1 /* Various processing of names.
3 Copyright (C) 1988, 1992, 1994, 1996, 1997, 1998, 1999, 2000, 2001,
4 2003, 2004, 2005, 2006 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 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. If filename_terminator
293 is NUL, CHANGE_DIRS is effectively always false.
295 Entries of type NELT_FMASK cause updates of the matching_flags
298 name_next_elt (int change_dirs
)
300 static struct name_elt entry
;
304 if (filename_terminator
== '\0')
307 while (name_index
!= names
)
312 ep
= &name_array
[name_index
++];
313 if (ep
->type
== NELT_FMASK
)
315 matching_flags
= ep
->v
.matching_flags
;
320 source_len
= strlen (source
);
321 if (name_buffer_length
< source_len
)
325 name_buffer_length
*= 2;
326 if (! name_buffer_length
)
329 while (name_buffer_length
< source_len
);
332 name_buffer
= xmalloc (name_buffer_length
+ 2);
334 strcpy (name_buffer
, source
);
336 /* Zap trailing slashes. */
338 cursor
= name_buffer
+ strlen (name_buffer
) - 1;
339 while (cursor
> name_buffer
&& ISSLASH (*cursor
))
342 if (change_dirs
&& ep
->type
== NELT_CHDIR
)
344 if (chdir (name_buffer
) < 0)
345 chdir_fatal (name_buffer
);
350 unquote_string (name_buffer
);
351 if (incremental_option
)
352 register_individual_file (name_buffer
);
353 entry
.type
= ep
->type
;
354 entry
.v
.name
= name_buffer
;
363 name_next (int change_dirs
)
365 struct name_elt
*nelt
= name_next_elt (change_dirs
);
366 return nelt
? nelt
->v
.name
: NULL
;
369 /* Gather names in a list for scanning. Could hash them later if we
372 If the names are already sorted to match the archive, we just read
373 them one by one. name_gather reads the first one, and it is called
374 by name_match as appropriate to read the next ones. At EOF, the
375 last name read is just left in the buffer. This option lets users
376 of small machines extract an arbitrary number of files by doing
377 "tar t" and editing down the list of files. */
382 /* Buffer able to hold a single name. */
383 static struct name
*buffer
;
384 static size_t allocated_size
;
388 if (same_order_option
)
390 static int change_dir
;
392 if (allocated_size
== 0)
394 allocated_size
= offsetof (struct name
, name
) + NAME_FIELD_SIZE
+ 1;
395 buffer
= xmalloc (allocated_size
);
396 /* FIXME: This memset is overkill, and ugly... */
397 memset (buffer
, 0, allocated_size
);
400 while ((ep
= name_next_elt (0)) && ep
->type
== NELT_CHDIR
)
401 change_dir
= chdir_arg (xstrdup (ep
->v
.name
));
407 buffer
->length
= strlen (ep
->v
.name
);
408 needed_size
= offsetof (struct name
, name
) + buffer
->length
+ 1;
409 if (allocated_size
< needed_size
)
414 if (! allocated_size
)
417 while (allocated_size
< needed_size
);
419 buffer
= xrealloc (buffer
, allocated_size
);
421 buffer
->change_dir
= change_dir
;
422 strcpy (buffer
->name
, ep
->v
.name
);
424 buffer
->found_count
= 0;
425 buffer
->matching_flags
= matching_flags
;
428 nametail
= &namelist
->next
;
431 addname (0, change_dir
);
435 /* Non sorted names -- read them all in. */
440 int change_dir0
= change_dir
;
441 while ((ep
= name_next_elt (0)) && ep
->type
== NELT_CHDIR
)
442 change_dir
= chdir_arg (xstrdup (ep
->v
.name
));
445 addname (ep
->v
.name
, change_dir
);
448 if (change_dir
!= change_dir0
)
449 addname (0, change_dir
);
456 /* Add a name to the namelist. */
458 addname (char const *string
, int change_dir
)
460 size_t length
= string
? strlen (string
) : 0;
461 struct name
*name
= xmalloc (offsetof (struct name
, name
) + length
+ 1);
464 strcpy (name
->name
, string
);
469 name
->length
= length
;
470 name
->found_count
= 0;
471 name
->matching_flags
= matching_flags
;
472 name
->change_dir
= change_dir
;
473 name
->dir_contents
= NULL
;
477 nametail
= &name
->next
;
481 /* Find a match for FILE_NAME (whose string length is LENGTH) in the name
484 namelist_match (char const *file_name
, size_t length
, bool exact
)
488 for (p
= namelist
; p
