1 /* vi: set sw=4 ts=4: */
3 * Mini tar implementation for busybox
5 * Modifed to use common extraction code used by ar, cpio, dpkg-deb, dpkg
6 * Glenn McGrath <bug1@optushome.com.au>
8 * Note, that as of BusyBox-0.43, tar has been completely rewritten from the
9 * ground up. It still has remnents of the old code lying about, but it is
10 * very different now (i.e., cleaner, less global variables, etc.)
12 * Copyright (C) 1999,2000 by Lineo, inc. and Erik Andersen
13 * Copyright (C) 1999,2000,2001 by Erik Andersen <andersee@debian.org>
15 * Based in part in the tar implementation in sash
16 * Copyright (c) 1999 by David I. Bell
17 * Permission is granted to use, distribute, or modify this source,
18 * provided that this copyright notice remains intact.
19 * Permission to distribute sash derived code under the GPL has been granted.
21 * Based in part on the tar implementation from busybox-0.28
22 * Copyright (C) 1995 Bruce Perens
23 * This is free software under the GNU General Public License.
25 * This program is free software; you can redistribute it and/or modify
26 * it under the terms of the GNU General Public License as published by
27 * the Free Software Foundation; either version 2 of the License, or
28 * (at your option) any later version.
30 * This program is distributed in the hope that it will be useful,
31 * but WITHOUT ANY WARRANTY; without even the implied warranty of
32 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
33 * General Public License for more details.
35 * You should have received a copy of the GNU General Public License
36 * along with this program; if not, write to the Free Software
37 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
50 #include "unarchive.h"
53 #ifdef CONFIG_FEATURE_TAR_CREATE
55 /* Tar file constants */
56 # define TAR_MAGIC "ustar" /* ustar and a null */
57 # define TAR_VERSION " " /* Be compatable with GNU tar format */
60 # define MAJOR(dev) (((dev)>>8)&0xff)
61 # define MINOR(dev) ((dev)&0xff)
64 static const int TAR_BLOCK_SIZE = 512;
65 static const int TAR_MAGIC_LEN = 6;
66 static const int TAR_VERSION_LEN = 2;
68 /* POSIX tar Header Block, from POSIX 1003.1-1990 */
69 enum { NAME_SIZE = 100 }; /* because gcc won't let me use 'static const int' */
72 char name[NAME_SIZE]; /* 0-99 */
73 char mode[8]; /* 100-107 */
74 char uid[8]; /* 108-115 */
75 char gid[8]; /* 116-123 */
76 char size[12]; /* 124-135 */
77 char mtime[12]; /* 136-147 */
78 char chksum[8]; /* 148-155 */
79 char typeflag; /* 156-156 */
80 char linkname[NAME_SIZE]; /* 157-256 */
81 char magic[6]; /* 257-262 */
82 char version[2]; /* 263-264 */
83 char uname[32]; /* 265-296 */
84 char gname[32]; /* 297-328 */
85 char devmajor[8]; /* 329-336 */
86 char devminor[8]; /* 337-344 */
87 char prefix[155]; /* 345-499 */
88 char padding[12]; /* 500-512 (pad to exactly the TAR_BLOCK_SIZE) */
90 typedef struct TarHeader TarHeader;
93 ** writeTarFile(), writeFileToTarball(), and writeTarHeader() are
94 ** the only functions that deal with the HardLinkInfo structure.
95 ** Even these functions use the xxxHardLinkInfo() functions.
