1 /* Various processing of names.
3 Copyright (C) 1988, 1992, 1994, 1996, 1997, 1998, 1999, 2000, 2001,
4 2003, 2004, 2005, 2006, 2007 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 3, 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 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
28 /* User and group names. */
30 struct group
*getgrnam ();
31 struct passwd
*getpwnam ();
32 #if ! HAVE_DECL_GETPWUID
33 struct passwd
*getpwuid ();
35 #if ! HAVE_DECL_GETGRGID
36 struct group
*getgrgid ();
39 /* Make sure you link with the proper libraries if you are running the
40 Yellow Peril (thanks for the good laugh, Ian J.!), or, euh... NIS.
41 This code should also be modified for non-UNIX systems to do something
44 static char *cached_uname
;
45 static char *cached_gname
;
47 static uid_t cached_uid
; /* valid only if cached_uname is not empty */
48 static gid_t cached_gid
; /* valid only if cached_gname is not empty */
50 /* These variables are valid only if nonempty. */
51 static char *cached_no_such_uname
;
52 static char *cached_no_such_gname
;
54 /* These variables are valid only if nonzero. It's not worth optimizing
55 the case for weird systems where 0 is not a valid uid or gid. */
56 static uid_t cached_no_such_uid
;
57 static gid_t cached_no_such_gid
;
59 static void register_individual_file (char const *name
);
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
)
69 *uname
= xstrdup ("");
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
;
84 *uname
= xstrdup ("");
88 *uname
= xstrdup (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
)
99 *gname
= xstrdup ("");
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
= xstrdup ("");
118 *gname
= xstrdup (cached_gname
);
121 /* Given UNAME, set the corresponding UID and return 1, or else, return 0. */
123 uname_to_uid (char const *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 const *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 */
187 /* File name arguments are processed in two stages: first a
188 name_array (see below) is filled, then the names from it
189 are moved into the namelist.
191 This awkward process is needed only to implement --same-order option,
192 which is meant to help process large archives on machines with
193 limited memory. With this option on, namelist contains at most one
194 entry, which diminishes the memory consumption.
196 However, I very much doubt if we still need this -- Sergey */
198 /* A name_array element contains entries of three types: */
200 #define NELT_NAME 0 /* File name */
201 #define NELT_CHDIR 1 /* Change directory request */
202 #define NELT_FMASK 2 /* Change fnmatch options request */
204 struct name_elt
/* A name_array element. */
206 char type
; /* Element type, see NELT_* constants above */
209 const char *name
; /* File or directory name */
210 int matching_flags
;/* fnmatch options if type == NELT_FMASK */
214 static struct name_elt
*name_array
; /* store an array of names */
215 static size_t allocated_names
; /* how big is the array? */
216 static size_t names
; /* how many entries does it have? */
217 static size_t name_index
; /* how many of the entries have we scanned? */
219 /* Check the size of name_array, reallocating it as necessary. */
223 if (names
== allocated_names
)
225 if (allocated_names
== 0)
226 allocated_names
= 10; /* Set initial allocation */
227 name_array
= x2nrealloc (name_array
, &allocated_names
,
228 sizeof (name_array
[0]));
232 /* Add to name_array the file NAME with fnmatch options MATCHING_FLAGS */
234 name_add_name (const char *name
, int matching_flags
)
236 static int prev_flags
= 0; /* FIXME: Or EXCLUDE_ANCHORED? */
240 ep
= &name_array
[names
++];
241 if (prev_flags
!= matching_flags
)
243 ep
->type
= NELT_FMASK
;
244 ep
->v
.matching_flags
= matching_flags
;
245 prev_flags
= matching_flags
;
247 ep
= &name_array
[names
++];
249 ep
->type
= NELT_NAME
;
253 /* Add to name_array a chdir request for the directory NAME */
255 name_add_dir (const char *name
)
259 ep
= &name_array
[names
++];
260 ep
->type
= NELT_CHDIR
;
265 /* Names from external name file. */
267 static char *name_buffer
; /* buffer to hold the current file name */
268 static size_t name_buffer_length
; /* allocated length of name_buffer */
270 /* Set up to gather file names for tar. They can either come from a
271 file or were saved from decoding arguments. */
275 name_buffer
= xmalloc (NAME_FIELD_SIZE
+ 2);
276 name_buffer_length
= NAME_FIELD_SIZE
;
286 static int matching_flags
; /* exclude_fnmatch options */
288 /* Get the next NELT_NAME element from name_array. Result is in
289 static storage and can't be relied upon across two calls.