; p
= p
->next
)
491 && exclude_fnmatch (p
->name
, file_name
, p
->matching_flags
))
498 /* Return true if and only if name FILE_NAME (from an archive) matches any
499 name from the namelist. */
501 name_match (const char *file_name
)
503 size_t length
= strlen (file_name
);
507 struct name
*cursor
= namelist
;
512 if (cursor
->name
[0] == 0)
514 chdir_do (cursor
->change_dir
);
516 nametail
= &namelist
;
520 cursor
= namelist_match (file_name
, length
, false);
523 if (!(ISSLASH (file_name
[cursor
->length
]) && recursion_option
)
524 || cursor
->found_count
== 0)
525 cursor
->found_count
++; /* remember it matched */
526 if (starting_file_option
)
530 nametail
= &namelist
;
532 chdir_do (cursor
->change_dir
);
534 /* We got a match. */
535 return ISFOUND (cursor
);
538 /* Filename from archive not found in namelist. If we have the whole
539 namelist here, just return 0. Otherwise, read the next name in and
540 compare it. If this was the last name, namelist->found_count will
541 remain on. If not, we loop to compare the newly read name. */
543 if (same_order_option
&& namelist
->found_count
)
545 name_gather (); /* read one more */
546 if (namelist
->found_count
)
554 /* Returns true if all names from the namelist were processed.
555 P is the stat_info of the most recently processed entry.
556 The decision is postponed until the next entry is read if:
558 1) P ended with a slash (i.e. it was a directory)
559 2) P matches any entry from the namelist *and* represents a subdirectory
560 or a file lying under this entry (in the terms of directory structure).
562 This is necessary to handle contents of directories. */
564 all_names_found (struct tar_stat_info
*p
)
566 struct name
const *cursor
;
569 if (test_label_option
)
571 if (!p
->file_name
|| occurrence_option
== 0 || p
->had_trailing_slash
)
573 len
= strlen (p
->file_name
);
574 for (cursor
= namelist
; cursor
; cursor
= cursor
->next
)
576 if (cursor
->matching_flags
/* FIXME: check this */
577 || (!WASFOUND (cursor
) && cursor
->name
[0])
578 || (len
>= cursor
->length
&& ISSLASH (p
->file_name
[cursor
->length
])))
585 is_pattern (const char *string
)
587 return strchr (string
, '*') || strchr (string
, '[') || strchr (string
, '?');
591 regex_usage_warning (const char *name
)
593 static int warned_once
= 0;
595 if (warn_regex_usage
&& is_pattern (name
))
599 /* TRANSLATORS: The following three msgids form a single sentence.
601 _("Pattern matching characters used in file names. Please,")));
603 _("use --wildcards to enable pattern matching, or --no-wildcards to")));
605 _("suppress this warning.")));
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 (!WASFOUND (cursor
) && cursor
->name
[0])
618 regex_usage_warning (cursor
->name
);
619 if (cursor
->found_count
== 0)
620 ERROR ((0, 0, _("%s: Not found in archive"),
621 quotearg_colon (cursor
->name
)));
623 ERROR ((0, 0, _("%s: Required occurrence not found in archive"),
624 quotearg_colon (cursor
->name
)));
627 /* Don't bother freeing the name list; we're about to exit. */
629 nametail
= &namelist
;
631 if (same_order_option
)
635 while ((name
= name_next (1)) != NULL
)
637 regex_usage_warning (name
);
638 ERROR ((0, 0, _("%s: Not found in archive"),
639 quotearg_colon (name
)));
644 /* Sorting name lists. */
646 /* Sort linked LIST of names, of given LENGTH, using COMPARE to order
647 names. Return the sorted list. Apart from the type `struct name'
648 and the definition of SUCCESSOR, this is a generic list-sorting
649 function, but it's too painful to make it both generic and portable
653 merge_sort (struct name
*list
, int length
,
654 int (*compare
) (struct name
const*, struct name
const*))
656 struct name
*first_list
;
657 struct name
*second_list
;
661 struct name
**merge_point
;
665 # define SUCCESSOR(name) ((name)->next)
672 if ((*compare
) (list
, SUCCESSOR (list
)) > 0)
674 result
= SUCCESSOR (list
);
675 SUCCESSOR (result
