]> Dogcows Code - chaz/tar/blob - src/list.c
Avoid compiler warnings.
[chaz/tar] / src / list.c
1 /* List a tar archive, with support routines for reading a tar archive.
2
3 Copyright (C) 1988, 1992, 1993, 1994, 1996, 1997, 1998, 1999, 2000,
4 2001, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
5
6 Written by John Gilmore, on 1985-08-26.
7
8 This program is free software; you can redistribute it and/or modify it
9 under the terms of the GNU General Public License as published by the
10 Free Software Foundation; either version 3, or (at your option) any later
11 version.
12
13 This program is distributed in the hope that it will be useful, but
14 WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
16 Public License for more details.
17
18 You should have received a copy of the GNU General Public License along
19 with this program; if not, write to the Free Software Foundation, Inc.,
20 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
21
22 #include <system.h>
23 #include <inttostr.h>
24 #include <quotearg.h>
25
26 #include "common.h"
27
28 #define max(a, b) ((a) < (b) ? (b) : (a))
29
30 union block *current_header; /* points to current archive header */
31 enum archive_format current_format; /* recognized format */
32 union block *recent_long_name; /* recent long name header and contents */
33 union block *recent_long_link; /* likewise, for long link */
34 size_t recent_long_name_blocks; /* number of blocks in recent_long_name */
35 size_t recent_long_link_blocks; /* likewise, for long link */
36
37 static uintmax_t from_header (const char *, size_t, const char *,
38 uintmax_t, uintmax_t, bool, bool);
39
40 /* Base 64 digits; see Internet RFC 2045 Table 1. */
41 static char const base_64_digits[64] =
42 {
43 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
44 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
45 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
46 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z',
47 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '+', '/'
48 };
49
50 /* Table of base-64 digit values indexed by unsigned chars.
51 The value is 64 for unsigned chars that are not base-64 digits. */
52 static char base64_map[UCHAR_MAX + 1];
53
54 static void
55 base64_init (void)
56 {
57 int i;
58 memset (base64_map, 64, sizeof base64_map);
59 for (i = 0; i < 64; i++)
60 base64_map[(int) base_64_digits[i]] = i;
61 }
62
63 /* Main loop for reading an archive. */
64 void
65 read_and (void (*do_something) (void))
66 {
67 enum read_header status = HEADER_STILL_UNREAD;
68 enum read_header prev_status;
69 struct timespec mtime;
70
71 base64_init ();
72 name_gather ();
73
74 open_archive (ACCESS_READ);
75 do
76 {
77 prev_status = status;
78 tar_stat_destroy (&current_stat_info);
79
80 status = read_header (false);
81 switch (status)
82 {
83 case HEADER_STILL_UNREAD:
84 case HEADER_SUCCESS_EXTENDED:
85 abort ();
86
87 case HEADER_SUCCESS:
88
89 /* Valid header. We should decode next field (mode) first.
90 Ensure incoming names are null terminated. */
91
92 if (! name_match (current_stat_info.file_name)
93 || (NEWER_OPTION_INITIALIZED (newer_mtime_option)
94 /* FIXME: We get mtime now, and again later; this causes
95 duplicate diagnostics if header.mtime is bogus. */
96 && ((mtime.tv_sec
97 = TIME_FROM_HEADER (current_header->header.mtime)),
98 /* FIXME: Grab fractional time stamps from
99 extended header. */
100 mtime.tv_nsec = 0,
101 current_stat_info.mtime = mtime,
102 OLDER_TAR_STAT_TIME (current_stat_info, m)))
103 || excluded_name (current_stat_info.file_name))
104 {
105 switch (current_header->header.typeflag)
106 {
107 case GNUTYPE_VOLHDR:
108 case GNUTYPE_MULTIVOL:
109 break;
110
111 case DIRTYPE:
112 if (show_omitted_dirs_option)
113 WARN ((0, 0, _("%s: Omitting"),
114 quotearg_colon (current_stat_info.file_name)));
115 /* Fall through. */
116 default:
117 decode_header (current_header,
118 &current_stat_info, &current_format, 0);
119 skip_member ();
120 continue;
121 }
122 }
123
124 (*do_something) ();
125 continue;
126
127 case HEADER_ZERO_BLOCK:
128 if (block_number_option)
129 {
130 char buf[UINTMAX_STRSIZE_BOUND];
131 fprintf (stdlis, _("block %s: ** Block of NULs **\n"),
132 STRINGIFY_BIGINT (current_block_ordinal (), buf));
133 }
134
135 set_next_block_after (current_header);
136
137 if (!ignore_zeros_option)
138 {
139 char buf[UINTMAX_STRSIZE_BOUND];
140
141 status = read_header (false);
142 if (status == HEADER_ZERO_BLOCK)
143 break;
144 WARN ((0, 0, _("A lone zero block at %s"),
145 STRINGIFY_BIGINT (current_block_ordinal (), buf)));
146 break;
147 }
148 status = prev_status;
149 continue;
150
151 case HEADER_END_OF_FILE:
152 if (block_number_option)
153 {
154 char buf[UINTMAX_STRSIZE_BOUND];
155 fprintf (stdlis, _("block %s: ** End of File **\n"),
156 STRINGIFY_BIGINT (current_block_ordinal (), buf));
157 }
158 break;
159
160 case HEADER_FAILURE:
161 /* If the previous header was good, tell them that we are
162 skipping bad ones. */
163 set_next_block_after (current_header);
164 switch (prev_status)