291 If CHANGE_DIRS is true, treat any entries of type NELT_CHDIR as
292 the request to change to the given directory.
294 Entries of type NELT_FMASK cause updates of the matching_flags
297 name_next_elt (int change_dirs
)
299 static struct name_elt entry
;
303 while (name_index
!= names
)
308 ep
= &name_array
[name_index
++];
309 if (ep
->type
== NELT_FMASK
)
311 matching_flags
= ep
->v
.matching_flags
;
316 source_len
= strlen (source
);
317 if (name_buffer_length
< source_len
)
321 name_buffer_length
*= 2;
322 if (! name_buffer_length
)
325 while (name_buffer_length
< source_len
);
328 name_buffer
= xmalloc (name_buffer_length
+ 2);
330 strcpy (name_buffer
, source
);
332 /* Zap trailing slashes. */
334 cursor
= name_buffer
+ strlen (name_buffer
) - 1;
335 while (cursor
> name_buffer
&& ISSLASH (*cursor
))
338 if (change_dirs
&& ep
->type
== NELT_CHDIR
)
340 if (chdir (name_buffer
) < 0)
341 chdir_fatal (name_buffer
);
346 unquote_string (name_buffer
);
347 if (incremental_option
)
348 register_individual_file (name_buffer
);
349 entry
.type
= ep
->type
;
350 entry
.v
.name
= name_buffer
;
359 name_next (int change_dirs
)
361 struct name_elt
*nelt
= name_next_elt (change_dirs
);
362 return nelt
? nelt
->v
.name
: NULL
;
365 /* Gather names in a list for scanning. Could hash them later if we
368 If the names are already sorted to match the archive, we just read
369 them one by one. name_gather reads the first one, and it is called
370 by name_match as appropriate to read the next ones. At EOF, the
371 last name read is just left in the buffer. This option lets users
372 of small machines extract an arbitrary number of files by doing
373 "tar t" and editing down the list of files. */
378 /* Buffer able to hold a single name. */
379 static struct name
*buffer
;
380 static size_t allocated_size
;
384 if (same_order_option
)
386 static int change_dir
;
388 if (allocated_size
== 0)
390 allocated_size
= offsetof (struct name
, name
) + NAME_FIELD_SIZE
+ 1;
391 buffer
= xzalloc (allocated_size
);
394 while ((ep
= name_next_elt (0)) && ep
->type
== NELT_CHDIR
)
395 change_dir
= chdir_arg (xstrdup (ep
->v
.name
));
401 buffer
->length
= strlen (ep
->v
.name
);
402 needed_size
= offsetof (struct name
, name
) + buffer
->length
+ 1;
403 if (allocated_size
< needed_size
)
408 if (! allocated_size
)
411 while (allocated_size
< needed_size
);
413 buffer
= xrealloc (buffer
, allocated_size
);
415 buffer
->change_dir
= change_dir
;
416 strcpy (buffer
->name
, ep
->v
.name
);
418 buffer
->found_count
= 0;
419 buffer
->matching_flags
= matching_flags
;
422 nametail
= &namelist
->next
;
425 addname (0, change_dir
);
429 /* Non sorted names -- read them all in. */
434 int change_dir0
= change_dir
;
435 while ((ep
= name_next_elt (0)) && ep
->type
== NELT_CHDIR
)
436 change_dir
= chdir_arg (xstrdup (ep
->v
.name
));
439 addname (ep
->v
.name
, change_dir
);
442 if (change_dir
!= change_dir0
)
443 addname (0, change_dir
);
450 /* Add a name to the namelist. */
452 addname (char const *string
, int change_dir
)
454 size_t length
= string
? strlen (string
) : 0;
455 struct name
*name
= xmalloc (offsetof (struct name
, name
) + length
+ 1);
458 strcpy (name
->name
, string
);
463 name
->length
= length
;
464 name
->found_count
= 0;
465 name
->matching_flags
= matching_flags
;
466 name
->change_dir
= change_dir
;
467 name
->dir_contents
= NULL
;
470 nametail
= &name
->next
;
474 /* Find a match for FILE_NAME (whose string length is LENGTH) in the name
477 namelist_match (char const *file_name
, size_t length
)
481 for (p
= namelist
; p
; p
= p
->next
)
484 && exclude_fnmatch (p
->name
, file_name
, p
->matching_flags
))
491 /* Return true if and only if name FILE_NAME (from an archive) matches any
492 name from the namelist. */
494 name_match (const char *file_name
)
496 size_t length
