1 /* Various processing of names.
3 Copyright (C) 1988, 1992, 1994, 1996, 1997, 1998, 1999, 2000, 2001,
4 2003 Free Software Foundation, Inc.
6 This program is free software; you can redistribute it and/or modify it
7 under the terms of the GNU General Public License as published by the
8 Free Software Foundation; either version 2, or (at your option) any later
11 This program is distributed in the hope that it will be useful, but
12 WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
14 Public License for more details.
16 You should have received a copy of the GNU General Public License along
17 with this program; if not, write to the Free Software Foundation, Inc.,
18 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
30 /* User and group names. */
32 struct group
*getgrnam ();
33 struct passwd
*getpwnam ();
34 #if ! HAVE_DECL_GETPWUID
35 struct passwd
*getpwuid ();
37 #if ! HAVE_DECL_GETGRGID
38 struct group
*getgrgid ();
41 /* Make sure you link with the proper libraries if you are running the
42 Yellow Peril (thanks for the good laugh, Ian J.!), or, euh... NIS.
43 This code should also be modified for non-UNIX systems to do something
46 static char *cached_uname
;
47 static char *cached_gname
;
49 static uid_t cached_uid
; /* valid only if cached_uname is not empty */
50 static gid_t cached_gid
; /* valid only if cached_gname is not empty */
52 /* These variables are valid only if nonempty. */
53 static char *cached_no_such_uname
;
54 static char *cached_no_such_gname
;
56 /* These variables are valid only if nonzero. It's not worth optimizing
57 the case for weird systems where 0 is not a valid uid or gid. */
58 static uid_t cached_no_such_uid
;
59 static gid_t cached_no_such_gid
;
61 /* Given UID, find the corresponding UNAME. */
63 uid_to_uname (uid_t uid
, char **uname
)
65 struct passwd
*passwd
;
67 if (uid
!= 0 && uid
== cached_no_such_uid
)
73 if (!cached_uname
|| uid
!= cached_uid
)
75 passwd
= getpwuid (uid
);
79 assign_string (&cached_uname
, passwd
->pw_name
);
83 cached_no_such_uid
= uid
;
88 *uname
= strdup (cached_uname
);
91 /* Given GID, find the corresponding GNAME. */
93 gid_to_gname (gid_t gid
, char **gname
)
97 if (gid
!= 0 && gid
== cached_no_such_gid
)
103 if (!cached_gname
|| gid
!= cached_gid
)
105 group
= getgrgid (gid
);
109 assign_string (&cached_gname
, group
->gr_name
);
113 cached_no_such_gid
= gid
;
114 *gname
= strdup ("");
118 *gname
= strdup (cached_gname
);
121 /* Given UNAME, set the corresponding UID and return 1, or else, return 0. */
123 uname_to_uid (char *uname
, uid_t
*uidp
)
125 struct passwd
*passwd
;
127 if (cached_no_such_uname
128 && strcmp (uname
, cached_no_such_uname
) == 0)
132 || uname
[0] != cached_uname
[0]
133 || strcmp (uname
, cached_uname
) != 0)
135 passwd
= getpwnam (uname
);
138 cached_uid
= passwd
->pw_uid
;
139 assign_string (&cached_uname
, passwd
->pw_name
);
143 assign_string (&cached_no_such_uname
, uname
);
151 /* Given GNAME, set the corresponding GID and return 1, or else, return 0. */
153 gname_to_gid (char *gname
, gid_t
*gidp
)
157 if (cached_no_such_gname
158 && strcmp (gname
, cached_no_such_gname
) == 0)
162 || gname
[0] != cached_gname
[0]
163 || strcmp (gname
, cached_gname
) != 0)
165 group
= getgrnam (gname
);
168 cached_gid
= group
->gr_gid
;
169 assign_string (&cached_gname
, gname
);
173 assign_string (&cached_no_such_gname
, gname
);
182 /* Names from the command call. */