97 typedef struct HardLinkInfo HardLinkInfo;
100 HardLinkInfo *next; /* Next entry in list */
101 dev_t dev; /* Device number */
102 ino_t ino; /* Inode number */
103 short linkCount; /* (Hard) Link Count */
104 char name[1]; /* Start of filename (must be last) */
107 /* Some info to be carried along when creating a new tarball */
110 char* fileName; /* File name of the tarball */
111 int tarFd; /* Open-for-write file descriptor
113 struct stat statBuf; /* Stat info for the tarball, letting
114 us know the inode and device that the
115 tarball lives, so we can avoid trying
116 to include the tarball into itself */
117 int verboseFlag; /* Whether to print extra stuff or not */
118 char** excludeList; /* List of files to not include */
119 HardLinkInfo *hlInfoHead; /* Hard Link Tracking Information */
120 HardLinkInfo *hlInfo; /* Hard Link Info for the current file */
122 typedef struct TarBallInfo TarBallInfo;
124 /* A nice enum with all the possible tar file content types */
127 REGTYPE = '0', /* regular file */
128 REGTYPE0 = '\0', /* regular file (ancient bug compat)*/
129 LNKTYPE = '1', /* hard link */
130 SYMTYPE = '2', /* symbolic link */
131 CHRTYPE = '3', /* character special */
132 BLKTYPE = '4', /* block special */
133 DIRTYPE = '5', /* directory */
134 FIFOTYPE = '6', /* FIFO special */
135 CONTTYPE = '7', /* reserved */
136 GNULONGLINK = 'K', /* GNU long (>100 chars) link name */
137 GNULONGNAME = 'L', /* GNU long (>100 chars) file name */
139 typedef enum TarFileType TarFileType;
141 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
143 addHardLinkInfo (HardLinkInfo **hlInfoHeadPtr, dev_t dev, ino_t ino,
144 short linkCount, const char *name)
146 /* Note: hlInfoHeadPtr can never be NULL! */
147 HardLinkInfo *hlInfo;
149 hlInfo = (HardLinkInfo *)xmalloc(sizeof(HardLinkInfo)+strlen(name)+1);
151 hlInfo->next = *hlInfoHeadPtr;
152 *hlInfoHeadPtr = hlInfo;
155 hlInfo->linkCount = linkCount;
156 strcpy(hlInfo->name, name);
162 freeHardLinkInfo (HardLinkInfo **hlInfoHeadPtr)
164 HardLinkInfo *hlInfo = NULL;
165 HardLinkInfo *hlInfoNext = NULL;
168 hlInfo = *hlInfoHeadPtr;
170 hlInfoNext = hlInfo->next;
174 *hlInfoHeadPtr = NULL;
179 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
180 static HardLinkInfo *
181 findHardLinkInfo (HardLinkInfo *hlInfo, dev_t dev, ino_t ino)
184 if ((ino == hlInfo->ino) && (dev == hlInfo->dev))
186 hlInfo = hlInfo->next;
191 /* Put an octal string into the specified buffer.
192 * The number is zero and space padded and possibly null padded.
193 * Returns TRUE if successful. */
194 static int putOctal (char *cp, int len, long value)
198 char *tempString = tempBuffer;
200 /* Create a string of the specified length with an initial space,
201 * leading zeroes and the octal number, and a trailing null. */
202 sprintf (tempString, "%0*lo", len - 1, value);
204 /* If the string is too large, suppress the leading space. */
205 tempLength = strlen (tempString) + 1;
206 if (tempLength > len) {
211 /* If the string is still too large, suppress the trailing null. */
212 if (tempLength > len)
215 /* If the string is still too large, fail. */
216 if (tempLength > len)
219 /* Copy the string to the field. */
220 memcpy (cp, tempString, len);
225 /* Write out a tar header for the specified file/directory/whatever */
227 writeTarHeader(struct TarBallInfo *tbInfo, const char *header_name,
228 const char *real_name, struct stat *statbuf)
231 struct TarHeader header;
232 const unsigned char *cp = (const unsigned char *) &header;
233 ssize_t size = sizeof(struct TarHeader);
235 memset( &header, 0, size);
237 strncpy(header.name, header_name, sizeof(header.name));
