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;
85 if (strncmp("./", path, 2) == 0) {
87 if (strlen(path) == 0) {
91 full_name = xmalloc(strlen(prefix) + strlen(path) + 1);
92 strcpy(full_name, prefix);
93 strcat(full_name, path);
95 full_name = file_entry->name;
97 if (function & extract_to_stdout) {
98 if (S_ISREG(file_entry->mode)) {
99 copy_file_chunk(src_stream, out_stream, file_entry->size);
100 archive_offset += file_entry->size;
103 else if (function & extract_one_to_buffer) {
104 if (S_ISREG(file_entry->mode)) {
105 buffer = (char *) xmalloc(file_entry->size + 1);
106 fread(buffer, 1, file_entry->size, src_stream);
107 buffer[file_entry->size] = '\0';
108 archive_offset += file_entry->size;
112 else if (function & extract_all_to_fs) {
115 stat_res = lstat (full_name, &oldfile);
116 if (stat_res == 0) { /* The file already exists */
117 if ((function & extract_unconditional) || (oldfile.st_mtime < file_entry->mtime)) {
118 if (!S_ISDIR(oldfile.st_mode)) {
119 unlink(full_name); /* Directories might not be empty etc */
122 if ((function & extract_quiet) != extract_quiet) {
123 error_msg("%s not created: newer or same age file exists", file_entry->name);
125 seek_sub_file(src_stream, file_entry->size);
129 if (function & extract_create_leading_dirs) { /* Create leading directories with default umask */
130 char *parent = dirname(full_name);
131 if (make_directory (parent, -1, FILEUTILS_RECUR) != 0) {
132 if ((function & extract_quiet) != extract_quiet) {
133 error_msg("couldn't create leading directories");
138 switch(file_entry->mode & S_IFMT) {
140 if (file_entry->link_name) { /* Found a cpio hard link */
141 if (link(file_entry->link_name, full_name) != 0) {
142 if ((function & extract_quiet) != extract_quiet) {
143 perror_msg("Cannot link from %s to '%s'",
144 file_entry->name, file_entry->link_name);
148 if ((dst_stream = wfopen(full_name, "w")) == NULL) {
149 seek_sub_file(src_stream, file_entry->size);
152 archive_offset += file_entry->size;
153 copy_file_chunk(src_stream, dst_stream, file_entry->size);
159 if (mkdir(full_name, file_entry->mode) < 0) {
160 if ((function & extract_quiet) != extract_quiet) {
161 perror_msg("extract_archive: ");
167 if (symlink(file_entry->link_name, full_name) < 0) {
168 if ((function & extract_quiet) != extract_quiet) {
169 perror_msg("Cannot create symlink from %s to '%s'", file_entry->name, file_entry->link_name);
178 if (mknod(full_name, file_entry->mode, file_entry->device) == -1) {
179 if ((function & extract_quiet) != extract_quiet) {
180 perror_msg("Cannot create node %s", file_entry->name);
187 /* Changing a symlink's properties normally changes the properties of the
188 * file pointed to, so dont try and change the date or mode, lchown does
189 * does the right thing, but isnt available in older versions of libc */
190 if (S_ISLNK(file_entry->mode)) {
191 #if (__GLIBC__ > 2) && (__GLIBC_MINOR__ > 1)
192 lchown(full_name, file_entry->uid, file_entry->gid);
195 if (function & extract_preserve_date) {
196 t.actime = file_entry->mtime;
197 t.modtime = file_entry->mtime;
198 utime(full_name, &t);
200 chmod(full_name, file_entry->mode);
201 chown(full_name, file_entry->uid, file_entry->gid);
204 /* If we arent extracting data we have to skip it,
205 * if data size is 0 then then just do it anyway
206 * (saves testing for it) */
207 seek_sub_file(src_stream, file_entry->size);
210 /* extract_list and extract_verbose_list can be used in conjunction
211 * with one of the above four extraction functions, so do this seperately */
212 if (function & extract_verbose_list) {
