2 * Copyright (C) 2000 by Glenn McGrath
3 * Copyright (C) 2001 by Laurence Anderson
5 * Based on previous work by busybox developers and others.
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU Library General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
30 #define CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY 1
31 #define CONFIG_FEATURE_TAR_GNU_EXTENSIONS
33 #ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
34 static char *longname
= NULL
;
35 static char *linkname
= NULL
;
38 extern void seek_sub_file(FILE *src_stream
, const int count
);
39 extern char *extract_archive(FILE *src_stream
, FILE *out_stream
, const file_header_t
*file_entry
,
40 const int function
, const char *prefix
);
41 extern ssize_t
seek_by_read(FILE* fd
, size_t len
);
44 #ifdef L_archive_offset
47 extern off_t archive_offset
;
50 #ifdef L_seek_sub_file
52 ssize_t
seek_by_read(FILE* fd
, size_t len
)
54 ssize_t cc
, total
= 0;
58 cc
= fread(buf
, sizeof(buf
[0]),
59 len
> SEEK_BUF
? SEEK_BUF
: len
,
65 if(feof(fd
) || ferror(fd
))
71 void seek_sub_file(FILE *src_stream
, const int count
)
73 /* Try to fseek as faster */
74 archive_offset
+= count
;
75 seek_by_read(src_stream
, count
);
82 #ifdef L_extract_archive
83 /* Extract the data postioned at src_stream to either filesystem, stdout or
84 * buffer depending on the value of 'function' which is defined in libbb.h
86 * prefix doesnt have to be just a directory, it may prefix the filename as well.
88 * e.g. '/var/lib/dpkg/info/dpkg.' will extract all files to the base bath
89 * '/var/lib/dpkg/info/' and all files/dirs created in that dir will have
90 * 'dpkg.' as their prefix
92 * For this reason if prefix does point to a dir then it must end with a
93 * trailing '/' or else the last dir will be assumed to be the file prefix
95 char *extract_archive(FILE *src_stream
, FILE *out_stream
, const file_header_t
*file_entry
,
96 const int function
, const char *prefix
)
98 FILE *dst_stream
= NULL
;
99 char *full_name
= NULL
;
100 char *full_link_name
= NULL
;
105 /* prefix doesnt have to be a proper path it may prepend
106 * the filename as well */
107 if (prefix
!= NULL
) {
108 /* strip leading '/' in filename to extract as prefix may not be dir */
109 /* Cant use concat_path_file here as prefix might not be a directory */
110 char *path
= file_entry
->name
;
111 if (strncmp("./", path
, 2) == 0) {
113 if (strlen(path
) == 0) {
117 full_name
= xmalloc(strlen(prefix
) + strlen(path
) + 1);
118 strcpy(full_name
, prefix
);
119 strcat(full_name
, path
);
120 if ( file_entry
->link_name
){
121 full_link_name
= xmalloc(strlen(prefix
) + strlen(file_entry
->link_name
) + 1);
122 strcpy(full_link_name
, prefix
);
123 strcat(full_link_name
, file_entry
->link_name
);
126 full_name
= strdup(file_entry
->name
);
127 if ( file_entry
->link_name
)
128 full_link_name
= strdup(file_entry
->link_name
);
132 if (function
& extract_to_stdout
) {
133 if (S_ISREG(file_entry
->mode
)) {
134 copy_file_chunk(src_stream
, out_stream
, file_entry
->size
);
135 archive_offset
+= file_entry
->size
;
138 else if (function
& extract_one_to_buffer
) {
139 if (S_ISREG(file_entry
->mode
)) {
140 buffer
= (char *) xmalloc(file_entry
->size
+ 1);
141 fread(buffer
, 1, file_entry
->size
, src_stream
);
142 buffer
[file_entry
->size
] = '\0';
143 archive_offset
+= file_entry
->size
;
147 else if (function
& extract_all_to_fs
) {
150 stat_res
= lstat (full_name
, &oldfile
);
151 if (stat_res
== 0) { /* The file already exists */
152 if ((function
& extract_unconditional
) || (oldfile
.st_mtime
< file_entry
->mtime
)) {
153 if (!S_ISDIR(oldfile
.st_mode
)) {
154 unlink(full_name