165 {
166 case HEADER_STILL_UNREAD:
167 ERROR ((0, 0, _("This does not look like a tar archive")));
168 /* Fall through. */
169
170 case HEADER_ZERO_BLOCK:
171 case HEADER_SUCCESS:
172 if (block_number_option)
173 {
174 char buf[UINTMAX_STRSIZE_BOUND];
175 off_t block_ordinal = current_block_ordinal ();
176 block_ordinal -= recent_long_name_blocks;
177 block_ordinal -= recent_long_link_blocks;
178 fprintf (stdlis, _("block %s: "),
179 STRINGIFY_BIGINT (block_ordinal, buf));
180 }
181 ERROR ((0, 0, _("Skipping to next header")));
182 break;
183
184 case HEADER_END_OF_FILE:
185 case HEADER_FAILURE:
186 /* We are in the middle of a cascade of errors. */
187 break;
188
189 case HEADER_SUCCESS_EXTENDED:
190 abort ();
191 }
192 continue;
193 }
194 break;
195 }
196 while (!all_names_found (&current_stat_info));
197
198 close_archive ();
199 names_notfound (); /* print names not found */
200 }
201
202 /* Print a header block, based on tar options. */
203 void
204 list_archive (void)
205 {
206 off_t block_ordinal = current_block_ordinal ();
207 /* Print the header block. */
208
209 decode_header (current_header, &current_stat_info, &current_format, 0);
210 if (verbose_option)
211 print_header (&current_stat_info, block_ordinal);
212
213 if (incremental_option)
214 {
215 if (verbose_option > 2)
216 {
217 if (is_dumpdir (&current_stat_info))
218 list_dumpdir (current_stat_info.dumpdir,
219 dumpdir_size (current_stat_info.dumpdir));
220 }
221 }
222
223 skip_member ();
224 }
225
226 /* Check header checksum */
227 /* The standard BSD tar sources create the checksum by adding up the
228 bytes in the header as type char. I think the type char was unsigned
229 on the PDP-11, but it's signed on the Next and Sun. It looks like the
230 sources to BSD tar were never changed to compute the checksum
231 correctly, so both the Sun and Next add the bytes of the header as
232 signed chars. This doesn't cause a problem until you get a file with
233 a name containing characters with the high bit set. So tar_checksum
234 computes two checksums -- signed and unsigned. */
235
236 enum read_header
237 tar_checksum (union block *header, bool silent)
238 {
239 size_t i;
240 int unsigned_sum = 0; /* the POSIX one :-) */
241 int signed_sum = 0; /* the Sun one :-( */
242 int recorded_sum;
243 uintmax_t parsed_sum;
244 char *p;
245
246 p = header->buffer;
247 for (i = sizeof *header; i-- != 0;)
248 {
249 unsigned_sum += (unsigned char) *p;
250 signed_sum += (signed char) (*p++);
251 }
252
253 if (unsigned_sum == 0)
254 return HEADER_ZERO_BLOCK;
255
256 /* Adjust checksum to count the "chksum" field as blanks. */
257
258 for (i = sizeof header->header.chksum; i-- != 0;)
259 {
260 unsigned_sum -= (unsigned char) header->header.chksum[i];
261 signed_sum -= (signed char) (header->header.chksum[i]);
262 }
263 unsigned_sum += ' ' * sizeof header->header.chksum;
264 signed_sum += ' ' * sizeof header->header.chksum;
265
266 parsed_sum = from_header (header->header.chksum,
267 sizeof header->header.chksum, 0,
268 (uintmax_t) 0,
269 (uintmax_t) TYPE_MAXIMUM (int), true, silent);
270 if (parsed_sum == (uintmax_t) -1)
271 return HEADER_FAILURE;
272
273 recorded_sum = parsed_sum;
274
275 if (unsigned_sum != recorded_sum && signed_sum != recorded_sum)
276 return HEADER_FAILURE;
277
278 return HEADER_SUCCESS;
279 }
280
281 /* Read a block that's supposed to be a header block. Return its
282 address in "current_header", and if it is good, the file's size
283 and names (file name, link name) in *info.
284
285 Return 1 for success, 0 if the checksum is bad, EOF on eof, 2 for a
286 block full of zeros (EOF marker).
287
288 If RAW_EXTENDED_HEADERS is nonzero, do not automagically fold the
289 GNU long name and link headers into later headers.
290
291 You must always set_next_block_after(current_header) to skip past
292 the header which this routine reads. */
293
294 enum read_header
295 read_header_primitive (bool raw_extended_headers, struct tar_stat_info *info)
296 {
297 union block *header;
298 union block *header_copy;
299 char *bp;
300 union block *data_block;
301 size_t size, written;
302 union block *next_long_name = 0;
303 union block *next_long_link = 0;
304 size_t next_long_name_blocks = 0;
305 size_t next_long_link_blocks = 0;
306
307 while (1)
308 {
309 enum read_header status;
310
311 header = find_next_block ();
312 current_header = header;
313 if (!header)
314 return HEADER_END_OF_FILE;
315
316 if ((status = tar_checksum (header, false)) != HEADER_SUCCESS)
317 return status;
318
319 /* Good block. Decode file size and return. */
320
321 if (header->header.typeflag == LNKTYPE)
322 info->stat.st_size = 0; /* links 0 size on tape */
323 else
324 info->stat.st_size = OFF_FROM_HEADER (header->header.size);
325
326 if (header->header.typeflag == GNUTYPE_LONGNAME
327 || header->header.typeflag == GNUTYPE_LONGLINK
328 || header->header.typeflag == XHDTYPE
329 || header->header.typeflag == XGLTYPE
330 || header->header.typeflag == SOLARIS_XHDTYPE)
331 {
332 if (raw_extended_headers)
333 return HEADER_SUCCESS_EXTENDED;
