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) 1999,2000,2001 by Lineo, inc.
13 * Written by Erik Andersen <andersen@lineo.com>, <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
52 #ifdef BB_FEATURE_TAR_CREATE
54 /* Tar file constants */
55 # define TAR_MAGIC "ustar" /* ustar and a null */
56 # define TAR_VERSION " " /* Be compatable with GNU tar format */
59 # define MAJOR(dev) (((dev)>>8)&0xff)
60 # define MINOR(dev) ((dev)&0xff)
63 static const int TAR_BLOCK_SIZE = 512;
64 static const int TAR_MAGIC_LEN = 6;
65 static const int TAR_VERSION_LEN = 2;
67 /* POSIX tar Header Block, from POSIX 1003.1-1990 */
68 enum { NAME_SIZE = 100 }; /* because gcc won't let me use 'static const int' */
71 char name[NAME_SIZE]; /* 0-99 */
72 char mode[8]; /* 100-107 */
73 char uid[8]; /* 108-115 */
74 char gid[8]; /* 116-123 */
75 char size[12]; /* 124-135 */
76 char mtime[12]; /* 136-147 */
77 char chksum[8]; /* 148-155 */
78 char typeflag; /* 156-156 */
79 char linkname[NAME_SIZE]; /* 157-256 */
80 char magic[6]; /* 257-262 */
81 char version[2]; /* 263-264 */
82 char uname[32]; /* 265-296 */
83 char gname[32]; /* 297-328 */
84 char devmajor[8]; /* 329-336 */
85 char devminor[8]; /* 337-344 */
86 char prefix[155]; /* 345-499 */
87 char padding[12]; /* 500-512 (pad to exactly the TAR_BLOCK_SIZE) */
89 typedef struct TarHeader TarHeader;
92 ** writeTarFile(), writeFileToTarball(), and writeTarHeader() are
93 ** the only functions that deal with the HardLinkInfo structure.
94 ** Even these functions use the xxxHardLinkInfo() functions.
96 typedef struct HardLinkInfo HardLinkInfo;
99 HardLinkInfo *next; /* Next entry in list */
100 dev_t dev; /* Device number */
101 ino_t ino; /* Inode number */
102 short linkCount; /* (Hard) Link Count */
103 char name[1]; /* Start of filename (must be last) */
106 /* Some info to be carried along when creating a new tarball */
109 char* fileName; /* File name of the tarball */
110 int tarFd; /* Open-for-write file descriptor
112 struct stat statBuf; /* Stat info for the tarball, letting
113 us know the inode and device that the
114 tarball lives, so we can avoid trying
115 to include the tarball into itself */
116 int verboseFlag; /* Whether to print extra stuff or not */
117 char** excludeList; /* List of files to not include */
118 HardLinkInfo *hlInfoHead; /* Hard Link Tracking Information */
119 HardLinkInfo *hlInfo; /* Hard Link Info for the current file */
121 typedef struct TarBallInfo TarBallInfo;
123 /* A nice enum with all the possible tar file content types */
126 REGTYPE = '0', /* regular file */
127 REGTYPE0 = '\0', /* regular file (ancient bug compat)*/
128 LNKTYPE = '1', /* hard link */
129 SYMTYPE = '2', /* symbolic link */
130 CHRTYPE = '3', /* character special */
131 BLKTYPE = '4', /* block special */
132 DIRTYPE = '5', /* directory */
133 FIFOTYPE = '6', /* FIFO special */
134 CONTTYPE = '7', /* reserved */
135 GNULONGLINK = 'K', /* GNU long (>100 chars) link name */
136 GNULONGNAME = 'L', /* GNU long (>100 chars) file name */
138 typedef enum TarFileType TarFileType;
140 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
142 addHardLinkInfo (HardLinkInfo **hlInfoHeadPtr, dev_t dev, ino_t ino,
143 short linkCount, const char *name)
145 /* Note: hlInfoHeadPtr can never be NULL! */
146 HardLinkInfo *hlInfo;
148 hlInfo = (HardLinkInfo *)xmalloc(sizeof(HardLinkInfo)+strlen(name)+1);
150 hlInfo->next = *hlInfoHeadPtr;
151 *hlInfoHeadPtr = hlInfo;
154 hlInfo->linkCount = linkCount;
155 strcpy(hlInfo->name, name);
161 freeHardLinkInfo (HardLinkInfo **hlInfoHeadPtr)
163 HardLinkInfo *hlInfo = NULL;
164 HardLinkInfo *hlInfoNext = NULL;
167 hlInfo = *hlInfoHeadPtr;
169 hlInfoNext = hlInfo->next;
173 *hlInfoHeadPtr = NULL;
178 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
179 static HardLinkInfo *
180 findHardLinkInfo (HardLinkInfo *hlInfo, dev_t dev, ino_t ino)
183 if ((ino == hlInfo->ino) && (dev == hlInfo->dev))
185 hlInfo = hlInfo->next;
190 /* Put an octal string into the specified buffer.