184 static struct name
*namelist
; /* first name in list, if any */
185 static struct name
**nametail
= &namelist
; /* end of name list */
186 static const char **name_array
; /* store an array of names */
187 static int allocated_names
; /* how big is the array? */
188 static int names
; /* how many entries does it have? */
189 static int name_index
; /* how many of the entries have we scanned? */
191 /* Initialize structures. */
195 allocated_names
= 10;
196 name_array
= xmalloc (sizeof (const char *) * allocated_names
);
200 /* Add NAME at end of name_array, reallocating it as necessary. */
202 name_add (const char *name
)
204 if (names
== allocated_names
)
206 allocated_names
*= 2;
208 xrealloc (name_array
, sizeof (const char *) * allocated_names
);
210 name_array
[names
++] = name
;
213 /* Names from external name file. */
215 static FILE *name_file
; /* file to read names from */
216 static char *name_buffer
; /* buffer to hold the current file name */
217 static size_t name_buffer_length
; /* allocated length of name_buffer */
219 /* FIXME: I should better check more closely. It seems at first glance that
220 is_pattern is only used when reading a file, and ignored for all
221 command line arguments. */
224 is_pattern (const char *string
)
226 return strchr (string
, '*') || strchr (string
, '[') || strchr (string
, '?');
229 /* Set up to gather file names for tar. They can either come from a
230 file or were saved from decoding arguments. */
232 name_init (int argc
, char *const *argv
)
234 name_buffer
= xmalloc (NAME_FIELD_SIZE
+ 2);
235 name_buffer_length
= NAME_FIELD_SIZE
;
237 if (files_from_option
)
239 if (!strcmp (files_from_option
, "-"))
241 request_stdin ("-T");
244 else if (name_file
= fopen (files_from_option
, "r"), !name_file
)
245 open_fatal (files_from_option
);
256 /* Read the next filename from name_file and null-terminate it. Put
257 it into name_buffer, reallocating and adjusting name_buffer_length
258 if necessary. Return 0 at end of file, 1 otherwise. */
260 read_name_from_file (void)
265 /* FIXME: getc may be called even if character was EOF the last time here. */
267 /* FIXME: This + 2 allocation might serve no purpose. */
269 while (character
= getc (name_file
),
270 character
!= EOF
&& character
!= filename_terminator
)
272 if (counter
== name_buffer_length
)
274 if (name_buffer_length
* 2 < name_buffer_length
)
276 name_buffer_length
*= 2;
277 name_buffer
= xrealloc (name_buffer
, name_buffer_length
+ 2);
279 name_buffer
[counter
++] = character
;
282 if (counter
== 0 && character
== EOF
)
285 if (counter
== name_buffer_length
)
287 if (name_buffer_length
* 2 < name_buffer_length
)
289 name_buffer_length
*= 2;
290 name_buffer
= xrealloc (name_buffer
, name_buffer_length
+ 2);
292 name_buffer
[counter
] = '\0';
297 /* Get the next name from ARGV or the file of names. Result is in
298 static storage and can't be relied upon across two calls.
300 If CHANGE_DIRS is true, treat a filename of the form "-C" as
301 meaning that the next filename is the name of a directory to change
302 to. If filename_terminator is NUL, CHANGE_DIRS is effectively
305 name_next (int change_dirs
)
311 if (filename_terminator
== '\0')
316 /* Get a name, either from file or from saved arguments. */
318 if (name_index
== names
)
322 if (! read_name_from_file ())
328 source
= name_array
[name_index
++];
329 source_len
= strlen (source
);