334 else if (header->header.typeflag == GNUTYPE_LONGNAME
335 || header->header.typeflag == GNUTYPE_LONGLINK)
336 {
337 size_t name_size = info->stat.st_size;
338 size_t n = name_size % BLOCKSIZE;
339 size = name_size + BLOCKSIZE;
340 if (n)
341 size += BLOCKSIZE - n;
342
343 if (name_size != info->stat.st_size || size < name_size)
344 xalloc_die ();
345
346 header_copy = xmalloc (size + 1);
347
348 if (header->header.typeflag == GNUTYPE_LONGNAME)
349 {
350 if (next_long_name)
351 free (next_long_name);
352 next_long_name = header_copy;
353 next_long_name_blocks = size / BLOCKSIZE;
354 }
355 else
356 {
357 if (next_long_link)
358 free (next_long_link);
359 next_long_link = header_copy;
360 next_long_link_blocks = size / BLOCKSIZE;
361 }
362
363 set_next_block_after (header);
364 *header_copy = *header;
365 bp = header_copy->buffer + BLOCKSIZE;
366
367 for (size -= BLOCKSIZE; size > 0; size -= written)
368 {
369 data_block = find_next_block ();
370 if (! data_block)
371 {
372 ERROR ((0, 0, _("Unexpected EOF in archive")));
373 break;
374 }
375 written = available_space_after (data_block);
376 if (written > size)
377 written = size;
378
379 memcpy (bp, data_block->buffer, written);
380 bp += written;
381 set_next_block_after ((union block *)
382 (data_block->buffer + written - 1));
383 }
384
385 *bp = '\0';
386 }
387 else if (header->header.typeflag == XHDTYPE
388 || header->header.typeflag == SOLARIS_XHDTYPE)
389 xheader_read (&info->xhdr, header,
390 OFF_FROM_HEADER (header->header.size));
391 else if (header->header.typeflag == XGLTYPE)
392 {
393 struct xheader xhdr;
394 memset (&xhdr, 0, sizeof xhdr);
395 xheader_read (&xhdr, header,
396 OFF_FROM_HEADER (header->header.size));
397 xheader_decode_global (&xhdr);
398 xheader_destroy (&xhdr);
399 }
400
401 /* Loop! */
402
403 }
404 else
405 {
406 char const *name;
407 struct posix_header const *h = &current_header->header;
408 char namebuf[sizeof h->prefix + 1 + NAME_FIELD_SIZE + 1];
409
410 if (recent_long_name)
411 free (recent_long_name);
412
413 if (next_long_name)
414 {
415 name = next_long_name->buffer + BLOCKSIZE;
416 recent_long_name = next_long_name;
417 recent_long_name_blocks = next_long_name_blocks;
418 }
419 else
420 {
421 /* Accept file names as specified by POSIX.1-1996
422 section 10.1.1. */
423 char *np = namebuf;
424
425 if (h->prefix[0] && strcmp (h->magic, TMAGIC) == 0)
426 {
427 memcpy (np, h->prefix, sizeof h->prefix);
428 np[sizeof h->prefix] = '\0';
429 np += strlen (np);
430 *np++ = '/';
431 }
432 memcpy (np, h->name, sizeof h->name);
433 np[sizeof h->name] = '\0';
434 name = namebuf;
435 recent_long_name = 0;
436 recent_long_name_blocks = 0;
437 }
438 assign_string (&info->orig_file_name, name);
439 assign_string (&info->file_name, name);
440 info->had_trailing_slash = strip_trailing_slashes (info->file_name);
441
442 if (recent_long_link)
443 free (recent_long_link);
444
445 if (next_long_link)
446 {
447 name = next_long_link->buffer + BLOCKSIZE;
448 recent_long_link = next_long_link;
449 recent_long_link_blocks = next_long_link_blocks;
450 }
451 else
452 {
453 memcpy (namebuf, h->linkname, sizeof h->linkname);
454 namebuf[sizeof h->linkname] = '\0';
455 name = namebuf;
456 recent_long_link = 0;
457 recent_long_link_blocks = 0;
458 }
459 assign_string (&info->link_name, name);
460
461 return HEADER_SUCCESS;
462 }
463 }
464 }
465
466 enum read_header
467 read_header (bool raw_extended_headers)
468 {
469 return read_header_primitive (raw_extended_headers, &current_stat_info);
470 }
471
472 static char *
473 decode_xform (char *file_name, void *data)
474 {
475 xform_type type = *(xform_type*)data;
476
477 switch (type)
478 {
479 case xform_symlink:
480 /* FIXME: It is not quite clear how and to which extent are the symbolic
481 links subject to filename transformation. In the absence of another
482 solution, symbolic links are exempt from component stripping and
483 name suffix normalization, but subject to filename transformation
484 proper. */
485 return file_name;
486
487 case xform_link:
488 file_name = safer_name_suffix (file_name, true, absolute_names_option);
489 break;
490
491 case xform_regfile:
492 file_name = safer_name_suffix (file_name, false, absolute_names_option);
493 break;
494 }
495
496 if (strip_name_components)
497 {
498 size_t prefix_len = stripped_prefix_len (file_name,
499 strip_name_components);
500 if (prefix_len == (size_t) -1)
501 prefix_len = strlen (file_name);
502 file_name += prefix_len;
503 }
504 return file_name;
505 }
506
507 bool
508 transform_member_name (char **pinput, xform_type type)
509 {
510 return transform_name_fp (pinput, decode_xform, &type);
511 }
512
513 #define ISOCTAL(c) ((c)>='0'&&(c)<='7')
514
515 /* Decode things from a file HEADER block into STAT_INFO, also setting
516 *FORMAT_POINTER depending on the header block format. If
517 DO_USER_GROUP, decode the user/group information (this is useful
518 for extraction, but waste time when merely listing).
519
520 read_header() has already decoded the checksum and length, so we don't.
