '7': case '8': case '9':
298 n = strtoul (cur, &cur, 10);
299 if (n > tf->regex.re_nsub)
300 USAGE_ERROR ((0, 0, _("Invalid transform replacement: back reference out of range")));
301 add_backref_segment (tf, n);
302 break;
304 case '\\':
305 add_char_segment (tf, '\\');
306 cur++;
307 break;
309 case 'a':
310 add_char_segment (tf, '\a');
311 cur++;
312 break;
314 case 'b':
315 add_char_segment (tf, '\b');
316 cur++;
317 break;
319 case 'f':
320 add_char_segment (tf, '\f');
321 cur++;
322 break;
324 case 'n':
325 add_char_segment (tf, '\n');
326 cur++;
327 break;
329 case 'r':
330 add_char_segment (tf, '\r');
331 cur++;
332 break;
334 case 't':
335 add_char_segment (tf, '\t');
336 cur++;
337 break;
339 case 'v':
340 add_char_segment (tf, '\v');
341 cur++;
342 break;
344 case '&':
345 add_char_segment (tf, '&');
346 cur++;
347 break;
349 case 'L':
350 /* Turn the replacement to lowercase until a '\U' or '\E'
351 is found, */
352 add_case_ctl_segment (tf, ctl_locase);
353 cur++;
354 break;
356 case 'l':
357 /* Turn the next character to lowercase, */
358 add_case_ctl_segment (tf, ctl_locase_next);
359 cur++;
360 break;
362 case 'U':
363 /* Turn the replacement to uppercase until a '\L' or '\E'
364 is found, */
365 add_case_ctl_segment (tf, ctl_upcase);
366 cur++;
367 break;
369 case 'u':
370 /* Turn the next character to uppercase, */
371 add_case_ctl_segment (tf, ctl_upcase_next);
372 cur++;
373 break;
375 case 'E':
376 /* Stop case conversion started by '\L' or '\U'. */
377 add_case_ctl_segment (tf, ctl_stop);
378 cur++;
379 break;
381 default:
382 if (*cur == delim)
383 add_char_segment (tf, delim);
384 else
386 char buf[2];
387 buf[0] = '\\';
388 buf[1] = *cur;
389 add_literal_segment (tf, buf, buf + 2);
391 cur++;
392 break;
394 beg = cur;
396 else if (*cur == '&')
398 add_literal_segment (tf, beg, cur);
399 add_backref_segment (tf, 0);
400 beg = ++cur;
402 else
403 cur++;
405 add_literal_segment (tf, beg, cur);
407 return p;
410 void
411 set_transform_expr (const char *expr)
413 while (*expr)
414 expr = parse_transform_expr (expr);
417 /* Run case conversion specified by CASE_CTL on array PTR of SIZE
418 characters. Returns pointer to statically allocated storage. */
419 static char *
420 run_case_conv (enum case_ctl_type case_ctl, char *ptr, size_t size)
422 static char *case_ctl_buffer;
423 static size_t case_ctl_bufsize;
424 char *p;
426 if (case_ctl_bufsize < size)
428 case_ctl_bufsize = size;
429 case_ctl_buffer = xrealloc (case_ctl_buffer, case_ctl_bufsize);
431 memcpy (case_ctl_buffer, ptr, size);
432 switch (case_ctl)
434 case ctl_upcase_next:
435 case_ctl_buffer[0] = toupper ((unsigned char) case_ctl_buffer[0]);
436 break;
438 case ctl_locase_next:
439 case_ctl_buffer[0] = tolower ((unsigned char) case_ctl_buffer[0]);
440 break;
442 case ctl_upcase:
443 for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
444 *p = toupper ((unsigned char) *p);
445 break;
447 case ctl_locase:
448 for (p = case_ctl_buffer; p < case_ctl_buffer + size; p++)
449 *p = tolower ((unsigned char) *p);
450 break;
452 case ctl_stop:
453 break;
455 return case_ctl_buffer;
459 static struct obstack stk;
460 static bool stk_init;
462 static void
463 _single_transform_name_to_obstack (struct transform *tf, char *input)