); /* Directories might not be empty etc */
157 if ((function
& extract_quiet
) != extract_quiet
) {
158 error_msg("%s not created: newer or same age file exists", file_entry
->name
);
160 seek_sub_file(src_stream
, file_entry
->size
);
164 if (function
& extract_create_leading_dirs
) { /* Create leading directories with default umask */
166 buf
= xstrdup(full_name
);
167 parent
= dirname(buf
);
168 if (make_directory (parent
, -1, FILEUTILS_RECUR
) != 0) {
169 if ((function
& extract_quiet
) != extract_quiet
) {
170 error_msg("couldn't create leading directories");
175 switch(file_entry
->mode
& S_IFMT
) {
177 if (file_entry
->link_name
) { /* Found a cpio hard link */
178 if (link(full_link_name
, full_name
) != 0) {
179 if ((function
& extract_quiet
) != extract_quiet
) {
180 perror_msg("Cannot link from %s to '%s'",
181 file_entry
->name
, file_entry
->link_name
);
185 if ((dst_stream
= wfopen(full_name
, "w")) == NULL
) {
186 seek_sub_file(src_stream
, file_entry
->size
);
189 archive_offset
+= file_entry
->size
;
190 copy_file_chunk(src_stream
, dst_stream
, file_entry
->size
);
196 if (mkdir(full_name
, file_entry
->mode
) < 0) {
197 if ((function
& extract_quiet
) != extract_quiet
) {
198 perror_msg("extract_archive: %s", full_name
);
204 if (symlink(file_entry
->link_name
, full_name
) < 0) {
205 if ((function
& extract_quiet
) != extract_quiet
) {
206 perror_msg("Cannot create symlink from %s to '%s'", file_entry
->name
, file_entry
->link_name
);
215 if (mknod(full_name
, file_entry
->mode
, file_entry
->device
) == -1) {
216 if ((function
& extract_quiet
) != extract_quiet
) {
217 perror_msg("Cannot create node %s", file_entry
->name
);
223 perror_msg("Don't know how to handle %s", full_name
);
227 /* Changing a symlink's properties normally changes the properties of the
228 * file pointed to, so dont try and change the date or mode, lchown does
229 * does the right thing, but isnt available in older versions of libc */
230 if (S_ISLNK(file_entry
->mode
)) {
231 #if (__GLIBC__ > 2) && (__GLIBC_MINOR__ > 1)
232 lchown(full_name
, file_entry
->uid
, file_entry
->gid
);
235 if (function
& extract_preserve_date
) {
236 t
.actime
= file_entry
->mtime
;
237 t
.modtime
= file_entry
->mtime
;
238 utime(full_name
, &t
);
240 chown(full_name
, file_entry
->uid
, file_entry
->gid
);
241 chmod(full_name
, file_entry
->mode
);
244 /* If we arent extracting data we have to skip it,
245 * if data size is 0 then then just do it anyway
246 * (saves testing for it) */
247 seek_sub_file(src_stream
, file_entry
->size
);
250 /* extract_list and extract_verbose_list can be used in conjunction
251 * with one of the above four extraction functions, so do this seperately */
252 if (function
& extract_verbose_list
) {
253 fprintf(out_stream
, "%s %d/%d %8d %s ", mode_string(file_entry
->mode
),
254 file_entry
->uid
, file_entry
->gid
,
255 (int) file_entry
->size
, time_string(file_entry
->mtime
));
257 if ((function
& extract_list
) || (function
& extract_verbose_list
)){
258 /* fputs doesnt add a trailing \n, so use fprintf */
259 fprintf(out_stream
, "%s\n", file_entry
->name
);
264 if ( full_link_name
)
265 free(full_link_name
);
267 return(buffer
); /* Maybe we should say if failed */
272 char *unarchive(FILE *src_stream
, FILE *out_stream
,
273 file_header_t
*(*get_headers
)(FILE *),
274 void (*free_headers
)(file_header_t
*),
275 const int extract_function
, const char *prefix
, const char **extract_names
)
277 file_header_t
*file_entry
;
283 while ((file_entry
= get_headers(src_stream
)) != NULL
) {
285 /* fprintf(stderr, __FUNCTION__ " getting headers\n"); */
286 if (extract_names
!= NULL
) {
287 int found_flag