521
522 This routine should *not* be called twice for the same block, since
523 the two calls might use different DO_USER_GROUP values and thus
524 might end up with different uid/gid for the two calls. If anybody
525 wants the uid/gid they should decode it first, and other callers
526 should decode it without uid/gid before calling a routine,
527 e.g. print_header, that assumes decoded data. */
528 void
529 decode_header (union block *header, struct tar_stat_info *stat_info,
530 enum archive_format *format_pointer, int do_user_group)
531 {
532 enum archive_format format;
533
534 if (strcmp (header->header.magic, TMAGIC) == 0)
535 {
536 if (header->star_header.prefix[130] == 0
537 && ISOCTAL (header->star_header.atime[0])
538 && header->star_header.atime[11] == ' '
539 && ISOCTAL (header->star_header.ctime[0])
540 && header->star_header.ctime[11] == ' ')
541 format = STAR_FORMAT;
542 else if (stat_info->xhdr.size)
543 format = POSIX_FORMAT;
544 else
545 format = USTAR_FORMAT;
546 }
547 else if (strcmp (header->header.magic, OLDGNU_MAGIC) == 0)
548 format = OLDGNU_FORMAT;
549 else
550 format = V7_FORMAT;
551 *format_pointer = format;
552
553 stat_info->stat.st_mode = MODE_FROM_HEADER (header->header.mode);
554 stat_info->mtime.tv_sec = TIME_FROM_HEADER (header->header.mtime);
555 stat_info->mtime.tv_nsec = 0;
556 assign_string (&stat_info->uname,
557 header->header.uname[0] ? header->header.uname : NULL);
558 assign_string (&stat_info->gname,
559 header->header.gname[0] ? header->header.gname : NULL);
560
561 if (format == OLDGNU_FORMAT && incremental_option)
562 {
563 stat_info->atime.tv_sec = TIME_FROM_HEADER (header->oldgnu_header.atime);
564 stat_info->ctime.tv_sec = TIME_FROM_HEADER (header->oldgnu_header.ctime);
565 stat_info->atime.tv_nsec = stat_info->ctime.tv_nsec = 0;
566 }
567 else if (format == STAR_FORMAT)
568 {
569 stat_info->atime.tv_sec = TIME_FROM_HEADER (header->star_header.atime);
570 stat_info->ctime.tv_sec = TIME_FROM_HEADER (header->star_header.ctime);
571 stat_info->atime.tv_nsec = stat_info->ctime.tv_nsec = 0;
572 }
573 else
574 stat_info->atime = stat_info->ctime = start_time;
575
576 if (format == V7_FORMAT)
577 {
578 stat_info->stat.st_uid = UID_FROM_HEADER (header->header.uid);
579 stat_info->stat.st_gid = GID_FROM_HEADER (header->header.gid);
580 stat_info->stat.st_rdev = 0;
581 }
582 else
583 {
584 if (do_user_group)
585 {
586 /* FIXME: Decide if this should somewhat depend on -p. */
587
588 if (numeric_owner_option
589 || !*header->header.uname
590 || !uname_to_uid (header->header.uname, &stat_info->stat.st_uid))
591 stat_info->stat.st_uid = UID_FROM_HEADER (header->header.uid);
592
593 if (numeric_owner_option
594 || !*header->header.gname
595 || !gname_to_gid (header->header.gname, &stat_info->stat.st_gid))
596 stat_info->stat.st_gid = GID_FROM_HEADER (header->header.gid);
597 }
598
599 switch (header->header.typeflag)
600 {
601 case BLKTYPE:
602 case CHRTYPE:
603 stat_info->stat.st_rdev =
604 makedev (MAJOR_FROM_HEADER (header->header.devmajor),
605 MINOR_FROM_HEADER (header->header.devminor));
606 break;
607
608 default:
609 stat_info->stat.st_rdev = 0;
610 }
611 }
612
613 stat_info->archive_file_size = stat_info->stat.st_size;
614 xheader_decode (stat_info);
615
616 if (sparse_member_p (stat_info))
617 {
618 sparse_fixup_header (stat_info);
619 stat_info->is_sparse = true;
620 }
621 else
622 {
623 stat_info->is_sparse = false;
624 if (((current_format == GNU_FORMAT
625 || current_format == OLDGNU_FORMAT)
626 && current_header->header.typeflag == GNUTYPE_DUMPDIR)
627 || stat_info->dumpdir)
628 stat_info->is_dumpdir = true;
629 }
630
631 transform_member_name (&stat_info->file_name, xform_regfile);
632 }
633
634 /* Convert buffer at WHERE0 of size DIGS from external format to
635 uintmax_t. DIGS must be positive. If TYPE is nonnull, the data
636 are of type TYPE. The buffer must represent a value in the range
637 -MINUS_MINVAL through MAXVAL. If OCTAL_ONLY, allow only octal
638 numbers instead of the other GNU extensions. Return -1 on error,
639 diagnosing the error if TYPE is nonnull and if !SILENT. */
640 static uintmax_t
641 from_header (char const *where0, size_t digs, char const *type,
642 uintmax_t minus_minval, uintmax_t maxval,
643 bool octal_only, bool silent)
644 {
645 uintmax_t value;
646 char const *where = where0;
647 char const *lim = where + digs;
648 int negative = 0;
649
650 /* Accommodate buggy tar of unknown vintage, which outputs leading
651 NUL if the previous field overflows. */
652 where += !*where;
653
654 /* Accommodate older tars, which output leading spaces. */
655 for (;;)
656 {
657 if (where == lim)
658 {
659 if (type && !silent)
660 ERROR ((0, 0,
661 /* TRANSLATORS: %s is type of the value (gid_t, uid_t, etc.) */