191 * The number is zero and space padded and possibly null padded.
192 * Returns TRUE if successful. */
193 static int putOctal (char *cp, int len, long value)
197 char *tempString = tempBuffer;
199 /* Create a string of the specified length with an initial space,
200 * leading zeroes and the octal number, and a trailing null. */
201 sprintf (tempString, "%0*lo", len - 1, value);
203 /* If the string is too large, suppress the leading space. */
204 tempLength = strlen (tempString) + 1;
205 if (tempLength > len) {
210 /* If the string is still too large, suppress the trailing null. */
211 if (tempLength > len)
214 /* If the string is still too large, fail. */
215 if (tempLength > len)
218 /* Copy the string to the field. */
219 memcpy (cp, tempString, len);
224 /* Write out a tar header for the specified file/directory/whatever */
226 writeTarHeader(struct TarBallInfo *tbInfo, const char *header_name,
227 const char *real_name, struct stat *statbuf)
230 struct TarHeader header;
231 const unsigned char *cp = (const unsigned char *) &header;
232 ssize_t size = sizeof(struct TarHeader);
234 memset( &header, 0, size);
236 strncpy(header.name, header_name, sizeof(header.name));
238 putOctal(header.mode, sizeof(header.mode), statbuf->st_mode);
239 putOctal(header.uid, sizeof(header.uid), statbuf->st_uid);
240 putOctal(header.gid, sizeof(header.gid), statbuf->st_gid);
241 putOctal(header.size, sizeof(header.size), 0); /* Regular file size is handled later */
242 putOctal(header.mtime, sizeof(header.mtime), statbuf->st_mtime);
243 strncpy(header.magic, TAR_MAGIC TAR_VERSION,
244 TAR_MAGIC_LEN + TAR_VERSION_LEN );
246 /* Enter the user and group names (default to root if it fails) */
247 my_getpwuid(header.uname, statbuf->st_uid);
249 strcpy(header.uname, "root");
250 my_getgrgid(header.gname, statbuf->st_gid);
252 strcpy(header.uname, "root");
254 if (tbInfo->hlInfo) {
255 /* This is a hard link */
256 header.typeflag = LNKTYPE;
257 strncpy(header.linkname, tbInfo->hlInfo->name, sizeof(header.linkname));
258 } else if (S_ISLNK(statbuf->st_mode)) {
259 char *lpath = xreadlink(real_name);
260 if (!lpath) /* Already printed err msg inside xreadlink() */
262 header.typeflag = SYMTYPE;
263 strncpy(header.linkname, lpath, sizeof(header.linkname));
265 } else if (S_ISDIR(statbuf->st_mode)) {
266 header.typeflag = DIRTYPE;
267 strncat(header.name, "/", sizeof(header.name));
268 } else if (S_ISCHR(statbuf->st_mode)) {
269 header.typeflag = CHRTYPE;
270 putOctal(header.devmajor, sizeof(header.devmajor), MAJOR(statbuf->st_rdev));
271 putOctal(header.devminor, sizeof(header.devminor), MINOR(statbuf->st_rdev));
272 } else if (S_ISBLK(statbuf->st_mode)) {
273 header.typeflag = BLKTYPE;
274 putOctal(header.devmajor, sizeof(header.devmajor), MAJOR(statbuf->st_rdev));
275 putOctal(header.devminor, sizeof(header.devminor), MINOR(statbuf->st_rdev));
276 } else if (S_ISFIFO(statbuf->st_mode)) {
277 header.typeflag = FIFOTYPE;
278 } else if (S_ISREG(statbuf->st_mode)) {
279 header.typeflag = REGTYPE;
280 putOctal(header.size, sizeof(header.size), statbuf->st_size);
282 error_msg("%s: Unknown file type", real_name);
286 /* Calculate and store the checksum (i.e., the sum of all of the bytes of
287 * the header). The checksum field must be filled with blanks for the
288 * calculation. The checksum field is formatted differently from the
289 * other fields: it has [6] digits, a null, then a space -- rather than
290 * digits, followed by a null like the other fields... */
291 memset(header.chksum, ' ', sizeof(header.chksum));
292 cp = (const unsigned char *) &header;
295 putOctal(header.chksum, 7, chksum);
297 /* Now write the header out to disk */