= FALSE
;
288 for(i
= 0; extract_names
[i
] != 0; i
++) {
289 if (strcmp(extract_names
[i
], file_entry
->name
) == 0) {
294 if (extract_function
& extract_exclude_list
) {
295 if (found_flag
== TRUE
) {
296 extract_flag
= FALSE
;
299 /* If its not found in the include list dont extract it */
300 if (found_flag
== FALSE
) {
301 extract_flag
= FALSE
;
307 if (extract_flag
== TRUE
) {
308 /* fprintf(stderr, __FUNCTION__ " extract?\n"); */
309 buffer
= extract_archive(src_stream
, out_stream
, file_entry
, extract_function
, prefix
);
310 /* fprintf(stderr, __FUNCTION__ " extracted\n"); */
312 /* seek past the data entry */
313 seek_sub_file(src_stream
, file_entry
->size
);
315 free_headers(file_entry
);
317 /*fprintf(stderr, __FUNCTION__ " goin home\n");*/
323 #ifdef L_get_header_ar
324 file_header_t
*get_header_ar(FILE *src_stream
)
326 file_header_t
*typed
;
339 static char *ar_long_names
;
341 if (fread(ar
.raw
, 1, 60, src_stream
) != 60) {
344 archive_offset
+= 60;
345 /* align the headers based on the header magic */
346 if ((ar
.formated
.magic
[0] != '`') || (ar
.formated
.magic
[1] != '\n')) {
347 /* some version of ar, have an extra '\n' after each data entry,
348 * this puts the next header out by 1 */
349 if (ar
.formated
.magic
[1] != '`') {
350 error_msg("Invalid magic");
353 /* read the next char out of what would be the data section,
354 * if its a '\n' then it is a valid header offset by 1*/
356 if (fgetc(src_stream
) != '\n') {
357 error_msg("Invalid magic");
360 /* fix up the header, we started reading 1 byte too early */
361 /* raw_header[60] wont be '\n' as it should, but it doesnt matter */
362 memmove(ar
.raw
, &ar
.raw
[1], 59);
365 typed
= (file_header_t
*) xcalloc(1, sizeof(file_header_t
));
367 typed
->size
= (size_t) atoi(ar
.formated
.size
);
368 /* long filenames have '/' as the first character */
369 if (ar
.formated
.name
[0] == '/') {
370 if (ar
.formated
.name
[1] == '/') {
371 /* If the second char is a '/' then this entries data section
372 * stores long filename for multiple entries, they are stored
373 * in static variable long_names for use in future entries */
374 ar_long_names
= (char *) xrealloc(ar_long_names
, typed
->size
);
375 fread(ar_long_names
, 1, typed
->size
, src_stream
);
376 archive_offset
+= typed
->size
;
377 /* This ar entries data section only contained filenames for other records
378 * they are stored in the static ar_long_names for future reference */
379 return (get_header_ar(src_stream
)); /* Return next header */
380 } else if (ar
.formated
.name
[1] == ' ') {
381 /* This is the index of symbols in the file for compilers */
382 seek_sub_file(src_stream
, typed
->size
);
383 return (get_header_ar(src_stream
)); /* Return next header */
385 /* The number after the '/' indicates the offset in the ar data section
386 (saved in variable long_name) that conatains the real filename */
387 if (!ar_long_names
) {
388 error_msg("Cannot resolve long file name");
391 typed
->name
= xstrdup(ar_long_names
+ atoi(&ar
.formated
.name
[1]));
394 /* short filenames */
395 typed
->name
= xcalloc(1, 16);
396 strncpy(typed
->name
, ar
.formated
.name
, 16);
398 typed
->name
[strcspn(typed
->name
, " /")]='\0';
400 /* convert the rest of the now valid char header to its typed struct */
401 parse_mode(ar
.formated
.mode
, &typed
->mode
);
402 typed
->mtime
= atoi(ar
.formated
.date
);
403 typed
->uid
= atoi(ar
.formated
.uid
);
404 typed
->gid
= atoi(ar
.formated
.gid
);
409 void free_header_ar(file_header_t
*ar_entry
)
411 if (ar_entry
== NULL
) {
415 free(ar_entry
->name
);
416 free(ar_entry
->link_name
);
422 #ifdef L_get_header_tar
423 file_header_t