213 fprintf(out_stream, "%s %d/%d %8d %s ", mode_string(file_entry->mode),
214 file_entry->uid, file_entry->gid,
215 (int) file_entry->size, time_string(file_entry->mtime));
217 if ((function & extract_list) || (function & extract_verbose_list)){
218 /* fputs doesnt add a trailing \n, so use fprintf */
219 fprintf(out_stream, "%s\n", file_entry->name);
224 return(NULL); /* Maybe we should say if failed */
229 char *unarchive(FILE *src_stream, FILE *out_stream, file_header_t *(*get_headers)(FILE *),
230 const int extract_function, const char *prefix, char **extract_names)
232 file_header_t *file_entry;
238 while ((file_entry = get_headers(src_stream)) != NULL) {
240 if (extract_names != NULL) {
241 int found_flag = FALSE;
242 for(i = 0; extract_names[i] != 0; i++) {
243 if (strcmp(extract_names[i], file_entry->name) == 0) {
248 if (extract_function & extract_exclude_list) {
249 if (found_flag == TRUE) {
250 extract_flag = FALSE;
253 /* If its not found in the include list dont extract it */
254 if (found_flag == FALSE) {
255 extract_flag = FALSE;
261 if (extract_flag == TRUE) {
262 buffer = extract_archive(src_stream, out_stream, file_entry, extract_function, prefix);
264 /* seek past the data entry */
265 seek_sub_file(src_stream, file_entry->size);
272 #ifdef L_get_header_ar
273 file_header_t *get_header_ar(FILE *src_stream)
275 file_header_t *typed;
288 static char *ar_long_names;
290 if (fread(ar.raw, 1, 60, src_stream) != 60) {
293 archive_offset += 60;
294 /* align the headers based on the header magic */
295 if ((ar.formated.magic[0] != '`') || (ar.formated.magic[1] != '\n')) {
296 /* some version of ar, have an extra '\n' after each data entry,
297 * this puts the next header out by 1 */
298 if (ar.formated.magic[1] != '`') {
299 error_msg("Invalid magic");
302 /* read the next char out of what would be the data section,
303 * if its a '\n' then it is a valid header offset by 1*/
305 if (fgetc(src_stream) != '\n') {
306 error_msg("Invalid magic");
309 /* fix up the header, we started reading 1 byte too early */
310 /* raw_header[60] wont be '\n' as it should, but it doesnt matter */
311 memmove(ar.raw, &ar.raw[1], 59);
314 typed = (file_header_t *) xcalloc(1, sizeof(file_header_t));
316 typed->size = (size_t) atoi(ar.formated.size);
317 /* long filenames have '/' as the first character */
318 if (ar.formated.name[0] == '/') {
319 if (ar.formated.name[1] == '/') {
320 /* If the second char is a '/' then this entries data section
321 * stores long filename for multiple entries, they are stored
322 * in static variable long_names for use in future entries */
323 ar_long_names = (char *) xrealloc(ar_long_names, typed->size);
324 fread(ar_long_names, 1, typed->size, src_stream);
325 archive_offset += typed->size;
326 /* This ar entries data section only contained filenames for other records
327 * they are stored in the static ar_long_names for future reference */
328 return (get_header_ar(src_stream)); /* Return next header */
329 } else if (ar.formated.name[1] == ' ') {
330 /* This is the index of symbols in the file for compilers */
331 seek_sub_file(src_stream, typed->size);
332 return (get_header_ar(src_stream)); /* Return next header */
334 /* The number after the '/' indicates the offset in the ar data section
335 (saved in variable long_name) that conatains the real filename */
336 if (!ar_long_names) {
337 error_msg("Cannot resolve long file name");
340 typed->name = xstrdup(ar_long_names + atoi(&ar.formated.name[1]));
343 /* short filenames */
344 typed->name = xcalloc(1, 16);
345 strncpy(typed->name, ar.formated.name, 16);
347 typed->name[strcspn(typed->name, " /")]='\0';