= strlen (file_name
);
500 struct name
*cursor
= namelist
;
505 if (cursor
->name
[0] == 0)
507 chdir_do (cursor
->change_dir
);
509 nametail
= &namelist
;
513 cursor
= namelist_match (file_name
, length
);
516 if (!(ISSLASH (file_name
[cursor
->length
]) && recursion_option
)
517 || cursor
->found_count
== 0)
518 cursor
->found_count
++; /* remember it matched */
519 if (starting_file_option
)
523 nametail
= &namelist
;
525 chdir_do (cursor
->change_dir
);
527 /* We got a match. */
528 return ISFOUND (cursor
);
531 /* Filename from archive not found in namelist. If we have the whole
532 namelist here, just return 0. Otherwise, read the next name in and
533 compare it. If this was the last name, namelist->found_count will
534 remain on. If not, we loop to compare the newly read name. */
536 if (same_order_option
&& namelist
->found_count
)
538 name_gather (); /* read one more */
539 if (namelist
->found_count
)
547 /* Returns true if all names from the namelist were processed.
548 P is the stat_info of the most recently processed entry.
549 The decision is postponed until the next entry is read if:
551 1) P ended with a slash (i.e. it was a directory)
552 2) P matches any entry from the namelist *and* represents a subdirectory
553 or a file lying under this entry (in the terms of directory structure).
555 This is necessary to handle contents of directories. */
557 all_names_found (struct tar_stat_info
*p
)
559 struct name
const *cursor
;
562 if (test_label_option
)
564 if (!p
->file_name
|| occurrence_option
== 0 || p
->had_trailing_slash
)
566 len
= strlen (p
->file_name
);
567 for (cursor
= namelist
; cursor
; cursor
= cursor
->next
)
569 if ((cursor
->name
[0] && !WASFOUND (cursor
))
570 || (len
>= cursor
->length
&& ISSLASH (p
->file_name
[cursor
->length
])))
577 is_pattern (const char *string
)
579 return strchr (string
, '*') || strchr (string
, '[') || strchr (string
, '?');
583 regex_usage_warning (const char *name
)
585 static int warned_once
= 0;
587 if (warn_regex_usage
&& is_pattern (name
))
591 /* TRANSLATORS: The following three msgids form a single sentence.
593 _("Pattern matching characters used in file names. Please,")));
595 _("use --wildcards to enable pattern matching, or --no-wildcards to")));
597 _("suppress this warning.")));
601 /* Print the names of things in the namelist that were not matched. */
603 names_notfound (void)
605 struct name
const *cursor
;
607 for (cursor
= namelist
; cursor
; cursor
= cursor
->next
)
608 if (!WASFOUND (cursor
) && cursor
->name
[0])
610 regex_usage_warning (cursor
->name
);
611 if (cursor
->found_count
== 0)
612 ERROR ((0, 0, _("%s: Not found in archive"),
613 quotearg_colon (cursor
->name
)));
615 ERROR ((0, 0, _("%s: Required occurrence not found in archive"),
616 quotearg_colon (cursor
->name
)));
619 /* Don't bother freeing the name list; we're about to exit. */
621 nametail
= &namelist
;
623 if (same_order_option
)
627 while ((name
= name_next (1)) != NULL
)
629 regex_usage_warning (name
);
630 ERROR ((0, 0, _("%s: Not found in archive"),
631 quotearg_colon (name
)));
636 /* Sorting name lists. */
638 /* Sort linked LIST of names, of given LENGTH, using COMPARE to order
639 names. Return the sorted list. Apart from the type `struct name'
640 and the definition of SUCCESSOR, this is a generic list-sorting
641 function, but it's too painful to make it both generic and portable
645 merge_sort (struct name
*list
, int length
,
646 int (*compare
) (struct name
const*, struct name
const*))
648 struct name
*first_list
;
649 struct name
*second_list
;
653 struct name
**merge_point
;
657 # define SUCCESSOR(name) ((name)->next)
664 if ((*compare
) (list
, SUCCESSOR (list
)) > 0)
666 result
= SUCCESSOR (list
);
667 SUCCESSOR (result
) = list
;
668 SUCCESSOR (list
) = 0;
675 first_length
= (length
+ 1) / 2;
676 second_length