330 if (name_buffer_length
< source_len
)
334 name_buffer_length
*= 2;
335 if (! name_buffer_length
)
338 while (name_buffer_length
< source_len
);
341 name_buffer
= xmalloc (name_buffer_length
+ 2);
343 strcpy (name_buffer
, source
);
346 /* Zap trailing slashes. */
348 cursor
= name_buffer
+ strlen (name_buffer
) - 1;
349 while (cursor
> name_buffer
&& ISSLASH (*cursor
))
354 if (chdir (name_buffer
) < 0)
355 chdir_fatal (name_buffer
);
358 else if (change_dirs
&& strcmp (name_buffer
, "-C") == 0)
362 unquote_string (name_buffer
);
367 /* No more names in file. */
369 if (name_file
&& chdir_flag
)
370 FATAL_ERROR ((0, 0, _("Missing file name after -C")));
375 /* Close the name file, if any. */
379 if (name_file
&& name_file
!= stdin
)
380 if (fclose (name_file
) != 0)
381 close_error (name_buffer
);
384 /* Gather names in a list for scanning. Could hash them later if we
387 If the names are already sorted to match the archive, we just read
388 them one by one. name_gather reads the first one, and it is called
389 by name_match as appropriate to read the next ones. At EOF, the
390 last name read is just left in the buffer. This option lets users
391 of small machines extract an arbitrary number of files by doing
392 "tar t" and editing down the list of files. */
397 /* Buffer able to hold a single name. */
398 static struct name
*buffer
;
399 static size_t allocated_size
;
403 if (same_order_option
)
405 static int change_dir
;
407 if (allocated_size
== 0)
409 allocated_size
= offsetof (struct name
, name
) + NAME_FIELD_SIZE
+ 1;
410 buffer
= xmalloc (allocated_size
);
411 /* FIXME: This memset is overkill, and ugly... */
412 memset (buffer
, 0, allocated_size
);
415 while ((name
= name_next (0)) && strcmp (name
, "-C") == 0)
417 char const *dir
= name_next (0);
419 FATAL_ERROR ((0, 0, _("Missing file name after -C")));
420 change_dir
= chdir_arg (xstrdup (dir
));
426 buffer
->length
= strlen (name
);
427 needed_size
= offsetof (struct name
, name
) + buffer
->length
+ 1;
428 if (allocated_size
< needed_size
)
433 if (! allocated_size
)
436 while (allocated_size
< needed_size
);
438 buffer
= xrealloc (buffer
, allocated_size
);
440 buffer
->change_dir
= change_dir
;
441 strcpy (buffer
->name
, name
);
446 nametail
= &namelist
->next
;
451 /* Non sorted names -- read them all in. */
456 int change_dir0
= change_dir
;
457 while ((name
= name_next (0)) && strcmp (name
, "-C") == 0)
459 char const *dir
= name_next (0);
461 FATAL_ERROR ((0, 0, _("Missing file name after -C")));
462 change_dir
= chdir_arg (xstrdup (dir
));
465 addname (name
, change_dir
);
468 if (change_dir
!= change_dir0
)
469 addname (0, change_dir
);
476 /* Add a name to the namelist. */
478 addname (char const *string
, int change_dir
)
480 size_t length
= string
? strlen (string
) : 0;
481 struct name
*name
= xmalloc (offsetof (struct name
, name
) + length
+ 1);
486 strcpy (name
->name
, string
);
492 /* FIXME: This initialization (and the byte of memory that it
493 initializes) is probably not needed, but we are currently in
494 bug-fix mode so we'll leave it in for now. */
499 name
->length
= length
;
501 name
->regexp
= 0; /* assume not a regular expression */
502 name
->firstch
= 1; /* assume first char is literal */
503 name
->change_dir
= change_dir
;
504 name
->dir_contents
= 0;
506 if (string
&& is_pattern (string
))
509 if (string
[0] == '*' || string
[0] == '[' || string
[0] == '?')