349 /* convert the rest of the now valid char header to its typed struct */
350 parse_mode(ar.formated.mode, &typed->mode);
351 typed->mtime = atoi(ar.formated.date);
352 typed->uid = atoi(ar.formated.uid);
353 typed->gid = atoi(ar.formated.gid);
359 #ifdef L_get_header_cpio
361 file_header_t *entry;
363 struct hardlinks *next;
366 file_header_t *get_header_cpio(FILE *src_stream)
368 file_header_t *cpio_entry = NULL;
369 char cpio_header[110];
372 int major, minor, nlink, inode;
373 static struct hardlinks *saved_hardlinks = NULL;
374 static int pending_hardlinks = 0;
376 if (pending_hardlinks) { /* Deal with any pending hardlinks */
377 struct hardlinks *tmp = saved_hardlinks, *oldtmp = NULL;
379 if (tmp->entry->link_name) { /* Found a hardlink ready to be extracted */
380 cpio_entry = tmp->entry;
381 if (oldtmp) oldtmp->next = tmp->next; /* Remove item from linked list */
382 else saved_hardlinks = tmp->next;
389 pending_hardlinks = 0; /* No more pending hardlinks, read next file entry */
392 /* There can be padding before archive header */
393 seek_sub_file(src_stream, (4 - (archive_offset % 4)) % 4);
394 if (fread(cpio_header, 1, 110, src_stream) == 110) {
395 archive_offset += 110;
396 if (strncmp(cpio_header, "07070", 5) != 0) {
397 error_msg("Unsupported format or invalid magic");
400 switch (cpio_header[5]) {
401 case '2': /* "crc" header format */
402 /* Doesnt do the crc check yet */
403 case '1': /* "newc" header format */
404 cpio_entry = (file_header_t *) xcalloc(1, sizeof(file_header_t));
405 sscanf(cpio_header, "%6c%8x%8x%8x%8x%8x%8lx%8lx%16c%8x%8x%8x%8c",
406 dummy, &inode, (unsigned int*)&cpio_entry->mode,
407 (unsigned int*)&cpio_entry->uid, (unsigned int*)&cpio_entry->gid,
408 &nlink, &cpio_entry->mtime, &cpio_entry->size,
409 dummy, &major, &minor, &namesize, dummy);
411 cpio_entry->name = (char *) xcalloc(1, namesize);
412 fread(cpio_entry->name, 1, namesize, src_stream); /* Read in filename */
413 archive_offset += namesize;
414 /* Skip padding before file contents */
415 seek_sub_file(src_stream, (4 - (archive_offset % 4)) % 4);
416 if (strcmp(cpio_entry->name, "TRAILER!!!") == 0) {
417 printf("%d blocks\n", (int) (archive_offset % 512 ? (archive_offset / 512) + 1 : archive_offset / 512)); /* Always round up */
418 if (saved_hardlinks) { /* Bummer - we still have unresolved hardlinks */
419 struct hardlinks *tmp = saved_hardlinks, *oldtmp = NULL;
421 error_msg("%s not created: cannot resolve hardlink", tmp->entry->name);
424 free (oldtmp->entry->name);
425 free (oldtmp->entry);
428 saved_hardlinks = NULL;
429 pending_hardlinks = 0;
434 if (S_ISLNK(cpio_entry->mode)) {
435 cpio_entry->link_name = (char *) xcalloc(1, cpio_entry->size + 1);
436 fread(cpio_entry->link_name, 1, cpio_entry->size, src_stream);
437 archive_offset += cpio_entry->size;
438 cpio_entry->size = 0; /* Stop possiable seeks in future */
440 if (nlink > 1 && !S_ISDIR(cpio_entry->mode)) {
441 if (cpio_entry->size == 0) { /* Put file on a linked list for later */
442 struct hardlinks *new = xmalloc(sizeof(struct hardlinks));
443 new->next = saved_hardlinks;
445 new->entry = cpio_entry;
446 saved_hardlinks = new;
447 return(get_header_cpio(src_stream)); /* Recurse to next file */
448 } else { /* Found the file with data in */
449 struct hardlinks *tmp = saved_hardlinks;
450 pending_hardlinks = 1;
452 if (tmp->inode == inode) {
453 tmp->entry->link_name = xstrdup(cpio_entry->name);
458 if (nlink > 1) error_msg("error resolving hardlink: did you create the archive with GNU cpio 2.0-2.2?");
461 cpio_entry->device = (major << 8) | minor;
464 error_msg("Unsupported format");
