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.
32 #define CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY 1
33 #define CONFIG_FEATURE_TAR_GNU_EXTENSIONS
35 #ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
36 static char *longname = NULL;
37 static char *linkname = NULL;
44 seek_by_read(FILE* fd, size_t len)
46 ssize_t cc, total = 0;
50 cc = fread(buf, sizeof(buf[0]),
51 len > SEEK_BUF ? SEEK_BUF : len,
57 if(feof(fd) || ferror(fd))
64 seek_sub_file(FILE *src_stream, const int count)
66 /* Try to fseek as faster */
67 archive_offset += count;
68 seek_by_read(src_stream, count);
73 /* Extract the data postioned at src_stream to either filesystem, stdout or
74 * buffer depending on the value of 'function' which is defined in libbb.h
76 * prefix doesnt have to be just a directory, it may prefix the filename as well.
78 * e.g. '/var/lib/dpkg/info/dpkg.' will extract all files to the base bath
79 * '/var/lib/dpkg/info/' and all files/dirs created in that dir will have
80 * 'dpkg.' as their prefix
82 * For this reason if prefix does point to a dir then it must end with a
83 * trailing '/' or else the last dir will be assumed to be the file prefix
86 extract_archive(FILE *src_stream, FILE *out_stream,
87 const file_header_t *file_entry, const int function,
90 FILE *dst_stream = NULL;
91 char *full_name = NULL;
92 char *full_link_name = NULL;
97 /* prefix doesnt have to be a proper path it may prepend
98 * the filename as well */
100 /* strip leading '/' in filename to extract as prefix may not be dir */
101 /* Cant use concat_path_file here as prefix might not be a directory */
102 char *path = file_entry->name;
103 if (strncmp("./", path, 2) == 0) {
105 if (strlen(path) == 0) {
109 full_name = xmalloc(strlen(prefix) + strlen(path) + 1);
110 strcpy(full_name, prefix);
111 strcat(full_name, path);
112 if ( file_entry->link_name ){
113 full_link_name = xmalloc(strlen(prefix) + strlen(file_entry->link_name) + 1);
114 strcpy(full_link_name, prefix);
115 strcat(full_link_name, file_entry->link_name);
118 full_name = xstrdup(file_entry->name);
119 if ( file_entry->link_name )
120 full_link_name = xstrdup(file_entry->link_name);
124 if (function & extract_to_stream) {
125 if (S_ISREG(file_entry->mode)) {
126 copy_file_chunk(src_stream, out_stream, file_entry->size);
127 archive_offset += file_entry->size;
130 else if (function & extract_one_to_buffer) {
131 if (S_ISREG(file_entry->mode)) {
132 buffer = (char *) xmalloc(file_entry->size + 1);
133 fread(buffer, 1, file_entry->size, src_stream);
134 buffer[file_entry->size] = '\0';
135 archive_offset += file_entry->size;
139 else if (function & extract_all_to_fs) {
142 stat_res = lstat (full_name, &oldfile);
143 if (stat_res == 0) { /* The file already exists */
144 if ((function & extract_unconditional) || (oldfile.st_mtime < file_entry->mtime)) {
145 if (!S_ISDIR(oldfile.st_mode)) {
146 unlink(full_name); /* Directories might not be empty etc */
149 if ((function & extract_quiet) != extract_quiet) {
150 error_msg("%s not created: newer or same age file exists", file_entry->name);
152 seek_sub_file(src_stream, file_entry->size);
156 if (function & extract_create_leading_dirs) { /* Create leading directories with default umask */
158 buf = xstrdup(full_name);
159 parent = dirname(buf);
160 if (make_directory (parent, -1, FILEUTILS_RECUR) != 0) {
161 if ((function & extract_quiet) != extract_quiet) {
