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 extern void seek_sub_file(FILE *src_stream, const int count);
31 extern char *extract_archive(FILE *src_stream, FILE *out_stream, const file_header_t *file_entry,
32 const int function, const char *prefix);
35 #ifdef L_archive_offset
38 extern off_t archive_offset;
41 #ifdef L_seek_sub_file
42 void seek_sub_file(FILE *src_stream, const int count)
45 /* Try to fseek as faster */
46 archive_offset += count;
47 if (fseek(src_stream, count, SEEK_CUR) != 0 && errno == ESPIPE) {
48 for (i = 0; i < count; i++) {
58 #ifdef L_extract_archive
59 /* Extract the data postioned at src_stream to either filesystem, stdout or
60 * buffer depending on the value of 'function' which is defined in libbb.h
62 * prefix doesnt have to be just a directory, it may prefix the filename as well.
64 * e.g. '/var/lib/dpkg/info/dpkg.' will extract all files to the base bath
65 * '/var/lib/dpkg/info/' and all files/dirs created in that dir will have
66 * 'dpkg.' as their prefix
68 * For this reason if prefix does point to a dir then it must end with a
69 * trailing '/' or else the last dir will be assumed to be the file prefix
71 char *extract_archive(FILE *src_stream, FILE *out_stream, const file_header_t *file_entry,
72 const int function, const char *prefix)
74 FILE *dst_stream = NULL;
75 char *full_name = NULL;
79 /* prefix doesnt have to be a proper path it may prepend
80 * the filename as well */
82 /* strip leading '/' in filename to extract as prefix may not be dir */
83 /* Cant use concat_path_file here as prefix might not be a directory */
84 char *path = file_entry->name;
88 full_name = xmalloc(strlen(prefix) + strlen(path) + 1);
89 strcpy(full_name, prefix);
90 strcat(full_name, path);
92 full_name = file_entry->name;
95 if (function & extract_to_stdout) {
96 if (S_ISREG(file_entry->mode)) {
97 copy_file_chunk(src_stream, out_stream, file_entry->size);
98 archive_offset += file_entry->size;
101 else if (function & extract_one_to_buffer) {
102 if (S_ISREG(file_entry->mode)) {
103 buffer = (char *) xmalloc(file_entry->size + 1);
104 fread(buffer, 1, file_entry->size, src_stream);
105 archive_offset += file_entry->size;
109 else if (function & extract_all_to_fs) {
112 stat_res = lstat (full_name, &oldfile);
113 if (stat_res == 0) { /* The file already exists */
114 if ((function & extract_unconditional) || (oldfile.st_mtime < file_entry->mtime)) {
115 if (!S_ISDIR(oldfile.st_mode)) {
116 unlink(full_name); /* Directories might not be empty etc */
119 error_msg("%s not created: newer or same age file exists", file_entry->name);
120 seek_sub_file(src_stream, file_entry->size);
124 if (function & extract_create_leading_dirs) { /* Create leading directories with default umask */
125 char *parent = dirname(full_name);
126 if (make_directory (parent, -1, FILEUTILS_RECUR) != 0) {
127 error_msg("couldn't create leading directories");
131 switch(file_entry->mode & S_IFMT) {
133 if (file_entry->link_name) { /* Found a cpio hard link */
134 if (link(file_entry->link_name, full_name) != 0) {
135 perror_msg("Cannot link from %s to '%s'",
136 file_entry->name, file_entry->link_name);
139 if ((dst_stream = wfopen(full_name, "w")) == NULL) {
140 seek_sub_file(src_stream, file_entry->size);
143 archive_offset += file_entry->size;
144 copy_file_chunk(src_stream, dst_stream, file_entry->size);
150 if (mkdir(full_name, file_entry->mode) < 0) {
151 perror_msg("extract_archive: ");
156 if (symlink(file_entry->link_name, full_name) < 0) {
157 perror_msg("Cannot create symlink from %s to '%s'", file_entry->name, file_entry->link_name);
165 if (mknod(full_name, file_entry->mode, file_entry->device) == -1) {
166 perror_msg("Cannot create node %s", file_entry->name);
172 /* Changing a symlink's properties normally changes the properties of the