) = list
;
676 SUCCESSOR (list
) = 0;
683 first_length
= (length
+ 1) / 2;
684 second_length
= length
/ 2;
685 for (cursor
= list
, counter
= first_length
- 1;
687 cursor
= SUCCESSOR (cursor
), counter
--)
689 second_list
= SUCCESSOR (cursor
);
690 SUCCESSOR (cursor
) = 0;
692 first_list
= merge_sort (first_list
, first_length
, compare
);
693 second_list
= merge_sort (second_list
, second_length
, compare
);
695 merge_point
= &result
;
696 while (first_list
&& second_list
)
697 if ((*compare
) (first_list
, second_list
) < 0)
699 cursor
= SUCCESSOR (first_list
);
700 *merge_point
= first_list
;
701 merge_point
= &SUCCESSOR (first_list
);
706 cursor
= SUCCESSOR (second_list
);
707 *merge_point
= second_list
;
708 merge_point
= &SUCCESSOR (second_list
);
709 second_list
= cursor
;
712 *merge_point
= first_list
;
714 *merge_point
= second_list
;
721 /* A comparison function for sorting names. Put found names last;
722 break ties by string comparison. */
725 compare_names (struct name
const *n1
, struct name
const *n2
)
727 int found_diff
= WASFOUND(n2
) - WASFOUND(n1
);
728 return found_diff
? found_diff
: strcmp (n1
->name
, n2
->name
);
731 /* Add all the dirs under NAME, which names a directory, to the namelist.
732 If any of the files is a directory, recurse on the subdirectory.
733 DEVICE is the device not to leave, if the -l option is specified. */
736 add_hierarchy_to_namelist (struct name
*name
, dev_t device
)
738 char *file_name
= name
->name
;
739 char *buffer
= get_directory_contents (file_name
, device
);
742 name
->dir_contents
= "\0\0\0\0";
745 size_t name_length
= name
->length
;
746 size_t allocated_length
= (name_length
>= NAME_FIELD_SIZE
747 ? name_length
+ NAME_FIELD_SIZE
749 char *namebuf
= xmalloc (allocated_length
+ 1);
750 /* FIXME: + 2 above? */
752 size_t string_length
;
753 int change_dir
= name
->change_dir
;
755 name
->dir_contents
= buffer
;
756 strcpy (namebuf
, file_name
);
757 if (! ISSLASH (namebuf
[name_length
- 1]))
759 namebuf
[name_length
++] = '/';
760 namebuf
[name_length
] = '\0';
763 for (string
= buffer
; *string
; string
+= string_length
+ 1)
765 string_length
= strlen (string
);
770 if (allocated_length
<= name_length
+ string_length
)
774 allocated_length
*= 2;
775 if (! allocated_length
)
778 while (allocated_length
<= name_length
+ string_length
);
780 namebuf
= xrealloc (namebuf
, allocated_length
+ 1);
782 strcpy (namebuf
+ name_length
, string
+ 1);
783 np
= addname (namebuf
, change_dir
);
785 add_hierarchy_to_namelist (np
, device
);
793 /* Collect all the names from argv[] (or whatever), expand them into a
794 directory tree, and sort them. This gets only subdirectories, not
798 collect_and_sort_names (void)
801 struct name
*next_name
;
807 if (listed_incremental_option
)
808 read_directory_file ();
813 for (name
= namelist
; name
; name
= next_name
)
815 next_name
= name
->next
;
816 if (name
->found_count
|| name
->dir_contents
)
818 if (name
->matching_flags
& EXCLUDE_WILDCARDS
)
819 /* NOTE: EXCLUDE_ANCHORED is not relevant here */
820 /* FIXME: just skip regexps for now */
822 chdir_do (name
->change_dir
);
823 if (name
->name
[0] == 0)
826 if (deref_stat (dereference_option
, name
->name
, &statbuf
) != 0)
828 stat_diag (name
->name
);
831 if (S_ISDIR (statbuf
.st_mode
))
834 add_hierarchy_to_namelist (name
, statbuf
.st_dev
);
839 for (name
= namelist
; name
; name
= name
->next
)
841 namelist
= merge_sort (namelist
, num_names
, compare_names
);
843 for (name
= namelist
; name
; name
= name
->next
)
844 name
->found_count
= 0;
846 if (listed_incremental_option
)
848 for (name
= namelist
; name
&& name
->name
[0] == 0; name
++)
851 name
->dir_contents
= append_incremental_renames (name
->dir_contents
);
855 /* This is like name_match, except that
856 1. It returns a pointer to the name it matched, and doesn't set FOUND
857 in structure. The caller will have to do that if it wants to.