662 _("Blanks in header where numeric %s value expected"),
663 type));
664 return -1;
665 }
666 if (!ISSPACE ((unsigned char) *where))
667 break;
668 where++;
669 }
670
671 value = 0;
672 if (ISODIGIT (*where))
673 {
674 char const *where1 = where;
675 uintmax_t overflow = 0;
676
677 for (;;)
678 {
679 value += *where++ - '0';
680 if (where == lim || ! ISODIGIT (*where))
681 break;
682 overflow |= value ^ (value << LG_8 >> LG_8);
683 value <<= LG_8;
684 }
685
686 /* Parse the output of older, unportable tars, which generate
687 negative values in two's complement octal. If the leading
688 nonzero digit is 1, we can't recover the original value
689 reliably; so do this only if the digit is 2 or more. This
690 catches the common case of 32-bit negative time stamps. */
691 if ((overflow || maxval < value) && '2' <= *where1 && type)
692 {
693 /* Compute the negative of the input value, assuming two's
694 complement. */
695 int digit = (*where1 - '0') | 4;
696 overflow = 0;
697 value = 0;
698 where = where1;
699 for (;;)
700 {
701 value += 7 - digit;
702 where++;
703 if (where == lim || ! ISODIGIT (*where))
704 break;
705 digit = *where - '0';
706 overflow |= value ^ (value << LG_8 >> LG_8);
707 value <<= LG_8;
708 }
709 value++;
710 overflow |= !value;
711
712 if (!overflow && value <= minus_minval)
713 {
714 if (!silent)
715 WARN ((0, 0,
716 /* TRANSLATORS: Second %s is a type name (gid_t,uid_t,etc.) */
717 _("Archive octal value %.*s is out of %s range; assuming two's complement"),
718 (int) (where - where1), where1, type));
719 negative = 1;
720 }
721 }
722
723 if (overflow)
724 {
725 if (type && !silent)
726 ERROR ((0, 0,
727 /* TRANSLATORS: Second %s is a type name (gid_t,uid_t,etc.) */
728 _("Archive octal value %.*s is out of %s range"),
729 (int) (where - where1), where1, type));
730 return -1;
731 }
732 }
733 else if (octal_only)
734 {
735 /* Suppress the following extensions. */
736 }
737 else if (*where == '-' || *where == '+')
738 {
739 /* Parse base-64 output produced only by tar test versions
740 1.13.6 (1999-08-11) through 1.13.11 (1999-08-23).
741 Support for this will be withdrawn in future releases. */
742 int dig;
743 if (!silent)
744 {
745 static bool warned_once;
746 if (! warned_once)
747 {
748 warned_once = true;
749 WARN ((0, 0, _("Archive contains obsolescent base-64 headers")));
750 }
751 }
752 negative = *where++ == '-';
753 while (where != lim
754 && (dig = base64_map[(unsigned char) *where]) < 64)
755 {
756 if (value << LG_64 >> LG_64 != value)
757 {
758 char *string = alloca (digs + 1);
759 memcpy (string, where0, digs);
760 string[digs] = '\0';
761 if (type && !silent)
762 ERROR ((0, 0,
763 _("Archive signed base-64 string %s is out of %s range"),
764 quote (string), type));
765 return -1;
766 }
767 value = (value << LG_64) | dig;
768 where++;
769 }
770 }
771 else if (*where == '\200' /* positive base-256 */
772 || *where == '\377' /* negative base-256 */)
773 {
774 /* Parse base-256 output. A nonnegative number N is
775 represented as (256**DIGS)/2 + N; a negative number -N is
776 represented as (256**DIGS) - N, i.e. as two's complement.
777 The representation guarantees that the leading bit is
778 always on, so that we don't confuse this format with the
779 others (assuming ASCII bytes of 8 bits or more). */
780 int signbit = *where & (1 << (LG_256 - 2));
781 uintmax_t topbits = (((uintmax_t) - signbit)
782 << (CHAR_BIT * sizeof (uintmax_t)
783 - LG_256 - (LG_256 - 2)));
784 value = (*where++ & ((1 << (LG_256 - 2)) - 1)) - signbit;
785 for (;;)
786 {
787 value = (value << LG_256) + (unsigned char) *where++;
788 if (where == lim)
789 break;
790 if (((value << LG_256 >> LG_256) | topbits) != value)
791 {
792 if (type && !silent)
793 ERROR ((0, 0,
794 _("Archive base-256 value is out of %s range"),
795 type));
796 return -1;
797 }
798 }
799 negative = signbit;
800 if (negative)
801 value = -value;
802 }
803
804 if (where != lim && *where && !ISSPACE ((unsigned char) *where))
805 {
806 if (type)
807 {
808 char buf[1000]; /* Big enough to represent any header. */
809 static struct quoting_options *o;
810
811 if (!o)
812 {
813 o = clone_quoting_options (0);
814 set_quoting_style (o, locale_quoting_style);
815 }
816
817 while (where0 != lim && ! lim[-1])
818 lim--;
819 quotearg_buffer (buf, sizeof buf, where0, lim - where, o);
820 if (!silent)
821 ERROR ((0, 0,
822 /* TRANSLATORS: Second %s is a type name (gid_t,uid_t,etc.) */
823 _("Archive contains %.*s where numeric %s value expected"),
824 (int) sizeof buf, buf, type));
825 }
826
827 return -1;
828 }
829
830 if (value <= (negative ? minus_minval : maxval))
831 return negative ? -value : value;
832
833 if (type && !silent)
834 {
835 char minval_buf[UINTMAX_STRSIZE_BOUND + 1];
836 char maxval_buf[UINTMAX_STRSIZE_BOUND];