173 * file pointed to, so dont try and change the date or mode, lchown does
174 * does the right thing, but isnt available in older versions of libc */
175 if (S_ISLNK(file_entry->mode)) {
176 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
177 lchown(full_name, file_entry->uid, file_entry->gid);
180 if (function & extract_preserve_date) {
181 t.actime = file_entry->mtime;
182 t.modtime = file_entry->mtime;
183 utime(full_name, &t);
185 chmod(full_name, file_entry->mode);
186 chown(full_name, file_entry->uid, file_entry->gid);
189 /* If we arent extracting data we have to skip it,
190 * if data size is 0 then then just do it anyway
191 * (saves testing for it) */
192 seek_sub_file(src_stream, file_entry->size);
195 /* extract_list and extract_verbose_list can be used in conjunction
196 * with one of the above four extraction functions, so do this seperately */
197 if (function & extract_verbose_list) {
198 fprintf(out_stream, "%s %d/%d %8d %s ", mode_string(file_entry->mode),
199 file_entry->uid, file_entry->gid,
200 (int) file_entry->size, time_string(file_entry->mtime));
202 if ((function & extract_list) || (function & extract_verbose_list)){
203 /* fputs doesnt add a trailing \n, so use fprintf */
204 fprintf(out_stream, "%s\n", file_entry->name);
209 return(NULL); /* Maybe we should say if failed */
214 char *unarchive(FILE *src_stream, FILE *out_stream, file_header_t *(*get_headers)(FILE *),
215 const int extract_function, const char *prefix, char **extract_names)
217 file_header_t *file_entry;
222 if (extract_names == NULL) {
226 while ((file_entry = get_headers(src_stream)) != NULL) {
228 if (extract_names[0] != NULL) {
229 for(i = 0; extract_names[i] != 0; i++) {
230 if (strcmp(extract_names[i], file_entry->name) == 0) {
235 /* seek past the data entry */
236 seek_sub_file(src_stream, file_entry->size);
240 buffer = extract_archive(src_stream, out_stream, file_entry, extract_function, prefix);
246 #ifdef L_get_header_ar
247 file_header_t *get_header_ar(FILE *src_stream)
249 file_header_t *typed;
262 static char *ar_long_names;
264 if (fread(ar.raw, 1, 60, src_stream) != 60) {
267 archive_offset += 60;
268 /* align the headers based on the header magic */
269 if ((ar.formated.magic[0] != '`') || (ar.formated.magic[1] != '\n')) {
270 /* some version of ar, have an extra '\n' after each data entry,
271 * this puts the next header out by 1 */
272 if (ar.formated.magic[1] != '`') {
273 error_msg("Invalid magic");
276 /* read the next char out of what would be the data section,
277 * if its a '\n' then it is a valid header offset by 1*/
279 if (fgetc(src_stream) != '\n') {
280 error_msg("Invalid magic");
283 /* fix up the header, we started reading 1 byte too early */
284 /* raw_header[60] wont be '\n' as it should, but it doesnt matter */
285 memmove(ar.raw, &ar.raw[1], 59);
288 typed = (file_header_t *) xcalloc(1, sizeof(file_header_t));
290 typed->size = (size_t) atoi(ar.formated.size);
291 /* long filenames have '/' as the first character */
292 if (ar.formated.name[0] == '/') {
293 if (ar.formated.name[1] == '/') {
294 /* If the second char is a '/' then this entries data section
295 * stores long filename for multiple entries, they are stored
296 * in static variable long_names for use in future entries */
297 ar_long_names = (char *) xrealloc(ar_long_names, typed->size);
298 fread(ar_long_names, 1, typed->size, src_stream);
299 archive_offset += typed->size;
300 /* This ar entries data section only contained filenames for other records
301 * they are stored in the static ar_long_names for future reference */
302 return (get_header_ar(src_stream)); /* Return next header */
303 } else if (ar.formated.name[1] == ' ') {
304 /* This is the index of symbols in the file for compilers */