162 error_msg("couldn't create leading directories");
167 switch(file_entry->mode & S_IFMT) {
169 if (file_entry->link_name) { /* Found a cpio hard link */
170 if (link(full_link_name, full_name) != 0) {
171 if ((function & extract_quiet) != extract_quiet) {
172 perror_msg("Cannot link from %s to '%s'",
173 file_entry->name, file_entry->link_name);
177 if ((dst_stream = wfopen(full_name, "w")) == NULL) {
178 seek_sub_file(src_stream, file_entry->size);
181 archive_offset += file_entry->size;
182 copy_file_chunk(src_stream, dst_stream, file_entry->size);
188 if (mkdir(full_name, file_entry->mode) < 0) {
189 if ((function & extract_quiet) != extract_quiet) {
190 perror_msg("extract_archive: %s", full_name);
196 if (symlink(file_entry->link_name, full_name) < 0) {
197 if ((function & extract_quiet) != extract_quiet) {
198 perror_msg("Cannot create symlink from %s to '%s'", file_entry->name, file_entry->link_name);
207 if (mknod(full_name, file_entry->mode, file_entry->device) == -1) {
208 if ((function & extract_quiet) != extract_quiet) {
209 perror_msg("Cannot create node %s", file_entry->name);
215 perror_msg("Don't know how to handle %s", full_name);
219 /* Changing a symlink's properties normally changes the properties of the
220 * file pointed to, so dont try and change the date or mode, lchown does
221 * does the right thing, but isnt available in older versions of libc */
222 if (S_ISLNK(file_entry->mode)) {
223 #if (__GLIBC__ > 2) && (__GLIBC_MINOR__ > 1)
224 lchown(full_name, file_entry->uid, file_entry->gid);
227 if (function & extract_preserve_date) {
228 t.actime = file_entry->mtime;
229 t.modtime = file_entry->mtime;
230 utime(full_name, &t);
232 chown(full_name, file_entry->uid, file_entry->gid);
233 chmod(full_name, file_entry->mode);
236 /* If we arent extracting data we have to skip it,
237 * if data size is 0 then then just do it anyway
238 * (saves testing for it) */
239 seek_sub_file(src_stream, file_entry->size);
242 /* extract_list and extract_verbose_list can be used in conjunction
243 * with one of the above four extraction functions, so do this seperately */
244 if (function & extract_verbose_list) {
245 fprintf(out_stream, "%s %d/%d %8d %s ", mode_string(file_entry->mode),
246 file_entry->uid, file_entry->gid,
247 (int) file_entry->size, time_string(file_entry->mtime));
249 if ((function & extract_list) || (function & extract_verbose_list)){
250 /* fputs doesnt add a trailing \n, so use fprintf */
251 fprintf(out_stream, "%s\n", file_entry->name);
256 if ( full_link_name )
257 free(full_link_name);
259 return(buffer); /* Maybe we should say if failed */
263 unarchive(FILE *src_stream, FILE *out_stream,
264 file_header_t *(*get_headers)(FILE *),
265 void (*free_headers)(file_header_t *),
266 const int extract_function,
268 const char **extract_names)
270 file_header_t *file_entry;
276 while ((file_entry = get_headers(src_stream)) != NULL) {
278 /* fprintf(stderr, __FUNCTION__ " getting headers\n"); */
279 if (extract_names != NULL) {
280 int found_flag = FALSE;
281 for(i = 0; extract_names[i] != 0; i++) {
282 if (strcmp(extract_names[i], file_entry->name) == 0) {
287 if (extract_function & extract_exclude_list) {
288 if (found_flag == TRUE) {
289 extract_flag = FALSE;
292 /* If its not found in the include list dont extract it */
293 if (found_flag == FALSE) {
294 extract_flag = FALSE;
300 if (extract_flag == TRUE) {
301 /* fprintf(stderr, __FUNCTION__ " extract?\n"); */