837 char value_buf[UINTMAX_STRSIZE_BOUND + 1];
838 char *minval_string = STRINGIFY_BIGINT (minus_minval, minval_buf + 1);
839 char *value_string = STRINGIFY_BIGINT (value, value_buf + 1);
840 if (negative)
841 *--value_string = '-';
842 if (minus_minval)
843 *--minval_string = '-';
844 /* TRANSLATORS: Second %s is type name (gid_t,uid_t,etc.) */
845 ERROR ((0, 0, _("Archive value %s is out of %s range %s..%s"),
846 value_string, type,
847 minval_string, STRINGIFY_BIGINT (maxval, maxval_buf)));
848 }
849
850 return -1;
851 }
852
853 gid_t
854 gid_from_header (const char *p, size_t s)
855 {
856 return from_header (p, s, "gid_t",
857 - (uintmax_t) TYPE_MINIMUM (gid_t),
858 (uintmax_t) TYPE_MAXIMUM (gid_t),
859 false, false);
860 }
861
862 major_t
863 major_from_header (const char *p, size_t s)
864 {
865 return from_header (p, s, "major_t",
866 - (uintmax_t) TYPE_MINIMUM (major_t),
867 (uintmax_t) TYPE_MAXIMUM (major_t), false, false);
868 }
869
870 minor_t
871 minor_from_header (const char *p, size_t s)
872 {
873 return from_header (p, s, "minor_t",
874 - (uintmax_t) TYPE_MINIMUM (minor_t),
875 (uintmax_t) TYPE_MAXIMUM (minor_t), false, false);
876 }
877
878 mode_t
879 mode_from_header (const char *p, size_t s)
880 {
881 /* Do not complain about unrecognized mode bits. */
882 unsigned u = from_header (p, s, "mode_t",
883 - (uintmax_t) TYPE_MINIMUM (mode_t),
884 TYPE_MAXIMUM (uintmax_t), false, false);
885 return ((u & TSUID ? S_ISUID : 0)
886 | (u & TSGID ? S_ISGID : 0)
887 | (u & TSVTX ? S_ISVTX : 0)
888 | (u & TUREAD ? S_IRUSR : 0)
889 | (u & TUWRITE ? S_IWUSR : 0)
890 | (u & TUEXEC ? S_IXUSR : 0)
891 | (u & TGREAD ? S_IRGRP : 0)
892 | (u & TGWRITE ? S_IWGRP : 0)
893 | (u & TGEXEC ? S_IXGRP : 0)
894 | (u & TOREAD ? S_IROTH : 0)
895 | (u & TOWRITE ? S_IWOTH : 0)
896 | (u & TOEXEC ? S_IXOTH : 0));
897 }
898
899 off_t
900 off_from_header (const char *p, size_t s)
901 {
902 /* Negative offsets are not allowed in tar files, so invoke
903 from_header with minimum value 0, not TYPE_MINIMUM (off_t). */
904 return from_header (p, s, "off_t", (uintmax_t) 0,
905 (uintmax_t) TYPE_MAXIMUM (off_t), false, false);
906 }
907
908 size_t
909 size_from_header (const char *p, size_t s)
910 {
911 return from_header (p, s, "size_t", (uintmax_t) 0,
912 (uintmax_t) TYPE_MAXIMUM (size_t), false, false);
913 }
914
915 time_t
916 time_from_header (const char *p, size_t s)
917 {
918 return from_header (p, s, "time_t",
919 - (uintmax_t) TYPE_MINIMUM (time_t),
920 (uintmax_t) TYPE_MAXIMUM (time_t), false, false);
921 }
922
923 uid_t
924 uid_from_header (const char *p, size_t s)
925 {
926 return from_header (p, s, "uid_t",
927 - (uintmax_t) TYPE_MINIMUM (uid_t),
928 (uintmax_t) TYPE_MAXIMUM (uid_t), false, false);
929 }
930
931 uintmax_t
932 uintmax_from_header (const char *p, size_t s)
933 {
934 return from_header (p, s, "uintmax_t", (uintmax_t) 0,
935 TYPE_MAXIMUM (uintmax_t), false, false);
936 }
937
938
939 /* Return a printable representation of T. The result points to
940 static storage that can be reused in the next call to this
941 function, to ctime, or to asctime. If FULL_TIME, then output the
942 time stamp to its full resolution; otherwise, just output it to
943 1-minute resolution. */
944 char const *
945 tartime (struct timespec t, bool full_time)
946 {
947 enum { fraclen = sizeof ".FFFFFFFFF" - 1 };
948 static char buffer[max (UINTMAX_STRSIZE_BOUND + 1,
949 INT_STRLEN_BOUND (int) + 16)
950 + fraclen];
951 struct tm *tm;
952 time_t s = t.tv_sec;
953 int ns = t.tv_nsec;
954 bool negative = s < 0;
955 char *p;
956
957 if (negative && ns != 0)
958 {
959 s++;
960 ns = 1000000000 - ns;
961 }
962
963 tm = utc_option ? gmtime (&s) : localtime (&s);
964 if (tm)
965 {
966 if (full_time)
967 {
968 sprintf (buffer, "%04ld-%02d-%02d %02d:%02d:%02d",
969 tm->tm_year + 1900L, tm->tm_mon + 1, tm->tm_mday,
970 tm->tm_hour, tm->tm_min, tm->tm_sec);
971 code_ns_fraction (ns, buffer + strlen (buffer));
972 }
973 else
974 sprintf (buffer, "%04ld-%02d-%02d %02d:%02d",
975 tm->tm_year + 1900L, tm->tm_mon + 1, tm->tm_mday,
976 tm->tm_hour, tm->tm_min);
977 return buffer;
978 }
979
980 /* The time stamp cannot be broken down, most likely because it
981 is out of range. Convert it as an integer,
982 right-adjusted in a field with the same width as the usual
983 4-year ISO time format. */
984 p = umaxtostr (negative ? - (uintmax_t) s : s,
985 buffer + sizeof buffer - UINTMAX_STRSIZE_BOUND - fraclen);
986 if (negative)
987 *--p = '-';
988 while ((buffer + sizeof buffer - sizeof "YYYY-MM-DD HH:MM"
989 + (full_time ? sizeof ":SS.FFFFFFFFF" - 1 : 0))
990 < p)
991 *--p = ' ';
992 if (full_time)
993 code_ns_fraction (ns, buffer + sizeof buffer - 1 - fraclen);
994 return p;
995 }
996
997 /* Actually print it.