298 if ((size=full_write(tbInfo->tarFd, (char*)&header, sizeof(struct TarHeader))) < 0) {
299 error_msg(io_error, real_name, strerror(errno));
302 /* Pad the header up to the tar block size */
303 for (; size<TAR_BLOCK_SIZE; size++) {
304 write(tbInfo->tarFd, "\0", 1);
306 /* Now do the verbose thing (or not) */
307 if (tbInfo->verboseFlag==TRUE) {
309 if (tbInfo->tarFd == fileno(stdout)) // If the archive goes to stdout, verbose to stderr
311 fprintf(vbFd, "%s\n", header.name);
317 static int exclude_file(char **excluded_files, const char *file)
321 if (excluded_files == NULL)
324 for (i = 0; excluded_files[i] != NULL; i++) {
325 if (excluded_files[i][0] == '/') {
326 if (fnmatch(excluded_files[i], file,
327 FNM_PATHNAME | FNM_LEADING_DIR) == 0)
332 for (p = file; p[0] != '\0'; p++) {
333 if ((p == file || p[-1] == '/') && p[0] != '/' &&
334 fnmatch(excluded_files[i], p,
335 FNM_PATHNAME | FNM_LEADING_DIR) == 0)
344 static int writeFileToTarball(const char *fileName, struct stat *statbuf, void* userData)
346 struct TarBallInfo *tbInfo = (struct TarBallInfo *)userData;
347 const char *header_name;
350 ** Check to see if we are dealing with a hard link.
352 ** Treat the first occurance of a given dev/inode as a file while
353 ** treating any additional occurances as hard links. This is done
354 ** by adding the file information to the HardLinkInfo linked list.
356 tbInfo->hlInfo = NULL;
357 if (statbuf->st_nlink > 1) {
358 tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf->st_dev,
360 if (tbInfo->hlInfo == NULL)
361 addHardLinkInfo (&tbInfo->hlInfoHead, statbuf->st_dev,
362 statbuf->st_ino, statbuf->st_nlink, fileName);
365 /* It is against the rules to archive a socket */
366 if (S_ISSOCK(statbuf->st_mode)) {
367 error_msg("%s: socket ignored", fileName);
371 /* It is a bad idea to store the archive we are in the process of creating,
372 * so check the device and inode to be sure that this particular file isn't
374 if (tbInfo->statBuf.st_dev == statbuf->st_dev &&
375 tbInfo->statBuf.st_ino == statbuf->st_ino) {
376 error_msg("%s: file is the archive; skipping", fileName);
380 header_name = fileName;
381 while (header_name[0] == '/') {
382 static int alreadyWarned=FALSE;
383 if (alreadyWarned==FALSE) {
384 error_msg("Removing leading '/' from member names");
390 if (strlen(fileName) >= NAME_SIZE) {
391 error_msg(name_longer_than_foo, NAME_SIZE);
395 if (header_name[0] == '\0')
398 # if defined BB_FEATURE_TAR_EXCLUDE
399 if (exclude_file(tbInfo->excludeList, header_name)) {
402 # endif //BB_FEATURE_TAR_EXCLUDE
404 if (writeTarHeader(tbInfo, header_name, fileName, statbuf)==FALSE) {
408 /* Now, if the file is a regular file, copy it out to the tarball */
409 if ((tbInfo->hlInfo == NULL)
410 && (S_ISREG(statbuf->st_mode))) {
413 ssize_t size=0, readSize=0;
415 /* open the file we want to archive, and make sure all is well */
416 if ((inputFileFd = open(fileName, O_RDONLY)) < 0) {
417 error_msg("%s: Cannot open: %s", fileName, strerror(errno));
421 /* write the file to the archive */
422 while ( (size = full_read(inputFileFd, buffer, sizeof(buffer))) > 0 ) {
423 if (full_write(tbInfo->tarFd, buffer, size) != size ) {
424 /* Output file seems to have a problem */
425 error_msg(io_error, fileName, strerror(errno));
431 error_msg(io_error, fileName, strerror(errno));
434 /* Pad the file up to the tar block size */
435 for (; (readSize%TAR_BLOCK_SIZE) != 0; readSize++) {
436 write(tbInfo->tarFd, "\0", 1);
444 static int writeTarFile(const char* tarName, int verboseFlag, char **argv,
450 struct TarBallInfo tbInfo;
451 tbInfo.verboseFlag = verboseFlag;
452 tbInfo.hlInfoHead = NULL;
454 /* Make sure there is at least one file to tar up. */