*get_header_tar(FILE *tar_stream
)
427 unsigned char raw
[512];
429 char name
[100]; /* 0-99 */
430 char mode
[8]; /* 100-107 */
431 char uid
[8]; /* 108-115 */
432 char gid
[8]; /* 116-123 */
433 char size
[12]; /* 124-135 */
434 char mtime
[12]; /* 136-147 */
435 char chksum
[8]; /* 148-155 */
436 char typeflag
; /* 156-156 */
437 char linkname
[100]; /* 157-256 */
438 char magic
[6]; /* 257-262 */
439 char version
[2]; /* 263-264 */
440 char uname
[32]; /* 265-296 */
441 char gname
[32]; /* 297-328 */
442 char devmajor
[8]; /* 329-336 */
443 char devminor
[8]; /* 337-344 */
444 char prefix
[155]; /* 345-499 */
445 char padding
[12]; /* 500-512 */
448 file_header_t
*tar_entry
= NULL
;
452 if (archive_offset
% 512 != 0) {
453 seek_sub_file(tar_stream
, 512 - (archive_offset
% 512));
456 if (fread(tar
.raw
, 1, 512, tar_stream
) != 512) {
457 /* Unfortunatly its common for tar files to have all sorts of
458 * trailing garbage, fail silently */
459 // error_msg("Couldnt read header");
462 archive_offset
+= 512;
464 /* Check header has valid magic, unfortunately some tar files
465 * have empty (0'ed) tar entries at the end, which will
466 * cause this to fail, so fail silently for now
468 if (strncmp(tar
.formated
.magic
, "ustar", 5) != 0) {
469 #ifdef CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY
470 if (strncmp(tar
.formated
.magic
, "\0\0\0\0\0", 5) != 0)
475 /* Do checksum on headers */
476 for (i
= 0; i
< 148 ; i
++) {
480 for (i
= 156; i
< 512 ; i
++) {
483 if (sum
!= strtol(tar
.formated
.chksum
, NULL
, 8)) {
484 if ( strtol(tar
.formated
.chksum
,NULL
,8) != 0 )
485 error_msg("Invalid tar header checksum");
489 /* convert to type'ed variables */
490 tar_entry
= xcalloc(1, sizeof(file_header_t
));
492 #ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
494 tar_entry
->name
= longname
;
498 tar_entry
->name
= linkname
;
503 tar_entry
->name
= xstrndup(tar
.formated
.name
, 100);
505 if (tar
.formated
.prefix
[0]) {
506 char *temp
= tar_entry
->name
;
507 char *prefixTemp
= xstrndup(tar
.formated
.prefix
, 155);
508 tar_entry
->name
= concat_path_file(prefixTemp
, temp
);
514 // tar_entry->name = xstrdup(tar.formated.name);
517 parse_mode(tar.formated.mode, &tar_entry->mode);
519 tar_entry
->mode
= 07777 & strtol(tar
.formated
.mode
, NULL
, 8);
521 tar_entry
->uid
= strtol(tar
.formated
.uid
, NULL
, 8);
522 tar_entry
->gid
= strtol(tar
.formated
.gid
, NULL
, 8);
523 tar_entry
->size
= strtol(tar
.formated
.size
, NULL
, 8);
524 tar_entry
->mtime
= strtol(tar
.formated
.mtime
, NULL
, 8);
525 tar_entry
->link_name
= *tar
.formated
.linkname
!= '\0' ? xstrndup(tar
.formated
.linkname
, 100) : NULL
;
526 tar_entry
->device
= (strtol(tar
.formated
.devmajor
, NULL
, 8) << 8) +
527 strtol(tar
.formated
.devminor
, NULL
, 8);
529 /* Fix mode, used by the old format */
530 switch (tar
.formated
.typeflag
) {
531 /* hard links are detected as regular files with 0 size and a link name */
533 tar_entry
->mode
|= S_IFREG
;
538 # ifdef CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY
539 if (last_char_is(tar_entry
->name
, '/')) {
540 tar_entry
->mode
|= S_IFDIR
;
543 tar_entry
->mode
|= S_IFREG
;
546 tar_entry
->mode
|= S_IFLNK
;
549 tar_entry
->mode
|= S_IFCHR
;
552 tar_entry
->mode
|= S_IFBLK
;
555 tar_entry
->mode
|= S_IFDIR
;
558 tar_entry
->mode
|= S_IFIFO
;
560 # ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
562 longname
= xmalloc(tar_entry
->size
+ 1);
563 if(fread(longname
, tar_entry
->size
, 1, tar_stream
) != 1)
565 longname
[tar_entry
->size
] = '\0';
566 archive_offset
+= tar_entry
->size
;
568 return(get_header_tar(tar_stream
));
571 linkname
= xmalloc(tar_entry