239 putOctal(header.mode, sizeof(header.mode), statbuf->st_mode);
240 putOctal(header.uid, sizeof(header.uid), statbuf->st_uid);
241 putOctal(header.gid, sizeof(header.gid), statbuf->st_gid);
242 putOctal(header.size, sizeof(header.size), 0); /* Regular file size is handled later */
243 putOctal(header.mtime, sizeof(header.mtime), statbuf->st_mtime);
244 strncpy(header.magic, TAR_MAGIC TAR_VERSION,
245 TAR_MAGIC_LEN + TAR_VERSION_LEN );
247 /* Enter the user and group names (default to root if it fails) */
248 my_getpwuid(header.uname, statbuf->st_uid);
250 strcpy(header.uname, "root");
251 my_getgrgid(header.gname, statbuf->st_gid);
253 strcpy(header.uname, "root");
255 if (tbInfo->hlInfo) {
256 /* This is a hard link */
257 header.typeflag = LNKTYPE;
258 strncpy(header.linkname, tbInfo->hlInfo->name, sizeof(header.linkname));
259 } else if (S_ISLNK(statbuf->st_mode)) {
260 char *lpath = xreadlink(real_name);
261 if (!lpath) /* Already printed err msg inside xreadlink() */
263 header.typeflag = SYMTYPE;
264 strncpy(header.linkname, lpath, sizeof(header.linkname));
266 } else if (S_ISDIR(statbuf->st_mode)) {
267 header.typeflag = DIRTYPE;
268 strncat(header.name, "/", sizeof(header.name));
269 } else if (S_ISCHR(statbuf->st_mode)) {
270 header.typeflag = CHRTYPE;
271 putOctal(header.devmajor, sizeof(header.devmajor), MAJOR(statbuf->st_rdev));
272 putOctal(header.devminor, sizeof(header.devminor), MINOR(statbuf->st_rdev));
273 } else if (S_ISBLK(statbuf->st_mode)) {
274 header.typeflag = BLKTYPE;
275 putOctal(header.devmajor, sizeof(header.devmajor), MAJOR(statbuf->st_rdev));
276 putOctal(header.devminor, sizeof(header.devminor), MINOR(statbuf->st_rdev));
277 } else if (S_ISFIFO(statbuf->st_mode)) {
278 header.typeflag = FIFOTYPE;
279 } else if (S_ISREG(statbuf->st_mode)) {
280 header.typeflag = REGTYPE;
281 putOctal(header.size, sizeof(header.size), statbuf->st_size);
283 error_msg("%s: Unknown file type", real_name);
287 /* Calculate and store the checksum (i.e., the sum of all of the bytes of
288 * the header). The checksum field must be filled with blanks for the
289 * calculation. The checksum field is formatted differently from the
290 * other fields: it has [6] digits, a null, then a space -- rather than
291 * digits, followed by a null like the other fields... */
292 memset(header.chksum, ' ', sizeof(header.chksum));
293 cp = (const unsigned char *) &header;
296 putOctal(header.chksum, 7, chksum);
298 /* Now write the header out to disk */
299 if ((size=full_write(tbInfo->tarFd, (char*)&header, sizeof(struct TarHeader))) < 0) {
300 error_msg(io_error, real_name, strerror(errno));
303 /* Pad the header up to the tar block size */
304 for (; size<TAR_BLOCK_SIZE; size++) {
305 write(tbInfo->tarFd, "\0", 1);
307 /* Now do the verbose thing (or not) */
308 if (tbInfo->verboseFlag==TRUE) {
310 if (tbInfo->tarFd == fileno(stdout)) // If the archive goes to stdout, verbose to stderr
312 fprintf(vbFd, "%s\n", header.name);
318 static int exclude_file(char **excluded_files, const char *file)
322 if (excluded_files == NULL)
325 for (i = 0; excluded_files[i] != NULL; i++) {
326 if (excluded_files[i][0] == '/') {
327 if (fnmatch(excluded_files[i], file,
328 FNM_PATHNAME | FNM_LEADING_DIR) == 0)
333 for (p = file; p[0] != '\0'; p++) {
334 if ((p == file || p[-1] == '/') && p[0] != '/' &&
335 fnmatch(excluded_files[i], p,
336 FNM_PATHNAME | FNM_LEADING_DIR) == 0)
345 static int writeFileToTarball(const char *fileName, struct stat *statbuf, void* userData)
347 struct TarBallInfo *tbInfo = (struct TarBallInfo *)userData;
348 const char *header_name;
351 ** Check to see if we are dealing with a hard link.