998
999 Plain and fancy file header block logging. Non-verbose just prints
1000 the name, e.g. for "tar t" or "tar x". This should just contain
1001 file names, so it can be fed back into tar with xargs or the "-T"
1002 option. The verbose option can give a bunch of info, one line per
1003 file. I doubt anybody tries to parse its format, or if they do,
1004 they shouldn't. Unix tar is pretty random here anyway. */
1005
1006
1007 /* FIXME: Note that print_header uses the globals HEAD, HSTAT, and
1008 HEAD_STANDARD, which must be set up in advance. Not very clean.. */
1009
1010 /* Width of "user/group size", with initial value chosen
1011 heuristically. This grows as needed, though this may cause some
1012 stairstepping in the output. Make it too small and the output will
1013 almost always look ragged. Make it too large and the output will
1014 be spaced out too far. */
1015 static int ugswidth = 19;
1016
1017 /* Width of printed time stamps. It grows if longer time stamps are
1018 found (typically, those with nanosecond resolution). Like
1019 USGWIDTH, some stairstepping may occur. */
1020 static int datewidth = sizeof "YYYY-MM-DD HH:MM" - 1;
1021
1022 void
1023 print_header (struct tar_stat_info *st, off_t block_ordinal)
1024 {
1025 char modes[11];
1026 char const *time_stamp;
1027 int time_stamp_len;
1028 char *temp_name;
1029
1030 /* These hold formatted ints. */
1031 char uform[UINTMAX_STRSIZE_BOUND], gform[UINTMAX_STRSIZE_BOUND];
1032 char *user, *group;
1033 char size[2 * UINTMAX_STRSIZE_BOUND];
1034 /* holds formatted size or major,minor */
1035 char uintbuf[UINTMAX_STRSIZE_BOUND];
1036 int pad;
1037 int sizelen;
1038
1039 if (test_label_option && current_header->header.typeflag != GNUTYPE_VOLHDR)
1040 return;
1041
1042 if (show_transformed_names_option)
1043 temp_name = st->file_name ? st->file_name : st->orig_file_name;
1044 else
1045 temp_name = st->orig_file_name ? st->orig_file_name : st->file_name;
1046
1047 if (block_number_option)
1048 {
1049 char buf[UINTMAX_STRSIZE_BOUND];
1050 if (block_ordinal < 0)
1051 block_ordinal = current_block_ordinal ();
1052 block_ordinal -= recent_long_name_blocks;
1053 block_ordinal -= recent_long_link_blocks;
1054 fprintf (stdlis, _("block %s: "),
1055 STRINGIFY_BIGINT (block_ordinal, buf));
1056 }
1057
1058 if (verbose_option <= 1)
1059 {
1060 /* Just the fax, mam. */
1061 fprintf (stdlis, "%s\n", quotearg (temp_name));
1062 }
1063 else
1064 {
1065 /* File type and modes. */
1066
1067 modes[0] = '?';
1068 switch (current_header->header.typeflag)
1069 {
1070 case GNUTYPE_VOLHDR:
1071 modes[0] = 'V';
1072 break;
1073
1074 case GNUTYPE_MULTIVOL:
1075 modes[0] = 'M';
1076 break;
1077
1078 case GNUTYPE_LONGNAME:
1079 case GNUTYPE_LONGLINK:
1080 modes[0] = 'L';
1081 ERROR ((0, 0, _("Unexpected long name header")));
1082 break;
1083
1084 case GNUTYPE_SPARSE:
1085 case REGTYPE:
1086 case AREGTYPE:
1087 modes[0] = '-';
1088 if (temp_name[strlen (temp_name) - 1] == '/')
1089 modes[0] = 'd';
1090 break;
1091 case LNKTYPE:
1092 modes[0] = 'h';
1093 break;
1094 case GNUTYPE_DUMPDIR:
1095 modes[0] = 'd';
1096 break;
1097 case DIRTYPE:
1098 modes[0] = 'd';
1099 break;
1100 case SYMTYPE:
1101 modes[0] = 'l';
1102 break;
1103 case BLKTYPE:
1104 modes[0] = 'b';
1105 break;
1106 case CHRTYPE:
1107 modes[0] = 'c';
1108 break;
1109 case FIFOTYPE:
1110 modes[0] = 'p';
1111 break;
1112 case CONTTYPE:
1113 modes[0] = 'C';
1114 break;
1115 }
1116
1117 pax_decode_mode (st->stat.st_mode, modes + 1);
1118
1119 /* Time stamp. */
1120
1121 time_stamp = tartime (st->mtime, false);
1122 time_stamp_len = strlen (time_stamp);
1123 if (datewidth < time_stamp_len)
1124 datewidth = time_stamp_len;
1125
1126 /* User and group names. */
1127
1128 if (st->uname
1129 && st->uname[0]
1130 && current_format != V7_FORMAT
1131 && !numeric_owner_option)
1132 user = st->uname;
1133 else
1134 {
1135 /* Try parsing it as an unsigned integer first, and as a
1136 uid_t if that fails. This method can list positive user
1137 ids that are too large to fit in a uid_t. */
1138 uintmax_t u = from_header (current_header->header.uid,
1139 sizeof current_header->header.uid, 0,
1140 (uintmax_t) 0,
1141 (uintmax_t) TYPE_MAXIMUM (uintmax_t),
1142 false, false);
1143 if (u != -1)
1144 user = STRINGIFY_BIGINT (u, uform);
1145 else
1146 {
1147 sprintf (uform, "%ld",
1148 (long) UID_FROM_HEADER (current_header->header.uid));
1149 user = uform;
1150 }
1151 }
1152
1153 if (st->gname
1154 && st->gname[0]
1155 && current_format != V7_FORMAT
1156 && !numeric_owner_option)
1157 group = st->gname;
1158 else
1159 {
1160 /* Try parsing it as an unsigned integer first, and as a