514 nametail
= &name
->next
;
518 /* Find a match for PATH (whose string length is LENGTH) in the name
521 namelist_match (char const *path
, size_t length
)
525 for (p
= namelist
; p
; p
= p
->next
)
527 /* If first chars don't match, quick skip. */
529 if (p
->firstch
&& p
->name
[0] != path
[0])
533 ? fnmatch (p
->name
, path
, recursion_option
) == 0
534 : (p
->length
<= length
535 && (path
[p
->length
] == '\0'
536 || (ISSLASH (path
[p
->length
]) && recursion_option
))
537 && memcmp (path
, p
->name
, p
->length
) == 0))
544 /* Return true if and only if name PATH (from an archive) matches any
545 name from the namelist. */
547 name_match (const char *path
)
549 size_t length
= strlen (path
);
553 struct name
*cursor
= namelist
;
556 return ! files_from_option
;
560 chdir_do (cursor
->change_dir
);
562 nametail
= &namelist
;
563 return ! files_from_option
;
566 cursor
= namelist_match (path
, length
);
569 cursor
->found
= 1; /* remember it matched */
570 if (starting_file_option
)
574 nametail
= &namelist
;
576 chdir_do (cursor
->change_dir
);
578 /* We got a match. */
582 /* Filename from archive not found in namelist. If we have the whole
583 namelist here, just return 0. Otherwise, read the next name in and
584 compare it. If this was the last name, namelist->found will remain
585 on. If not, we loop to compare the newly read name. */
587 if (same_order_option
&& namelist
->found
)
589 name_gather (); /* read one more */
598 /* Returns true if all names from the namelist were processed */
602 struct name
const *cursor
;
603 for (cursor
= namelist
; cursor
; cursor
= cursor
->next
)
604 if (!cursor
->found
&& !cursor
->fake
)
609 /* Print the names of things in the namelist that were not matched. */
611 names_notfound (void)
613 struct name
const *cursor
;
615 for (cursor
= namelist
; cursor
; cursor
= cursor
->next
)
616 if (!cursor
->found
&& !cursor
->fake
)
617 ERROR ((0, 0, _("%s: Not found in archive"),
618 quotearg_colon (cursor
->name
)));
620 /* Don't bother freeing the name list; we're about to exit. */
622 nametail
= &namelist
;
624 if (same_order_option
)
628 while (name
= name_next (1), name
)
629 ERROR ((0, 0, _("%s: Not found in archive"),
630 quotearg_colon (name
)));
634 /* Sorting name lists. */
636 /* Sort linked LIST of names, of given LENGTH, using COMPARE to order
637 names. Return the sorted list. Apart from the type `struct name'
638 and the definition of SUCCESSOR, this is a generic list-sorting
639 function, but it's too painful to make it both generic and portable
643 merge_sort (struct name
*list
, int length
,
644 int (*compare
) (struct name
const*, struct name
const*))
646 struct name
*first_list
;
647 struct name
*second_list
;
651 struct name
**merge_point
;
655 # define SUCCESSOR(name) ((name)->next)
662 if ((*compare
) (list
, SUCCESSOR (list
)) > 0)
664 result
= SUCCESSOR (list
);
665 SUCCESSOR (result
) = list
;
666 SUCCESSOR (list
) = 0;
673 first_length
= (length
+ 1) / 2;
674 second_length
= length
/ 2;
675 for (cursor
= list
, counter
= first_length
- 1;
677 cursor
= SUCCESSOR (cursor
), counter
--)
679 second_list
= SUCCESSOR (cursor
);
680 SUCCESSOR (cursor
) = 0;
682 first_list
= merge_sort (first_list
, first_length
, compare
);
683 second_list
= merge_sort (second_list
, second_length
, compare
);
685 merge_point
= &result
;
686 while (first_list
&& second_list
)
687 if ((*compare
) (first_list
, second_list
) < 0)
689 cursor
= SUCCESSOR (first_list
);
690 *merge_point
= first_list
;
691 merge_point
= &SUCCESSOR (first_list
);
696 cursor
= SUCCESSOR (second_list
);
697 *merge_point
= second_list
;
698 merge_point
= &SUCCESSOR (second_list
);
699 second_list
= cursor
;
702 *merge_point
= first_list
;
704 *merge_point
= second_list
;
711 /* A comparison function for sorting names. Put found names last;
712 break ties by string comparison. */
715 compare_names (struct name
const *n1
, struct name
const *n2
)
717 int found_diff
= n2
->found
- n1
->found
;
718 return found_diff
? found_diff
: strcmp (n1
->name
, n2
->name
);
721 /* Add all the dirs under NAME, which names a directory, to the namelist.