353 ** Treat the first occurance of a given dev/inode as a file while
354 ** treating any additional occurances as hard links. This is done
355 ** by adding the file information to the HardLinkInfo linked list.
357 tbInfo->hlInfo = NULL;
358 if (statbuf->st_nlink > 1) {
359 tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf->st_dev,
361 if (tbInfo->hlInfo == NULL)
362 addHardLinkInfo (&tbInfo->hlInfoHead, statbuf->st_dev,
363 statbuf->st_ino, statbuf->st_nlink, fileName);
366 /* It is against the rules to archive a socket */
367 if (S_ISSOCK(statbuf->st_mode)) {
368 error_msg("%s: socket ignored", fileName);
372 /* It is a bad idea to store the archive we are in the process of creating,
373 * so check the device and inode to be sure that this particular file isn't
375 if (tbInfo->statBuf.st_dev == statbuf->st_dev &&
376 tbInfo->statBuf.st_ino == statbuf->st_ino) {
377 error_msg("%s: file is the archive; skipping", fileName);
381 header_name = fileName;
382 while (header_name[0] == '/') {
383 static int alreadyWarned=FALSE;
384 if (alreadyWarned==FALSE) {
385 error_msg("Removing leading '/' from member names");
391 if (strlen(fileName) >= NAME_SIZE) {
392 error_msg(name_longer_than_foo, NAME_SIZE);
396 if (header_name[0] == '\0')
399 # if defined CONFIG_FEATURE_TAR_EXCLUDE
400 if (exclude_file(tbInfo->excludeList, header_name)) {
403 # endif //CONFIG_FEATURE_TAR_EXCLUDE
405 if (writeTarHeader(tbInfo, header_name, fileName, statbuf)==FALSE) {
409 /* Now, if the file is a regular file, copy it out to the tarball */
410 if ((tbInfo->hlInfo == NULL)
411 && (S_ISREG(statbuf->st_mode))) {
414 ssize_t size=0, readSize=0;
416 /* open the file we want to archive, and make sure all is well */
417 if ((inputFileFd = open(fileName, O_RDONLY)) < 0) {
418 error_msg("%s: Cannot open: %s", fileName, strerror(errno));
422 /* write the file to the archive */
423 while ( (size = full_read(inputFileFd, buffer, sizeof(buffer))) > 0 ) {
424 if (full_write(tbInfo->tarFd, buffer, size) != size ) {
425 /* Output file seems to have a problem */
426 error_msg(io_error, fileName, strerror(errno));
432 error_msg(io_error, fileName, strerror(errno));
435 /* Pad the file up to the tar block size */
436 for (; (readSize%TAR_BLOCK_SIZE) != 0; readSize++) {
437 write(tbInfo->tarFd, "\0", 1);
445 static int writeTarFile(const char* tarName, int verboseFlag, char **argv,
451 struct TarBallInfo tbInfo;
452 tbInfo.verboseFlag = verboseFlag;
453 tbInfo.hlInfoHead = NULL;
455 /* Make sure there is at least one file to tar up. */
457 error_msg_and_die("Cowardly refusing to create an empty archive");
459 /* Open the tar file for writing. */
460 if (!strcmp(tarName, "-"))
461 tbInfo.tarFd = fileno(stdout);
463 tbInfo.tarFd = open (tarName, O_WRONLY | O_CREAT | O_TRUNC, 0644);
464 if (tbInfo.tarFd < 0) {
465 perror_msg( "Error opening '%s'", tarName);
466 freeHardLinkInfo(&tbInfo.hlInfoHead);
469 tbInfo.excludeList=excludeList;
470 /* Store the stat info for the tarball's file, so
471 * can avoid including the tarball into itself.... */
472 if (fstat(tbInfo.tarFd, &tbInfo.statBuf) < 0)