302 buffer = extract_archive(src_stream, out_stream, file_entry, extract_function, prefix);
303 /* fprintf(stderr, __FUNCTION__ " extracted\n"); */
305 /* seek past the data entry */
306 seek_sub_file(src_stream, file_entry->size);
308 free_headers(file_entry);
310 /*fprintf(stderr, __FUNCTION__ " goin home\n");*/
315 static file_header_t *
316 get_header_ar(FILE *src_stream)
318 file_header_t *typed;
331 static char *ar_long_names;
333 if (fread(ar.raw, 1, 60, src_stream) != 60) {
336 archive_offset += 60;
337 /* align the headers based on the header magic */
338 if ((ar.formated.magic[0] != '`') || (ar.formated.magic[1] != '\n')) {
339 /* some version of ar, have an extra '\n' after each data entry,
340 * this puts the next header out by 1 */
341 if (ar.formated.magic[1] != '`') {
342 error_msg("Invalid magic");
345 /* read the next char out of what would be the data section,
346 * if its a '\n' then it is a valid header offset by 1*/
348 if (fgetc(src_stream) != '\n') {
349 error_msg("Invalid magic");
352 /* fix up the header, we started reading 1 byte too early */
353 /* raw_header[60] wont be '\n' as it should, but it doesnt matter */
354 memmove(ar.raw, &ar.raw[1], 59);
357 typed = (file_header_t *) xcalloc(1, sizeof(file_header_t));
359 typed->size = (size_t) atoi(ar.formated.size);
360 /* long filenames have '/' as the first character */
361 if (ar.formated.name[0] == '/') {
362 if (ar.formated.name[1] == '/') {
363 /* If the second char is a '/' then this entries data section
364 * stores long filename for multiple entries, they are stored
365 * in static variable long_names for use in future entries */
366 ar_long_names = (char *) xrealloc(ar_long_names, typed->size);
367 fread(ar_long_names, 1, typed->size, src_stream);
368 archive_offset += typed->size;
369 /* This ar entries data section only contained filenames for other records
370 * they are stored in the static ar_long_names for future reference */
371 return (get_header_ar(src_stream)); /* Return next header */
372 } else if (ar.formated.name[1] == ' ') {
373 /* This is the index of symbols in the file for compilers */
374 seek_sub_file(src_stream, typed->size);
375 return (get_header_ar(src_stream)); /* Return next header */
377 /* The number after the '/' indicates the offset in the ar data section
378 (saved in variable long_name) that conatains the real filename */
379 if (!ar_long_names) {
380 error_msg("Cannot resolve long file name");
383 typed->name = xstrdup(ar_long_names + atoi(&ar.formated.name[1]));
386 /* short filenames */
387 typed->name = xcalloc(1, 16);
388 strncpy(typed->name, ar.formated.name, 16);
390 typed->name[strcspn(typed->name, " /")]='\0';
392 /* convert the rest of the now valid char header to its typed struct */
393 parse_mode(ar.formated.mode, &typed->mode);
394 typed->mtime = atoi(ar.formated.date);
395 typed->uid = atoi(ar.formated.uid);
396 typed->gid = atoi(ar.formated.gid);
402 free_header_ar(file_header_t *ar_entry)
404 if (ar_entry == NULL)
407 free(ar_entry->name);
408 if (ar_entry->link_name)
409 free(ar_entry->link_name);
415 static file_header_t *
416 get_header_tar(FILE *tar_stream)
419 unsigned char raw[512];
421 char name[100]; /* 0-99 */
422 char mode[8]; /* 100-107 */
423 char uid[8]; /* 108-115 */
424 char gid[8]; /* 116-123 */
425 char size[12]; /* 124-135 */
426 char mtime[12]; /* 136-147 */
427 char chksum[8]; /* 148-155 */
428 char typeflag; /* 156-156 */