1161 gid_t if that fails. This method can list positive group
1162 ids that are too large to fit in a gid_t. */
1163 uintmax_t g = from_header (current_header->header.gid,
1164 sizeof current_header->header.gid, 0,
1165 (uintmax_t) 0,
1166 (uintmax_t) TYPE_MAXIMUM (uintmax_t),
1167 false, false);
1168 if (g != -1)
1169 group = STRINGIFY_BIGINT (g, gform);
1170 else
1171 {
1172 sprintf (gform, "%ld",
1173 (long) GID_FROM_HEADER (current_header->header.gid));
1174 group = gform;
1175 }
1176 }
1177
1178 /* Format the file size or major/minor device numbers. */
1179
1180 switch (current_header->header.typeflag)
1181 {
1182 case CHRTYPE:
1183 case BLKTYPE:
1184 strcpy (size,
1185 STRINGIFY_BIGINT (major (st->stat.st_rdev), uintbuf));
1186 strcat (size, ",");
1187 strcat (size,
1188 STRINGIFY_BIGINT (minor (st->stat.st_rdev), uintbuf));
1189 break;
1190
1191 default:
1192 /* st->stat.st_size keeps stored file size */
1193 strcpy (size, STRINGIFY_BIGINT (st->stat.st_size, uintbuf));
1194 break;
1195 }
1196
1197 /* Figure out padding and print the whole line. */
1198
1199 sizelen = strlen (size);
1200 pad = strlen (user) + 1 + strlen (group) + 1 + sizelen;
1201 if (pad > ugswidth)
1202 ugswidth = pad;
1203
1204 fprintf (stdlis, "%s %s/%s %*s %-*s",
1205 modes, user, group, ugswidth - pad + sizelen, size,
1206 datewidth, time_stamp);
1207
1208 fprintf (stdlis, " %s", quotearg (temp_name));
1209
1210 switch (current_header->header.typeflag)
1211 {
1212 case SYMTYPE:
1213 fprintf (stdlis, " -> %s\n", quotearg (st->link_name));
1214 break;
1215
1216 case LNKTYPE:
1217 fprintf (stdlis, _(" link to %s\n"), quotearg (st->link_name));
1218 break;
1219
1220 default:
1221 {
1222 char type_string[2];
1223 type_string[0] = current_header->header.typeflag;
1224 type_string[1] = '\0';
1225 fprintf (stdlis, _(" unknown file type %s\n"),
1226 quote (type_string));
1227 }
1228 break;
1229
1230 case AREGTYPE:
1231 case REGTYPE:
1232 case GNUTYPE_SPARSE:
1233 case CHRTYPE:
1234 case BLKTYPE:
1235 case DIRTYPE:
1236 case FIFOTYPE:
1237 case CONTTYPE:
1238 case GNUTYPE_DUMPDIR:
1239 putc ('\n', stdlis);
1240 break;
1241
1242 case GNUTYPE_LONGLINK:
1243 fprintf (stdlis, _("--Long Link--\n"));
1244 break;
1245
1246 case GNUTYPE_LONGNAME:
1247 fprintf (stdlis, _("--Long Name--\n"));
1248 break;
1249
1250 case GNUTYPE_VOLHDR:
1251 fprintf (stdlis, _("--Volume Header--\n"));
1252 break;
1253
1254 case GNUTYPE_MULTIVOL:
1255 strcpy (size,
1256 STRINGIFY_BIGINT
1257 (UINTMAX_FROM_HEADER (current_header->oldgnu_header.offset),
1258 uintbuf));
1259 fprintf (stdlis, _("--Continued at byte %s--\n"), size);
1260 break;
1261 }
1262 }
1263 fflush (stdlis);
1264 }
1265
1266 /* Print a similar line when we make a directory automatically. */
1267 void
1268 print_for_mkdir (char *dirname, int length, mode_t mode)
1269 {
1270 char modes[11];
1271
1272 if (verbose_option > 1)
1273 {
1274 /* File type and modes. */
1275
1276 modes[0] = 'd';
1277 pax_decode_mode (mode, modes + 1);
1278
1279 if (block_number_option)
1280 {
1281 char buf[UINTMAX_STRSIZE_BOUND];
1282 fprintf (stdlis, _("block %s: "),
1283 STRINGIFY_BIGINT (current_block_ordinal (), buf));
1284 }
1285
1286 fprintf (stdlis, "%s %*s %.*s\n", modes, ugswidth + 1 + datewidth,
1287 _("Creating directory:"), length, quotearg (dirname));
1288 }
1289 }
1290
1291 /* Skip over SIZE bytes of data in blocks in the archive. */
1292 void
1293 skip_file (off_t size)
1294 {
1295 union block *x;
1296
1297 /* FIXME: Make sure mv_begin is always called before it */
1298
1299 if (seekable_archive)
1300 {
1301 off_t nblk = seek_archive (size);
1302 if (nblk >= 0)
1303 size -= nblk * BLOCKSIZE;
1304 else
1305 seekable_archive = false;
1306 }
1307
1308 mv_size_left (size);
1309
1310 while (size > 0)
1311 {
1312 x = find_next_block ();
1313 if (! x)
1314 FATAL_ERROR ((0, 0, _("Unexpected EOF in archive")));
1315
1316 set_next_block_after (x);
1317 size -= BLOCKSIZE;
1318 mv_size_left (size);
1319 }
1320 }
1321
1322 /* Skip the current member in the archive.
1323 NOTE: Current header must be decoded before calling this function. */
1324 void
1325 skip_member (void)
1326 {
1327 if (!current_stat_info.skipped)
1328 {
1329 char save_typeflag = current_header->header.typeflag;
1330 set_next_block_after (current_header);
1331
1332 mv_begin (&current_stat_info);
1333
1334 if (current_stat_info.is_sparse)
1335 sparse_skip_file (&current_stat_info);
1336 else if (save_typeflag != DIRTYPE)
1337 skip_file (current_stat_info.stat.st_size);
1338
1339 mv_end ();
1340 }
1341 }
This page took 0.102603 seconds and 5 git commands to generate.