305 seek_sub_file(src_stream, typed->size);
306 return (get_header_ar(src_stream)); /* Return next header */
308 /* The number after the '/' indicates the offset in the ar data section
309 (saved in variable long_name) that conatains the real filename */
310 if (!ar_long_names) {
311 error_msg("Cannot resolve long file name");
314 typed->name = xstrdup(ar_long_names + atoi(&ar.formated.name[1]));
317 /* short filenames */
318 typed->name = xcalloc(1, 16);
319 strncpy(typed->name, ar.formated.name, 16);
321 typed->name[strcspn(typed->name, " /")]='\0';
323 /* convert the rest of the now valid char header to its typed struct */
324 parse_mode(ar.formated.mode, &typed->mode);
325 typed->mtime = atoi(ar.formated.date);
326 typed->uid = atoi(ar.formated.uid);
327 typed->gid = atoi(ar.formated.gid);
333 #ifdef L_get_header_cpio
335 file_header_t *entry;
337 struct hardlinks *next;
340 file_header_t *get_header_cpio(FILE *src_stream)
342 file_header_t *cpio_entry = NULL;
343 char cpio_header[110];
346 int major, minor, nlink, inode;
347 static struct hardlinks *saved_hardlinks = NULL;
348 static int pending_hardlinks = 0;
350 if (pending_hardlinks) { /* Deal with any pending hardlinks */
351 struct hardlinks *tmp = saved_hardlinks, *oldtmp = NULL;
353 if (tmp->entry->link_name) { /* Found a hardlink ready to be extracted */
354 cpio_entry = tmp->entry;
355 if (oldtmp) oldtmp->next = tmp->next; /* Remove item from linked list */
356 else saved_hardlinks = tmp->next;
363 pending_hardlinks = 0; /* No more pending hardlinks, read next file entry */
366 /* There can be padding before archive header */
367 seek_sub_file(src_stream, (4 - (archive_offset % 4)) % 4);
368 if (fread(cpio_header, 1, 110, src_stream) == 110) {
369 archive_offset += 110;
370 if (strncmp(cpio_header, "07070", 5) != 0) {
371 error_msg("Unsupported format or invalid magic");
374 switch (cpio_header[5]) {
375 case '2': /* "crc" header format */
376 /* Doesnt do the crc check yet */
377 case '1': /* "newc" header format */
378 cpio_entry = (file_header_t *) xcalloc(1, sizeof(file_header_t));
379 sscanf(cpio_header, "%6c%8x%8x%8x%8x%8x%8lx%8lx%16c%8x%8x%8x%8c",
380 dummy, &inode, &cpio_entry->mode, &cpio_entry->uid, &cpio_entry->gid,
381 &nlink, &cpio_entry->mtime, &cpio_entry->size,
382 dummy, &major, &minor, &namesize, dummy);
384 cpio_entry->name = (char *) xcalloc(1, namesize);
385 fread(cpio_entry->name, 1, namesize, src_stream); /* Read in filename */
386 archive_offset += namesize;
387 /* Skip padding before file contents */
388 seek_sub_file(src_stream, (4 - (archive_offset % 4)) % 4);
389 if (strcmp(cpio_entry->name, "TRAILER!!!") == 0) {
390 printf("%d blocks\n", (int) (archive_offset % 512 ? (archive_offset / 512) + 1 : archive_offset / 512)); /* Always round up */
391 if (saved_hardlinks) { /* Bummer - we still have unresolved hardlinks */
392 struct hardlinks *tmp = saved_hardlinks, *oldtmp = NULL;
394 error_msg("%s not created: cannot resolve hardlink", tmp->entry->name);
397 free (oldtmp->entry->name);
398 free (oldtmp->entry);
401 saved_hardlinks = NULL;
402 pending_hardlinks = 0;
407 if (S_ISLNK(cpio_entry->mode)) {
408 cpio_entry->link_name = (char *) xcalloc(1, cpio_entry->size + 1);
409 fread(cpio_entry->link_name, 1, cpio_entry->size, src_stream);
410 archive_offset += cpio_entry->size;
411 cpio_entry->size = 0; /* Stop possiable seeks in future */
413 if (nlink > 1 && !S_ISDIR(cpio_entry->mode)) {
414 if (cpio_entry->size == 0) { /* Put file on a linked list for later */
415 struct hardlinks *new = xmalloc(sizeof(struct hardlinks));
416 new->next = saved_hardlinks;
418 new->entry = cpio_entry;
419 saved_hardlinks = new;
420 return(get_header_cpio(src_stream)); /* Recurse to next file */
421 } else { /* Found the file with data in */