429 char linkname[100]; /* 157-256 */
430 char magic[6]; /* 257-262 */
431 char version[2]; /* 263-264 */
432 char uname[32]; /* 265-296 */
433 char gname[32]; /* 297-328 */
434 char devmajor[8]; /* 329-336 */
435 char devminor[8]; /* 337-344 */
436 char prefix[155]; /* 345-499 */
437 char padding[12]; /* 500-512 */
440 file_header_t *tar_entry = NULL;
444 if (archive_offset % 512 != 0) {
445 seek_sub_file(tar_stream, 512 - (archive_offset % 512));
448 if (fread(tar.raw, 1, 512, tar_stream) != 512) {
449 /* Unfortunatly its common for tar files to have all sorts of
450 * trailing garbage, fail silently */
451 // error_msg("Couldnt read header");
454 archive_offset += 512;
456 /* Check header has valid magic, unfortunately some tar files
457 * have empty (0'ed) tar entries at the end, which will
458 * cause this to fail, so fail silently for now
460 if (strncmp(tar.formated.magic, "ustar", 5) != 0) {
461 #ifdef CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY
462 if (strncmp(tar.formated.magic, "\0\0\0\0\0", 5) != 0)
467 /* Do checksum on headers */
468 for (i = 0; i < 148 ; i++) {
472 for (i = 156; i < 512 ; i++) {
475 if (sum != strtol(tar.formated.chksum, NULL, 8)) {
476 if ( strtol(tar.formated.chksum,NULL,8) != 0 )
477 error_msg("Invalid tar header checksum");
481 /* convert to type'ed variables */
482 tar_entry = xcalloc(1, sizeof(file_header_t));
484 #ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
486 tar_entry->name = longname;
490 tar_entry->name = linkname;
495 tar_entry->name = xstrndup(tar.formated.name, 100);
497 if (tar.formated.prefix[0]) {
498 char *temp = tar_entry->name;
499 char *prefixTemp = xstrndup(tar.formated.prefix, 155);
500 tar_entry->name = concat_path_file(prefixTemp, temp);
506 // tar_entry->name = xstrdup(tar.formated.name);
509 parse_mode(tar.formated.mode, &tar_entry->mode);
511 tar_entry->mode = 07777 & strtol(tar.formated.mode, NULL, 8);
513 tar_entry->uid = strtol(tar.formated.uid, NULL, 8);
514 tar_entry->gid = strtol(tar.formated.gid, NULL, 8);
515 tar_entry->size = strtol(tar.formated.size, NULL, 8);
516 tar_entry->mtime = strtol(tar.formated.mtime, NULL, 8);
517 tar_entry->link_name = *tar.formated.linkname != '\0' ? xstrndup(tar.formated.linkname, 100) : NULL;
518 tar_entry->device = (strtol(tar.formated.devmajor, NULL, 8) << 8) +
519 strtol(tar.formated.devminor, NULL, 8);
521 /* Fix mode, used by the old format */
522 switch (tar.formated.typeflag) {
523 /* hard links are detected as regular files with 0 size and a link name */
525 tar_entry->mode |= S_IFREG ;
530 # ifdef CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY
531 if (last_char_is(tar_entry->name, '/')) {
532 tar_entry->mode |= S_IFDIR;
535 tar_entry->mode |= S_IFREG;
538 tar_entry->mode |= S_IFLNK;
541 tar_entry->mode |= S_IFCHR;
544 tar_entry->mode |= S_IFBLK;
547 tar_entry->mode |= S_IFDIR;
550 tar_entry->mode |= S_IFIFO;
552 # ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
554 longname = xmalloc(tar_entry->size + 1);
555 if(fread(longname, tar_entry->size, 1, tar_stream) != 1)
557 longname[tar_entry->size] = '\0';
558 archive_offset += tar_entry->size;
560 return(get_header_tar(tar_stream));
563 linkname = xmalloc(tar_entry->size + 1);
564 if(fread(linkname, tar_entry->size, 1, tar_stream) != 1)
566 linkname[tar_entry->size] = '\0';
567 archive_offset += tar_entry->size;
569 tar_entry->name = linkname;
570 return(get_header_tar(tar_stream));