465 regmatch_t *rmp;
466 int rc;
467 size_t nmatches = 0;
468 enum case_ctl_type case_ctl = ctl_stop, /* Current case conversion op */
469 save_ctl = ctl_stop; /* Saved case_ctl for \u and \l */
471 /* Reset case conversion after a single-char operation */
472 #define CASE_CTL_RESET() if (case_ctl == ctl_upcase_next \
473 || case_ctl == ctl_locase_next) \
475 case_ctl = save_ctl; \
476 save_ctl = ctl_stop; \
479 rmp = xmalloc ((tf->regex.re_nsub + 1) * sizeof (*rmp));
481 while (*input)
483 size_t disp;
484 char *ptr;
486 rc = regexec (&tf->regex, input, tf->regex.re_nsub + 1, rmp, 0);
488 if (rc == 0)
490 struct replace_segm *segm;
492 disp = rmp[0].rm_eo;
494 if (rmp[0].rm_so)
495 obstack_grow (&stk, input, rmp[0].rm_so);
497 nmatches++;
498 if (tf->match_number && nmatches < tf->match_number)
500 obstack_grow (&stk, input, disp);
501 input += disp;
502 continue;
505 for (segm = tf->repl_head; segm; segm = segm->next)
507 switch (segm->type)
509 case segm_literal: /* Literal segment */
510 if (case_ctl == ctl_stop)
511 ptr = segm->v.literal.ptr;
512 else
514 ptr = run_case_conv (case_ctl,
515 segm->v.literal.ptr,
516 segm->v.literal.size);
517 CASE_CTL_RESET();
519 obstack_grow (&stk, ptr, segm->v.literal.size);
520 break;
522 case segm_backref: /* Back-reference segment */
523 if (rmp[segm->v.ref].rm_so != -1
524 && rmp[segm->v.ref].rm_eo != -1)
526 size_t size = rmp[segm->v.ref].rm_eo
527 - rmp[segm->v.ref].rm_so;
528 ptr = input + rmp[segm->v.ref].rm_so;
529 if (case_ctl != ctl_stop)
531 ptr = run_case_conv (case_ctl, ptr, size);
532 CASE_CTL_RESET();
535 obstack_grow (&stk, ptr, size);
537 break;
539 case segm_case_ctl:
540 switch (segm->v.ctl)
542 case ctl_upcase_next:
543 case ctl_locase_next:
544 switch (save_ctl)
546 case ctl_stop:
547 case ctl_upcase:
548 case ctl_locase:
549 save_ctl = case_ctl;
550 default:
551 break;
553 /*FALL THROUGH*/
555 case ctl_upcase:
556 case ctl_locase:
557 case ctl_stop:
558 case_ctl = segm->v.ctl;
563 else
565 disp = strlen (input);
566 obstack_grow (&stk, input, disp);
569 input += disp;
571 if (tf->transform_type == transform_first)
573 obstack_grow (&stk, input, strlen (input));
574 break;
578 obstack_1grow (&stk, 0);
579 free (rmp);
582 static bool
583 _transform_name_to_obstack (int flags, char *input, char **output)
585 struct transform *tf;
586 bool alloced = false;
588 if (!stk_init)
590 obstack_init (&stk);
591 stk_init = true;
594 for (tf = transform_head; tf; tf = tf->next)
596 if (tf->flags & flags)
598 _single_transform_name_to_obstack (tf, input);
599 input = obstack_finish (&stk);
600 alloced = true;
603 *output = input;
604 return alloced;
607 bool
608 transform_name_fp (char **pinput, int flags,
609 char *(*fun)(char *, void *), void *dat)
611 char *str;
612 bool ret = _transform_name_to_obstack (flags, *pinput, &str);
613 if (ret)
615 assign_string (pinput, fun ? fun (str, dat) : str);
616 obstack_free (&stk, str);
618 else if (fun)
620 *pinput = NULL;
621 assign_string (pinput, fun (str, dat));
622 free (str);
623 ret = true;
625 return ret;
628 bool
629 transform_name (char **pinput, int type)
631 return transform_name_fp (pinput, type, NULL, NULL);
634 bool
635 transform_program_p (void)
637 return transform_head != NULL;