473 error_msg_and_die(io_error, tarName, strerror(errno));
475 /* Read the directory/files and iterate over them one at a time */
476 while (*argv != NULL) {
477 if (recursive_action(*argv++, TRUE, FALSE, FALSE,
478 writeFileToTarball, writeFileToTarball,
479 (void*) &tbInfo) == FALSE) {
483 /* Write two empty blocks to the end of the archive */
484 for (size=0; size<(2*TAR_BLOCK_SIZE); size++) {
485 write(tbInfo.tarFd, "\0", 1);
488 /* To be pedantically correct, we would check if the tarball
489 * is smaller than 20 tar blocks, and pad it if it was smaller,
490 * but that isn't necessary for GNU tar interoperability, and
491 * so is considered a waste of space */
493 /* Hang up the tools, close up shop, head home */
495 if (errorFlag == TRUE) {
496 error_msg("Error exit delayed from previous errors");
497 freeHardLinkInfo(&tbInfo.hlInfoHead);
500 freeHardLinkInfo(&tbInfo.hlInfoHead);
505 void append_file_to_list(const char *new_name, char ***list, int *list_count)
507 *list = realloc(*list, sizeof(char *) * (*list_count + 2));
508 (*list)[*list_count] = xstrdup(new_name);
510 (*list)[*list_count] = NULL;
513 void append_file_list_to_list(char *filename, char ***name_list, int *num_of_entries)
519 src_stream = xfopen(filename, "r");
520 while ((line = get_line_from_file(src_stream)) != NULL) {
521 line_ptr = last_char_is(line, '\n');
525 append_file_to_list(line, name_list, num_of_entries);
531 #ifdef CONFIG_FEATURE_TAR_EXCLUDE
533 * Create a list of names that are in the include list AND NOT in the exclude lists
535 char **list_and_not_list(char **include_list, char **exclude_list)
537 char **new_include_list = NULL;
538 int new_include_count = 0;
539 int include_count = 0;
542 if (include_list == NULL) {
546 while (include_list[include_count] != NULL) {
549 while (exclude_list[exclude_count] != NULL) {
550 if (strcmp(include_list[include_count], exclude_list[exclude_count]) == 0) {
557 if (found == FALSE) {
558 new_include_list = realloc(new_include_list, sizeof(char *) * (include_count + 2));
559 new_include_list[new_include_count] = include_list[include_count];
562 free(include_list[include_count]);
566 new_include_list[new_include_count] = NULL;
567 return(new_include_list);
571 int tar_main(int argc, char **argv)
574 /* These are optional */
576 untar_from_stdin = 2,
578 /* Require one and only one of these */
584 FILE *src_stream = NULL;
585 FILE *uncompressed_stream = NULL;
586 char **include_list = NULL;
587 char **exclude_list = NULL;
588 char *src_filename = NULL;
589 char *dst_prefix = NULL;
590 char *file_list_name = NULL;
592 unsigned short untar_funct = 0;
593 unsigned short untar_funct_required = 0;
594 unsigned short extract_function = 0;
595 int include_list_count = 0;
596 int exclude_list_count = 0;
604 /* Prepend '-' to the first argument if required */
605 if (argv[1][0] != '-') {
606 char *tmp = xmalloc(strlen(argv[1]) + 2);
608 strcpy(tmp + 1, argv[1]);
612 while ((opt = getopt(argc, argv, "ctxT:X:C:f:Opvz")) != -1) {
615 /* One and only one of these is required */
617 untar_funct_required |= untar_create;
620 untar_funct_required |= untar_list;