->size
+ 1);
572 if(fread(linkname
, tar_entry
->size
, 1, tar_stream
) != 1)
574 linkname
[tar_entry
->size
] = '\0';
575 archive_offset
+= tar_entry
->size
;
577 tar_entry
->name
= linkname
;
578 return(get_header_tar(tar_stream
));
585 perror_msg("Ignoring GNU extension type %c", tar
.formated
.typeflag
);
588 perror_msg("Unknown typeflag: 0x%x", tar
.formated
.typeflag
);
596 void free_header_tar(file_header_t
*tar_entry
)
598 if (tar_entry
== NULL
) {
602 free(tar_entry
->name
);
603 free(tar_entry
->link_name
);
611 char *deb_extract(const char *package_filename
, FILE *out_stream
,
612 const int extract_function
, const char *prefix
, const char *filename
)
614 FILE *deb_stream
= NULL
;
615 file_header_t
*ar_header
= NULL
;
616 const char **file_list
= NULL
;
617 char *output_buffer
= NULL
;
618 char *ared_file
= NULL
;
621 if (filename
!= NULL
) {
622 file_list
= xmalloc(sizeof(char *) * 2);
623 file_list
[0] = filename
;
627 if (extract_function
& extract_control_tar_gz
) {
628 ared_file
= "control.tar.gz";
630 else if (extract_function
& extract_data_tar_gz
) {
631 ared_file
= "data.tar.gz";
633 fprintf(stderr
, "no file specified to extract -- extract_function=%x\n", extract_function
);
637 /* open the debian package to be worked on */
638 deb_stream
= wfopen(package_filename
, "r");
639 if (deb_stream
== NULL
) {
642 /* set the buffer size */
643 setvbuf(deb_stream
, NULL
, _IOFBF
, 0x8000);
646 fread(ar_magic
, 1, 8, deb_stream
);
647 /*fprintf(stderr, "deb_extract ar_magic=%08x\n", *(long *)ar_magic);*/
648 if (strncmp(ar_magic
,"!<arch>",7) == 0) {
651 while ((ar_header
= get_header_ar(deb_stream
)) != NULL
) {
652 if (strcmp(ared_file
, ar_header
->name
) == 0) {
654 FILE *uncompressed_stream
;
655 /* open a stream of decompressed data */
656 uncompressed_stream
= gz_open(deb_stream
, &gunzip_pid
);
657 if (uncompressed_stream
== NULL
) {
662 output_buffer
= unarchive(uncompressed_stream
, out_stream
, get_header_tar
, free_header_tar
, extract_function
, prefix
, file_list
);
663 fclose(uncompressed_stream
);
664 gz_close(gunzip_pid
);
667 seek_sub_file(deb_stream
, ar_header
->size
);
668 free (ar_header
->name
);
672 } else if (strncmp(ar_magic
, "\037\213", 2) == 0) {
673 /* it's a gz file, let's assume it's an opkg */
674 int unzipped_opkg_pid
;
675 FILE *unzipped_opkg_stream
;
676 file_header_t
*tar_header
;
678 fseek(deb_stream
, 0, SEEK_SET
);
679 unzipped_opkg_stream
= gz_open(deb_stream
, &unzipped_opkg_pid
);
680 if (unzipped_opkg_stream
== NULL
) {
684 /*fprintf(stderr, __FUNCTION__ ": processing opkg %s -- ared_file=%s\n", package_filename, ared_file);*/
685 /* walk through outer tar file to find ared_file */
686 while ((tar_header
= get_header_tar(unzipped_opkg_stream
)) != NULL
) {
688 if (strncmp(tar_header
->name
, "./", 2) == 0)
690 if (strcmp(ared_file
, tar_header
->name
+name_offset
) == 0) {
692 FILE *uncompressed_stream
;
693 /* open a stream of decompressed data */
694 uncompressed_stream
= gz_open(unzipped_opkg_stream
, &gunzip_pid
);
695 if (uncompressed_stream
== NULL
) {
699 /*fprintf(stderr, __FUNCTION__ ":%d: here -- found file\n", __LINE__);*/
700 output_buffer
= unarchive(uncompressed_stream
,
707 /*fprintf(stderr, __FUNCTION__ ":%d: unarchive complete\n", __LINE__);*/
708 free_header_tar(tar_header
);
709 fclose(uncompressed_stream
);
710 gz_close(gunzip_pid
);
713 seek_sub_file(unzipped_opkg_stream
, tar_header
->size
);
714 free_header_tar(tar_header
);
716 fclose(unzipped_opkg_stream
);
717 gz_close(unzipped_opkg_pid
);
718 /*fprintf(stderr, __FUNCTION__ ":%d: done\n", __LINE__);*/
721 error_msg_and_die("invalid magic");
730 return output_buffer
;