456 error_msg_and_die("Cowardly refusing to create an empty archive");
458 /* Open the tar file for writing. */
459 if (!strcmp(tarName, "-"))
460 tbInfo.tarFd = fileno(stdout);
462 tbInfo.tarFd = open (tarName, O_WRONLY | O_CREAT | O_TRUNC, 0644);
463 if (tbInfo.tarFd < 0) {
464 perror_msg( "Error opening '%s'", tarName);
465 freeHardLinkInfo(&tbInfo.hlInfoHead);
468 tbInfo.excludeList=excludeList;
469 /* Store the stat info for the tarball's file, so
470 * can avoid including the tarball into itself.... */
471 if (fstat(tbInfo.tarFd, &tbInfo.statBuf) < 0)
472 error_msg_and_die(io_error, tarName, strerror(errno));
474 /* Read the directory/files and iterate over them one at a time */
475 while (*argv != NULL) {
476 if (recursive_action(*argv++, TRUE, FALSE, FALSE,
477 writeFileToTarball, writeFileToTarball,
478 (void*) &tbInfo) == FALSE) {
482 /* Write two empty blocks to the end of the archive */
483 for (size=0; size<(2*TAR_BLOCK_SIZE); size++) {
484 write(tbInfo.tarFd, "\0", 1);
487 /* To be pedantically correct, we would check if the tarball
488 * is smaller than 20 tar blocks, and pad it if it was smaller,
489 * but that isn't necessary for GNU tar interoperability, and
490 * so is considered a waste of space */
492 /* Hang up the tools, close up shop, head home */
494 if (errorFlag == TRUE) {
495 error_msg("Error exit delayed from previous errors");
496 freeHardLinkInfo(&tbInfo.hlInfoHead);
499 freeHardLinkInfo(&tbInfo.hlInfoHead);
504 void append_file_to_list(const char *new_name, char ***list, int *list_count)
506 *list = realloc(*list, sizeof(char *) * (*list_count + 2));
507 if (last_char_is(new_name, '/')) {
508 (*list)[*list_count] = concat_path_file(new_name, "*");
510 (*list)[*list_count] = xstrdup(new_name);
513 (*list)[*list_count] = NULL;
516 void append_file_list_to_list(char *filename, char ***name_list, int *num_of_entries)
522 src_stream = xfopen(filename, "r");
523 while ((line = get_line_from_file(src_stream)) != NULL) {
524 line_ptr = last_char_is(line, '\n');
528 append_file_to_list(line, name_list, num_of_entries);
534 #ifdef BB_FEATURE_TAR_EXCLUDE
536 * Create a list of names that are in the include list AND NOT in the exclude lists
538 char **list_and_not_list(char **include_list, char **exclude_list)
540 char **new_include_list = NULL;
541 int new_include_count = 0;
542 int include_count = 0;
545 if (include_list == NULL) {
549 while (include_list[include_count] != NULL) {
552 while (exclude_list[exclude_count] != NULL) {
553 if (strcmp(include_list[include_count], exclude_list[exclude_count]) == 0) {
560 if (found == FALSE) {
561 new_include_list = realloc(new_include_list, sizeof(char *) * (include_count + 2));
562 new_include_list[new_include_count] = include_list[include_count];
565 free(include_list[include_count]);
569 new_include_list[new_include_count] = NULL;
570 return(new_include_list);
574 int tar_main(int argc, char **argv)
577 /* These are optional */
579 untar_from_stdin = 2,
581 /* Require one and only one of these */
587 #ifdef BB_FEATURE_TAR_EXCLUDE
588 char **exclude_list = NULL;
589 int exclude_list_count = 0;
592 FILE *src_stream = NULL;
593 FILE *uncompressed_stream = NULL;
594 char **include_list = NULL;
595 char **the_real_list = NULL;
596 char *src_filename = NULL;
597 char *dst_prefix = NULL;
598 char *file_list_name = NULL;
600 unsigned short untar_funct = 0;
601 unsigned short untar_funct_required = 0;
602 unsigned short extract_function = 0;
603 int include_list_count = 0;
611 /* Prepend '-' to the first argument if required */
612 if (argv[1][0] != '-') {
613 char *tmp = xmalloc(strlen(argv[1]) + 2);
615 strcpy(tmp + 1, argv[1]);
619 while ((opt = getopt(argc, argv, "ctxT:X:C:f:Opvz")) != -1) {