722 If any of the files is a directory, recurse on the subdirectory.
723 DEVICE is the device not to leave, if the -l option is specified. */
726 add_hierarchy_to_namelist (struct name
*name
, dev_t device
)
728 char *path
= name
->name
;
729 char *buffer
= get_directory_contents (path
, device
);
732 name
->dir_contents
= "\0\0\0\0";
735 size_t name_length
= name
->length
;
736 size_t allocated_length
= (name_length
>= NAME_FIELD_SIZE
737 ? name_length
+ NAME_FIELD_SIZE
739 char *name_buffer
= xmalloc (allocated_length
+ 1);
740 /* FIXME: + 2 above? */
742 size_t string_length
;
743 int change_dir
= name
->change_dir
;
745 name
->dir_contents
= buffer
;
746 strcpy (name_buffer
, path
);
747 if (! ISSLASH (name_buffer
[name_length
- 1]))
749 name_buffer
[name_length
++] = '/';
750 name_buffer
[name_length
] = '\0';
753 for (string
= buffer
; *string
; string
+= string_length
+ 1)
755 string_length
= strlen (string
);
758 if (allocated_length
<= name_length
+ string_length
)
762 allocated_length
*= 2;
763 if (! allocated_length
)
766 while (allocated_length
<= name_length
+ string_length
);
768 name_buffer
= xrealloc (name_buffer
, allocated_length
+ 1);
770 strcpy (name_buffer
+ name_length
, string
+ 1);
771 add_hierarchy_to_namelist (addname (name_buffer
, change_dir
),
780 /* Collect all the names from argv[] (or whatever), expand them into a
781 directory tree, and sort them. This gets only subdirectories, not
785 collect_and_sort_names (void)
788 struct name
*next_name
;
794 if (listed_incremental_option
)
795 read_directory_file ();
800 for (name
= namelist
; name
; name
= next_name
)
802 next_name
= name
->next
;
803 if (name
->found
|| name
->dir_contents
)
805 if (name
->regexp
) /* FIXME: just skip regexps for now */
807 chdir_do (name
->change_dir
);
811 if (deref_stat (dereference_option
, name
->name
, &statbuf
) != 0)
813 if (ignore_failed_read_option
)
814 stat_warn (name
->name
);
816 stat_error (name
->name
);
819 if (S_ISDIR (statbuf
.st_mode
))
822 add_hierarchy_to_namelist (name
, statbuf
.st_dev
);
827 for (name
= namelist
; name
; name
= name
->next
)
829 namelist
= merge_sort (namelist
, num_names
, compare_names
);
831 for (name
= namelist
; name
; name
= name
->next
)
835 /* This is like name_match, except that it returns a pointer to the
836 name it matched, and doesn't set FOUND in structure. The caller
837 will have to do that if it wants to. Oh, and if the namelist is
838 empty, it returns null, unlike name_match, which returns TRUE. */
840 name_scan (const char *path
)
842 size_t length
= strlen (path
);
846 struct name
*cursor
= namelist_match (path
, length
);
850 /* Filename from archive not found in namelist. If we have the whole
851 namelist here, just return 0. Otherwise, read the next name in and
852 compare it. If this was the last name, namelist->found will remain
853 on. If not, we loop to compare the newly read name. */
855 if (same_order_option
&& namelist
&& namelist
->found
)
857 name_gather (); /* read one more */
866 /* This returns a name from the namelist which doesn't have ->found
867 set. It sets ->found before returning, so successive calls will
868 find and return all the non-found names in the namelist. */
869 struct name
*gnu_list_name
;
872 name_from_list (void)
875 gnu_list_name
= namelist
;
876 while (gnu_list_name
&& (gnu_list_name
->found
| gnu_list_name
->fake
))
877 gnu_list_name
= gnu_list_name
->next
;
880 gnu_list_name
->found
= 1;
881 chdir_do (gnu_list_name
->change_dir
);
882 return gnu_list_name
->name
;
888 blank_name_list (void)
893 for (name
= namelist
; name
; name
= name
->next
)
897 /* Yield a newly allocated file name consisting of PATH concatenated to
898 NAME, with an intervening slash if PATH does not already end in one. */
900 new_name (const char *path
, const char *name