= length
/ 2;
677 for (cursor
= list
, counter
= first_length
- 1;
679 cursor
= SUCCESSOR (cursor
), counter
--)
681 second_list
= SUCCESSOR (cursor
);
682 SUCCESSOR (cursor
) = 0;
684 first_list
= merge_sort (first_list
, first_length
, compare
);
685 second_list
= merge_sort (second_list
, second_length
, compare
);
687 merge_point
= &result
;
688 while (first_list
&& second_list
)
689 if ((*compare
) (first_list
, second_list
) < 0)
691 cursor
= SUCCESSOR (first_list
);
692 *merge_point
= first_list
;
693 merge_point
= &SUCCESSOR (first_list
);
698 cursor
= SUCCESSOR (second_list
);
699 *merge_point
= second_list
;
700 merge_point
= &SUCCESSOR (second_list
);
701 second_list
= cursor
;
704 *merge_point
= first_list
;
706 *merge_point
= second_list
;
713 /* A comparison function for sorting names. Put found names last;
714 break ties by string comparison. */
717 compare_names (struct name
const *n1
, struct name
const *n2
)
719 int found_diff
= WASFOUND(n2
) - WASFOUND(n1
);
720 return found_diff
? found_diff
: strcmp (n1
->name
, n2
->name
);
723 /* Add all the dirs under NAME, which names a directory, to the namelist.
724 If any of the files is a directory, recurse on the subdirectory.
725 DEVICE is the device not to leave, if the -l option is specified. */
728 add_hierarchy_to_namelist (struct name
*name
, dev_t device
)
730 char *file_name
= name
->name
;
731 const char *buffer
= get_directory_contents (file_name
, device
);
734 name
->dir_contents
= "\0\0\0\0";
737 size_t name_length
= name
->length
;
738 size_t allocated_length
= (name_length
>= NAME_FIELD_SIZE
739 ? name_length
+ NAME_FIELD_SIZE
741 char *namebuf
= xmalloc (allocated_length
+ 1);
742 /* FIXME: + 2 above? */
744 size_t string_length
;
745 int change_dir
= name
->change_dir
;
747 name
->dir_contents
= buffer
;
748 strcpy (namebuf
, file_name
);
749 if (! ISSLASH (namebuf
[name_length
- 1]))
751 namebuf
[name_length
++] = '/';
752 namebuf
[name_length
] = '\0';
755 for (string
= buffer
; *string
; string
+= string_length
+ 1)
757 string_length
= strlen (string
);
762 if (allocated_length
<= name_length
+ string_length
)
766 allocated_length
*= 2;
767 if (! allocated_length
)
770 while (allocated_length
<= name_length
+ string_length
);
772 namebuf
= xrealloc (namebuf
, allocated_length
+ 1);
774 strcpy (namebuf
+ name_length
, string
+ 1);
775 np
= addname (namebuf
, change_dir
);
776 add_hierarchy_to_namelist (np
, device
);
784 /* Collect all the names from argv[] (or whatever), expand them into a
785 directory tree, and sort them. This gets only subdirectories, not
789 collect_and_sort_names (void)
792 struct name
*next_name
;
798 if (listed_incremental_option
)
799 read_directory_file ();
804 for (name
= namelist
; name
; name
= next_name
)
806 next_name
= name
->next
;
807 if (name
->found_count
|| name
->dir_contents
)
809 if (name
->matching_flags
& EXCLUDE_WILDCARDS
)
810 /* NOTE: EXCLUDE_ANCHORED is not relevant here */
811 /* FIXME: just skip regexps for now */
813 chdir_do (name
->change_dir
);
814 if (name
->name
[0] == 0)
817 if (deref_stat (dereference_option
, name
->name
, &statbuf
) != 0)
819 stat_diag (name
->name
);
822 if (S_ISDIR (statbuf
.st_mode
))
825 add_hierarchy_to_namelist (name
, statbuf
.st_dev
);
830 for (name
= namelist
; name
; name
= name
->next
)
832 namelist
= merge_sort (namelist
, num_names
, compare_names
);
834 for (name
= namelist
; name
; name
= name
->next
)
835 name
->found_count
= 0;
837 if (listed_incremental_option
)
839 for (name
= namelist
; name
&& name
->name
[0] == 0; name
++)
842 name
->dir_contents
= append_incremental_renames (name
->dir_contents
);
846 /* This is like name_match, except that
847 1. It returns a pointer to the name it matched, and doesn't set FOUND
848 in structure. The caller will have to do that if it wants to.