622 /* One and only one of these is required */
624 untar_funct_required |= untar_create;
627 untar_funct_required |= untar_list;
628 extract_function |= extract_list |extract_unconditional;
631 untar_funct_required |= untar_extract;
632 extract_function |= (extract_all_to_fs | extract_unconditional | extract_create_leading_dirs);
635 /* These are optional */
636 /* Exclude or Include files listed in <filename>*/
637 #ifdef BB_FEATURE_TAR_EXCLUDE
639 append_file_list_to_list(optarg, &exclude_list, &exclude_list_count);
643 // by default a list is an include list
644 append_file_list_to_list(optarg, &include_list, &include_list_count);
647 case 'C': // Change to dir <optarg>
648 /* Make sure dst_prefix ends in a '/' */
649 dst_prefix = concat_path_file(optarg, "/");
651 case 'f': // archive filename
652 if (strcmp(optarg, "-") == 0) {
653 // Untar from stdin to stdout
654 untar_funct |= untar_from_stdin;
656 untar_funct |= untar_from_file;
657 src_filename = xstrdup(optarg);
661 extract_function |= extract_to_stdout;
666 if (extract_function & extract_list) {
667 extract_function |= extract_verbose_list;
669 extract_function |= extract_list;
671 #ifdef BB_FEATURE_TAR_GZIP
673 untar_funct |= untar_unzip;
681 /* Make sure the valid arguments were passed */
682 if (untar_funct_required == 0) {
683 error_msg_and_die("You must specify one of the `-ctx' options");
685 if ((untar_funct_required != untar_create) &&
686 (untar_funct_required != untar_extract) &&
687 (untar_funct_required != untar_list)) {
688 error_msg_and_die("You may not specify more than one `ctx' option.");
690 untar_funct |= untar_funct_required;
692 /* Setup an array of filenames to work with */
693 while (optind < argc) {
694 append_file_to_list(argv[optind], &include_list, &include_list_count);
698 /* By default the include list is the list we act on */
699 the_real_list = include_list;
701 #ifdef BB_FEATURE_TAR_EXCLUDE
702 if (exclude_list != NULL) {
703 if (include_list == NULL) {
704 /* the_real_list is an exclude list */
705 extract_function |= extract_exclude_list;
706 the_real_list = exclude_list;
708 /* Both an exclude and include file list was present,
709 * its an exclude from include list only.
710 * Remove excluded files from the include list
712 the_real_list = list_and_not_list(include_list, exclude_list);
717 if (extract_function & (extract_list | extract_all_to_fs)) {
718 if (dst_prefix == NULL) {
719 dst_prefix = xstrdup("./");
722 /* Setup the source of the tar data */
723 if (untar_funct & untar_from_file) {
724 src_stream = xfopen(src_filename, "r");
728 #ifdef BB_FEATURE_TAR_GZIP
729 /* Get a binary tree of all the tar file headers */
730 if (untar_funct & untar_unzip) {
731 uncompressed_stream = gz_open(src_stream, &gunzip_pid);
733 #endif // BB_FEATURE_TAR_GZIP
734 uncompressed_stream = src_stream;
736 /* extract or list archive */
737 unarchive(uncompressed_stream, stdout, &get_header_tar, extract_function, dst_prefix, the_real_list);
738 fclose(uncompressed_stream);
740 #ifdef BB_FEATURE_TAR_CREATE
741 /* create an archive */
742 else if (untar_funct & untar_create) {
743 int verboseFlag = FALSE;
745 #ifdef BB_FEATURE_TAR_GZIP
746 if (untar_funct && untar_unzip) {
747 error_msg_and_die("Creation of compressed tarfile not internally support by tar, pipe to busybox gunzip");
749 #endif // BB_FEATURE_TAR_GZIP
750 if (extract_function & extract_verbose_list) {
753 writeTarFile(src_filename, verboseFlag, &argv[argc - 1], the_real_list);
755 #endif // BB_FEATURE_TAR_CREATE
758 #ifdef BB_FEATURE_TAR_GZIP
759 if (untar_funct & untar_unzip) {
762 gz_close(gunzip_pid);
764 #endif // BB_FEATURE_TAR_GZIP
768 if (file_list_name) {
769 free(file_list_name);
771 return(EXIT_SUCCESS);