467 if (ferror(src_stream) || feof(src_stream)) {
468 perror_msg("Stream error");
476 #ifdef L_get_header_tar
477 file_header_t *get_header_tar(FILE *tar_stream)
480 unsigned char raw[512];
482 char name[100]; /* 0-99 */
483 char mode[8]; /* 100-107 */
484 char uid[8]; /* 108-115 */
485 char gid[8]; /* 116-123 */
486 char size[12]; /* 124-135 */
487 char mtime[12]; /* 136-147 */
488 char chksum[8]; /* 148-155 */
489 char typeflag; /* 156-156 */
490 char linkname[100]; /* 157-256 */
491 char magic[6]; /* 257-262 */
492 char version[2]; /* 263-264 */
493 char uname[32]; /* 265-296 */
494 char gname[32]; /* 297-328 */
495 char devmajor[8]; /* 329-336 */
496 char devminor[8]; /* 337-344 */
497 char prefix[155]; /* 345-499 */
498 char padding[12]; /* 500-512 */
501 file_header_t *tar_entry = NULL;
505 if (archive_offset % 512 != 0) {
506 seek_sub_file(tar_stream, 512 - (archive_offset % 512));
509 if (fread(tar.raw, 1, 512, tar_stream) != 512) {
510 /* Unfortunatly its common for tar files to have all sorts of
511 * trailing garbage, fail silently */
512 // error_msg("Couldnt read header");
515 archive_offset += 512;
517 /* Check header has valid magic, unfortunately some tar files
518 * have empty (0'ed) tar entries at the end, which will
519 * cause this to fail, so fail silently for now
521 if (strncmp(tar.formated.magic, "ustar", 5) != 0) {
525 /* Do checksum on headers */
526 for (i = 0; i < 148 ; i++) {
530 for (i = 156; i < 512 ; i++) {
533 if (sum != strtol(tar.formated.chksum, NULL, 8)) {
534 error_msg("Invalid tar header checksum");
538 /* convert to type'ed variables */
539 tar_entry = xcalloc(1, sizeof(file_header_t));
540 tar_entry->name = xstrdup(tar.formated.name);
542 parse_mode(tar.formated.mode, &tar_entry->mode);
543 tar_entry->uid = strtol(tar.formated.uid, NULL, 8);
544 tar_entry->gid = strtol(tar.formated.gid, NULL, 8);
545 tar_entry->size = strtol(tar.formated.size, NULL, 8);
546 tar_entry->mtime = strtol(tar.formated.mtime, NULL, 8);
547 tar_entry->link_name = strlen(tar.formated.linkname) ?
548 xstrdup(tar.formated.linkname) : NULL;
549 tar_entry->device = (strtol(tar.formated.devmajor, NULL, 8) << 8) +
550 strtol(tar.formated.devminor, NULL, 8);
557 char *deb_extract(const char *package_filename, FILE *out_stream,
558 const int extract_function, const char *prefix, const char *filename)
561 FILE *uncompressed_stream = NULL;
562 file_header_t *ar_header = NULL;
563 char **file_list = NULL;
564 char *output_buffer = NULL;
565 char *ared_file = NULL;
569 if (filename != NULL) {
570 file_list = xmalloc(sizeof(char *) * 2);
571 file_list[0] = xstrdup(filename);
575 if (extract_function & extract_control_tar_gz) {
576 ared_file = xstrdup("control.tar.gz");
578 else if (extract_function & extract_data_tar_gz) {
579 ared_file = xstrdup("data.tar.gz");
582 /* open the debian package to be worked on */
583 deb_stream = wfopen(package_filename, "r");
584 if (deb_stream == NULL) {
587 /* set the buffer size */
588 setvbuf(deb_stream, NULL, _IOFBF, 0x8000);
591 fread(ar_magic, 1, 8, deb_stream);
592 if (strncmp(ar_magic,"!<arch>",7) != 0) {
593 error_msg_and_die("invalid magic");
597 while ((ar_header = get_header_ar(deb_stream)) != NULL) {
598 if (strcmp(ared_file, ar_header->name) == 0) {
599 /* open a stream of decompressed data */
600 uncompressed_stream = gz_open(deb_stream, &gunzip_pid);
602 output_buffer = unarchive(uncompressed_stream, out_stream, get_header_tar, extract_function, prefix, file_list);
604 seek_sub_file(deb_stream, ar_header->size);
606 gz_close(gunzip_pid);
608 fclose(uncompressed_stream);
610 return(output_buffer);