577 perror_msg("Ignoring GNU extension type %c", tar.formated.typeflag);
580 perror_msg("Unknown typeflag: 0x%x", tar.formated.typeflag);
589 free_header_tar(file_header_t *tar_entry)
591 if (tar_entry == NULL)
594 free(tar_entry->name);
595 if (tar_entry->link_name);
596 free(tar_entry->link_name);
602 deb_extract(const char *package_filename, FILE *out_stream,
603 const int extract_function, const char *prefix, const char *filename)
605 FILE *deb_stream = NULL;
606 file_header_t *ar_header = NULL;
607 const char **file_list = NULL;
608 char *output_buffer = NULL;
609 char *ared_file = NULL;
612 if (filename != NULL) {
613 file_list = xmalloc(sizeof(char *) * 2);
614 file_list[0] = filename;
618 if (extract_function & extract_control_tar_gz) {
619 ared_file = "control.tar.gz";
621 else if (extract_function & extract_data_tar_gz) {
622 ared_file = "data.tar.gz";
624 fprintf(stderr, "no file specified to extract -- extract_function=%x\n", extract_function);
628 /* open the debian package to be worked on */
629 deb_stream = wfopen(package_filename, "r");
630 if (deb_stream == NULL) {
633 /* set the buffer size */
634 setvbuf(deb_stream, NULL, _IOFBF, 0x8000);
637 fread(ar_magic, 1, 8, deb_stream);
638 /*fprintf(stderr, "deb_extract ar_magic=%08x\n", *(long *)ar_magic);*/
639 if (strncmp(ar_magic,"!<arch>",7) == 0) {
642 while ((ar_header = get_header_ar(deb_stream)) != NULL) {
643 if (strcmp(ared_file, ar_header->name) == 0) {
645 FILE *uncompressed_stream;
646 /* open a stream of decompressed data */
647 uncompressed_stream = gz_open(deb_stream, &gunzip_pid);
648 if (uncompressed_stream == NULL) {
653 output_buffer = unarchive(uncompressed_stream, out_stream, get_header_tar, free_header_tar, extract_function, prefix, file_list);
654 fclose(uncompressed_stream);
655 gz_close(gunzip_pid);
656 free_header_ar(ar_header);
659 seek_sub_file(deb_stream, ar_header->size);
660 free_header_ar(ar_header);
663 } else if (strncmp(ar_magic, "\037\213", 2) == 0) {
664 /* it's a gz file, let's assume it's an opkg */
665 int unzipped_opkg_pid;
666 FILE *unzipped_opkg_stream;
667 file_header_t *tar_header;
669 fseek(deb_stream, 0, SEEK_SET);
670 unzipped_opkg_stream = gz_open(deb_stream, &unzipped_opkg_pid);
671 if (unzipped_opkg_stream == NULL) {
675 /*fprintf(stderr, __FUNCTION__ ": processing opkg %s -- ared_file=%s\n", package_filename, ared_file);*/
676 /* walk through outer tar file to find ared_file */
677 while ((tar_header = get_header_tar(unzipped_opkg_stream)) != NULL) {
679 if (strncmp(tar_header->name, "./", 2) == 0)
681 if (strcmp(ared_file, tar_header->name+name_offset) == 0) {
683 FILE *uncompressed_stream;
684 /* open a stream of decompressed data */
685 uncompressed_stream = gz_open(unzipped_opkg_stream, &gunzip_pid);
686 if (uncompressed_stream == NULL) {
690 /*fprintf(stderr, __FUNCTION__ ":%d: here -- found file\n", __LINE__);*/
691 output_buffer = unarchive(uncompressed_stream,
698 /*fprintf(stderr, __FUNCTION__ ":%d: unarchive complete\n", __LINE__);*/
699 free_header_tar(tar_header);
700 fclose(uncompressed_stream);
701 gz_close(gunzip_pid);
702 free_header_tar(tar_header);
705 seek_sub_file(unzipped_opkg_stream, tar_header->size);
706 free_header_tar(tar_header);
708 fclose(unzipped_opkg_stream);
709 gz_close(unzipped_opkg_pid);
710 /*fprintf(stderr, __FUNCTION__ ":%d: done\n", __LINE__);*/
713 error_msg_and_die("invalid magic");
722 return output_buffer;