)
902 size_t pathlen
= strlen (path
);
903 size_t namesize
= strlen (name
) + 1;
904 int slash
= pathlen
&& ! ISSLASH (path
[pathlen
- 1]);
905 char *buffer
= xmalloc (pathlen
+ slash
+ namesize
);
906 memcpy (buffer
, path
, pathlen
);
907 buffer
[pathlen
] = '/';
908 memcpy (buffer
+ pathlen
+ slash
, name
, namesize
);
912 /* Return nonzero if file NAME is excluded. */
914 excluded_name (char const *name
)
916 return excluded_filename (excluded
, name
+ FILESYSTEM_PREFIX_LEN (name
));
919 /* Hash tables of strings. */
921 /* Calculate the hash of a string. */
923 hash_string_hasher (void const *name
, unsigned n_buckets
)
925 return hash_string (name
, n_buckets
);
928 /* Compare two strings for equality. */
930 hash_string_compare (void const *name1
, void const *name2
)
932 return strcmp (name1
, name2
) == 0;
935 /* Return zero if TABLE contains a copy of STRING; otherwise, insert a
936 copy of STRING to TABLE and return 1. */
938 hash_string_insert (Hash_table
**table
, char const *string
)
940 Hash_table
*t
= *table
;
941 char *s
= xstrdup (string
);
945 || (*table
= t
= hash_initialize (0, 0, hash_string_hasher
,
946 hash_string_compare
, 0)))
947 && (e
= hash_insert (t
, s
))))
959 /* Return 1 if TABLE contains STRING. */
961 hash_string_lookup (Hash_table
const *table
, char const *string
)
963 return table
&& hash_lookup (table
, string
);
966 /* Names to avoid dumping. */
967 static Hash_table
*avoided_name_table
;
969 /* Remember to not archive NAME. */
971 add_avoided_name (char const *name
)
973 hash_string_insert (&avoided_name_table
, name
);
976 /* Should NAME be avoided when archiving? */
978 is_avoided_name (char const *name
)
980 return hash_string_lookup (avoided_name_table
, name
);
983 /* Return a safer suffix of FILE_NAME, or "." if it has no safer
984 suffix. Check for fully specified file names and other atrocities.
985 Warn the user if we do not return NAME. If LINK_TARGET is 1,
986 FILE_NAME is the target of a hard link, not a member name. */
989 safer_name_suffix (char const *file_name
, bool link_target
)
993 if (absolute_names_option
)
997 /* Skip file system prefixes, leading pathnames that contain
998 "..", and leading slashes. */
1000 size_t prefix_len
= FILESYSTEM_PREFIX_LEN (file_name
);
1002 for (p
= file_name
+ prefix_len
; *p
; )
1004 if (p
[0] == '.' && p
[1] == '.' && (ISSLASH (p
[2]) || !p
[2]))
1005 prefix_len
= p
+ 2 - file_name
;
1016 for (p
= file_name
+ prefix_len
; ISSLASH (*p
); p
++)
1018 prefix_len
= p
- file_name
;
1022 static Hash_table
*prefix_table
[2];
1023 char *prefix
= alloca (prefix_len
+ 1);
1024 memcpy (prefix
, file_name
, prefix_len
);
1025 prefix
[prefix_len
] = '\0';
1027 if (hash_string_insert (&prefix_table
[link_target
], prefix
))
1029 static char const *const diagnostic
[] =
1031 N_("Removing leading `%s' from member names"),
1032 N_("Removing leading `%s' from hard link targets")
1034 WARN ((0, 0, _(diagnostic
[link_target
]), prefix
));
1043 static char const *const diagnostic
[] =
1045 N_("Substituting `.' for empty member name"),
1046 N_("Substituting `.' for empty hard link target")
1048 WARN ((0, 0, _(diagnostic
[link_target
])));
1057 /* Return the size of the prefix of FILE_NAME that is removed after
1058 stripping NUM leading path name components. NUM must be
1062 stripped_prefix_len (char const *file_name
, size_t num
)
1064 char const *p
= file_name
+ FILESYSTEM_PREFIX_LEN (file_name
);
1065 while (ISSLASH (*p
))
1069 bool slash
= ISSLASH (*p
);
1074 return p
- file_name
;
1075 while (ISSLASH (*p
))
1082 /* Return nonzero if NAME contains ".." as a path name component. */
1084 contains_dot_dot (char const *name
)
1086 char const *p
= name
+ FILESYSTEM_PREFIX_LEN (name
);
1090 if (p
[0] == '.' && p
[1] == '.' && (ISSLASH (p
[2]) || !p
[2]))
1098 while (! ISSLASH (*p
));