422 struct hardlinks *tmp = saved_hardlinks;
423 pending_hardlinks = 1;
425 if (tmp->inode == inode) {
426 tmp->entry->link_name = xstrdup(cpio_entry->name);
431 if (nlink > 1) error_msg("error resolving hardlink: did you create the archive with GNU cpio 2.0-2.2?");
434 cpio_entry->device = (major << 8) | minor;
437 error_msg("Unsupported format");
440 if (ferror(src_stream) || feof(src_stream)) {
441 perror_msg("Stream error");
449 #ifdef L_get_header_tar
450 file_header_t *get_header_tar(FILE *tar_stream)
453 unsigned char raw[512];
455 char name[100]; /* 0-99 */
456 char mode[8]; /* 100-107 */
457 char uid[8]; /* 108-115 */
458 char gid[8]; /* 116-123 */
459 char size[12]; /* 124-135 */
460 char mtime[12]; /* 136-147 */
461 char chksum[8]; /* 148-155 */
462 char typeflag; /* 156-156 */
463 char linkname[100]; /* 157-256 */
464 char magic[6]; /* 257-262 */
465 char version[2]; /* 263-264 */
466 char uname[32]; /* 265-296 */
467 char gname[32]; /* 297-328 */
468 char devmajor[8]; /* 329-336 */
469 char devminor[8]; /* 337-344 */
470 char prefix[155]; /* 345-499 */
471 char padding[12]; /* 500-512 */
474 file_header_t *tar_entry = NULL;
478 if (archive_offset % 512 != 0) {
479 seek_sub_file(tar_stream, 512 - (archive_offset % 512));
482 if (fread(tar.raw, 1, 512, tar_stream) != 512) {
483 error_msg("Couldnt read header");
486 archive_offset += 512;
488 /* Check header has valid magic, unfortunately some tar files
489 * have empty (0'ed) tar entries at the end, which will
490 * cause this to fail, so fail silently for now
492 if (strncmp(tar.formated.magic, "ustar", 5) != 0) {
496 /* Do checksum on headers */
497 for (i = 0; i < 148 ; i++) {
501 for (i = 156; i < 512 ; i++) {
504 if (sum != strtol(tar.formated.chksum, NULL, 8)) {
505 error_msg("Invalid tar header checksum");
509 /* convert to type'ed variables */
510 tar_entry = xcalloc(1, sizeof(file_header_t));
511 tar_entry->name = xstrdup(tar.formated.name);
513 parse_mode(tar.formated.mode, &tar_entry->mode);
514 tar_entry->uid = strtol(tar.formated.uid, NULL, 8);
515 tar_entry->gid = strtol(tar.formated.gid, NULL, 8);
516 tar_entry->size = strtol(tar.formated.size, NULL, 8);
517 tar_entry->mtime = strtol(tar.formated.mtime, NULL, 8);
518 tar_entry->link_name = strlen(tar.formated.linkname) ?
519 xstrdup(tar.formated.linkname) : NULL;
520 tar_entry->device = (strtol(tar.formated.devmajor, NULL, 8) << 8) +
521 strtol(tar.formated.devminor, NULL, 8);
528 char *deb_extract(const char *package_filename, FILE *out_stream,
529 const int extract_function, const char *prefix, const char *filename)
532 FILE *uncompressed_stream = NULL;
533 file_header_t *ar_header = NULL;
534 char *output_buffer = NULL;
535 char *ared_file = NULL;
540 if (filename == NULL) {
541 file_list = xmalloc(sizeof(char *));
544 file_list = xmalloc(sizeof(char *) * 3);
545 file_list[0] = xstrdup(filename);
549 if (extract_function & extract_control_tar_gz) {
550 ared_file = xstrdup("control.tar.gz");
552 else if (extract_function & extract_data_tar_gz) {
553 ared_file = xstrdup("data.tar.gz");
556 /* open the debian package to be worked on */
557 deb_stream = wfopen(package_filename, "r");
560 fread(ar_magic, 1, 8, deb_stream);
561 if (strncmp(ar_magic,"!<arch>",7) != 0) {
562 error_msg_and_die("invalid magic");
566 while ((ar_header = get_header_ar(deb_stream)) != NULL) {
567 if (strcmp(ared_file, ar_header->name) == 0) {
568 /* open a stream of decompressed data */
569 uncompressed_stream = gz_open(deb_stream, &gunzip_pid);
571 output_buffer = unarchive(uncompressed_stream, out_stream, get_header_tar, extract_function, prefix, file_list);
573 seek_sub_file(deb_stream, ar_header->size);
575 gz_close(gunzip_pid);
577 fclose(uncompressed_stream);
579 return(output_buffer);