858 2. If the namelist is empty, it returns null, unlike name_match, which
860 3. The second argument (EXACT) controls matching algorithm. If it
861 is TRUE, the exact matching is used. However, regular expressions are
862 always matched as such, no matter what the value of EXACT is. */
864 name_scan (const char *file_name
, bool exact
)
866 size_t length
= strlen (file_name
);
870 struct name
*cursor
= namelist_match (file_name
, length
, exact
);
874 /* Filename from archive not found in namelist. If we have the whole
875 namelist here, just return 0. Otherwise, read the next name in and
876 compare it. If this was the last name, namelist->found_count will
877 remain on. If not, we loop to compare the newly read name. */
879 if (same_order_option
&& namelist
&& namelist
->found_count
)
881 name_gather (); /* read one more */
882 if (namelist
->found_count
)
890 /* This returns a name from the namelist which doesn't have ->found
891 set. It sets ->found before returning, so successive calls will
892 find and return all the non-found names in the namelist. */
893 struct name
*gnu_list_name
;
896 name_from_list (void)
899 gnu_list_name
= namelist
;
901 && (gnu_list_name
->found_count
|| gnu_list_name
->name
[0] == 0))
902 gnu_list_name
= gnu_list_name
->next
;
905 gnu_list_name
->found_count
++;
906 chdir_do (gnu_list_name
->change_dir
);
907 return gnu_list_name
->name
;
913 blank_name_list (void)
918 for (name
= namelist
; name
; name
= name
->next
)
919 name
->found_count
= 0;
922 /* Yield a newly allocated file name consisting of FILE_NAME concatenated to
923 NAME, with an intervening slash if FILE_NAME does not already end in one. */
925 new_name (const char *file_name
, const char *name
)
927 size_t file_name_len
= strlen (file_name
);
928 size_t namesize
= strlen (name
) + 1;
929 int slash
= file_name_len
&& ! ISSLASH (file_name
[file_name_len
- 1]);
930 char *buffer
= xmalloc (file_name_len
+ slash
+ namesize
);
931 memcpy (buffer
, file_name
, file_name_len
);
932 buffer
[file_name_len
] = '/';
933 memcpy (buffer
+ file_name_len
+ slash
, name
, namesize
);
937 /* Return nonzero if file NAME is excluded. */
939 excluded_name (char const *name
)
941 return excluded_file_name (excluded
, name
+ FILE_SYSTEM_PREFIX_LEN (name
));
944 /* Names to avoid dumping. */
945 static Hash_table
*avoided_name_table
;
947 /* Remember to not archive NAME. */
949 add_avoided_name (char const *name
)
951 hash_string_insert (&avoided_name_table
, name
);
954 /* Should NAME be avoided when archiving? */
956 is_avoided_name (char const *name
)
958 return hash_string_lookup (avoided_name_table
, name
);
962 static Hash_table
*individual_file_table
;
965 register_individual_file (char const *name
)
969 if (deref_stat (dereference_option
, name
, &st
) != 0)
970 return; /* Will be complained about later */
971 if (S_ISDIR (st
.st_mode
))
974 hash_string_insert (&individual_file_table
, name
);
978 is_individual_file (char const *name
)
980 return hash_string_lookup (individual_file_table
, name
);
985 /* Return the size of the prefix of FILE_NAME that is removed after
986 stripping NUM leading file name components. NUM must be
990 stripped_prefix_len (char const *file_name
, size_t num
)
992 char const *p
= file_name
+ FILE_SYSTEM_PREFIX_LEN (file_name
);
997 bool slash
= ISSLASH (*p
);
1002 return p
- file_name
;
1003 while (ISSLASH (*p
))
1010 /* Return nonzero if NAME contains ".." as a file name component. */
1012 contains_dot_dot (char const *name
)
1014 char const *p
= name
+ FILE_SYSTEM_PREFIX_LEN (name
);
1018 if (p
[0] == '.' && p
[1] == '.' && (ISSLASH (p
[2]) || !p
[2]))
1026 while (! ISSLASH (*p
));