621 extract_function |= extract_list |extract_unconditional;
624 untar_funct_required |= untar_extract;
625 extract_function |= (extract_all_to_fs | extract_unconditional | extract_create_leading_dirs);
628 /* These are optional */
629 /* Exclude or Include files listed in <filename>*/
630 #ifdef CONFIG_FEATURE_TAR_EXCLUDE
632 append_file_list_to_list(optarg, &exclude_list, &exclude_list_count);
636 // by default a list is an include list
637 append_file_list_to_list(optarg, &include_list, &include_list_count);
640 case 'C': // Change to dir <optarg>
641 /* Make sure dst_prefix ends in a '/' */
642 dst_prefix = concat_path_file(optarg, "/");
644 case 'f': // archive filename
645 if (strcmp(optarg, "-") == 0) {
646 // Untar from stdin to stdout
647 untar_funct |= untar_from_stdin;
649 untar_funct |= untar_from_file;
650 src_filename = xstrdup(optarg);
654 extract_function |= extract_to_stdout;
659 if (extract_function & extract_list) {
660 extract_function |= extract_verbose_list;
662 extract_function |= extract_list;
664 #ifdef CONFIG_FEATURE_TAR_GZIP
666 untar_funct |= untar_unzip;
674 /* Make sure the valid arguments were passed */
675 if (untar_funct_required == 0) {
676 error_msg_and_die("You must specify one of the `-ctx' options");
678 if ((untar_funct_required != untar_create) &&
679 (untar_funct_required != untar_extract) &&
680 (untar_funct_required != untar_list)) {
681 error_msg_and_die("You may not specify more than one `ctx' option.");
683 untar_funct |= untar_funct_required;
685 /* Setup an array of filenames to work with */
686 while (optind < argc) {
687 append_file_to_list(argv[optind], &include_list, &include_list_count);
691 if (extract_function & (extract_list | extract_all_to_fs)) {
692 if (dst_prefix == NULL) {
693 dst_prefix = xstrdup("./");
696 /* Setup the source of the tar data */
697 if (untar_funct & untar_from_file) {
698 src_stream = xfopen(src_filename, "r");
702 #ifdef CONFIG_FEATURE_TAR_GZIP
703 /* Get a binary tree of all the tar file headers */
704 if (untar_funct & untar_unzip) {
705 uncompressed_stream = gz_open(src_stream, &gunzip_pid);
707 #endif // CONFIG_FEATURE_TAR_GZIP
708 uncompressed_stream = src_stream;
710 /* extract or list archive */
711 unarchive(uncompressed_stream, stdout, &get_header_tar, extract_function, dst_prefix, include_list, exclude_list);
712 fclose(uncompressed_stream);
714 #ifdef CONFIG_FEATURE_TAR_CREATE
715 /* create an archive */
716 else if (untar_funct & untar_create) {
717 int verboseFlag = FALSE;
719 #ifdef CONFIG_FEATURE_TAR_GZIP
720 if (untar_funct && untar_unzip) {
721 error_msg_and_die("Creation of compressed tarfile not internally support by tar, pipe to busybox gunzip");
723 #endif // CONFIG_FEATURE_TAR_GZIP
724 if (extract_function & extract_verbose_list) {
727 writeTarFile(src_filename, verboseFlag, &argv[argc - 1], include_list);
729 #endif // CONFIG_FEATURE_TAR_CREATE
732 #ifdef CONFIG_FEATURE_TAR_GZIP
733 if (untar_funct & untar_unzip) {
736 gz_close(gunzip_pid);
738 #endif // CONFIG_FEATURE_TAR_GZIP
742 if (file_list_name) {
743 free(file_list_name);
745 return(EXIT_SUCCESS);