849 2. If the namelist is empty, it returns null, unlike name_match, which
852 name_scan (const char *file_name
)
854 size_t length
= strlen (file_name
);
858 struct name
*cursor
= namelist_match (file_name
, length
);
862 /* Filename from archive not found in namelist. If we have the whole
863 namelist here, just return 0. Otherwise, read the next name in and
864 compare it. If this was the last name, namelist->found_count will
865 remain on. If not, we loop to compare the newly read name. */
867 if (same_order_option
&& namelist
&& namelist
->found_count
)
869 name_gather (); /* read one more */
870 if (namelist
->found_count
)
878 /* This returns a name from the namelist which doesn't have ->found
879 set. It sets ->found before returning, so successive calls will
880 find and return all the non-found names in the namelist. */
881 struct name
*gnu_list_name
;
884 name_from_list (void)
887 gnu_list_name
= namelist
;
889 && (gnu_list_name
->found_count
|| gnu_list_name
->name
[0] == 0))
890 gnu_list_name
= gnu_list_name
->next
;
893 gnu_list_name
->found_count
++;
894 chdir_do (gnu_list_name
->change_dir
);
895 return gnu_list_name
->name
;
901 blank_name_list (void)
906 for (name
= namelist
; name
; name
= name
->next
)
907 name
->found_count
= 0;
910 /* Yield a newly allocated file name consisting of FILE_NAME concatenated to
911 NAME, with an intervening slash if FILE_NAME does not already end in one. */
913 new_name (const char *file_name
, const char *name
)
915 size_t file_name_len
= strlen (file_name
);
916 size_t namesize
= strlen (name
) + 1;
917 int slash
= file_name_len
&& ! ISSLASH (file_name
[file_name_len
- 1]);
918 char *buffer
= xmalloc (file_name_len
+ slash
+ namesize
);
919 memcpy (buffer
, file_name
, file_name_len
);
920 buffer
[file_name_len
] = '/';
921 memcpy (buffer
+ file_name_len
+ slash
, name
, namesize
);
925 /* Return nonzero if file NAME is excluded. */
927 excluded_name (char const *name
)
929 return excluded_file_name (excluded
, name
+ FILE_SYSTEM_PREFIX_LEN (name
));
932 /* Names to avoid dumping. */
933 static Hash_table
*avoided_name_table
;
935 /* Remember to not archive NAME. */
937 add_avoided_name (char const *name
)
939 hash_string_insert (&avoided_name_table
, name
);
942 /* Should NAME be avoided when archiving? */
944 is_avoided_name (char const *name
)
946 return hash_string_lookup (avoided_name_table
, name
);
950 static Hash_table
*individual_file_table
;
953 register_individual_file (char const *name
)
957 if (deref_stat (dereference_option
, name
, &st
) != 0)
958 return; /* Will be complained about later */
959 if (S_ISDIR (st
.st_mode
))
962 hash_string_insert (&individual_file_table
, name
);
966 is_individual_file (char const *name
)
968 return hash_string_lookup (individual_file_table
, name
);
973 /* Return the size of the prefix of FILE_NAME that is removed after
974 stripping NUM leading file name components. NUM must be
978 stripped_prefix_len (char const *file_name
, size_t num
)
980 char const *p
= file_name
+ FILE_SYSTEM_PREFIX_LEN (file_name
);
985 bool slash
= ISSLASH (*p
);
990 return p
- file_name
;
998 /* Return nonzero if NAME contains ".." as a file name component. */
1000 contains_dot_dot (char const *name
)
1002 char const *p
= name
+ FILE_SYSTEM_PREFIX_LEN (name
);
1006 if (p
[0] == '.' && p
[1] == '.' && (ISSLASH (p
[2]) || !p
[2]))
1009 while (! ISSLASH (*p
))