libbb/unarchive.c

   1 /*
   2  *  Copyright (C) 2000 by Glenn McGrath
   3  *  Copyright (C) 2001 by Laurence Anderson
   4  *
   5  *  Based on previous work by busybox developers and others.
   6  *
   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.
  11  *
  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.
  16  *
  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.
  20  */
  21
  22 #include <stdio.h>
  23 #include <errno.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #include <unistd.h>
  27 #include <utime.h>
  28 #include "libbb.h"
  29
  30 typedef struct file_headers_s {
  31         char *name;
  32         char *link_name;
  33         off_t size;
  34         uid_t uid;
  35         gid_t gid;
  36         mode_t mode;
  37         time_t mtime;
  38         dev_t device;
  39 } file_header_t;
  40
  41
  42 extern void seek_sub_file(FILE *src_stream, const int count);
  43 extern char *extract_archive(FILE *src_stream, FILE *out_stream, const file_header_t *file_entry,
  44  const int function, const char *prefix);
  45
  46
  47 #ifdef L_archive_offset
  48 off_t archive_offset;
  49 #else
  50 extern off_t archive_offset;
  51 #endif
  52
  53 #ifdef L_seek_sub_file
  54 void seek_sub_file(FILE *src_stream, const int count)
  55 {
  56         int i;
  57         /* Try to fseek as faster */
  58         archive_offset += count;
  59         if (fseek(src_stream, count, SEEK_CUR) != 0 && errno == ESPIPE) {
  60         for (i = 0; i < count; i++) {
  61                 fgetc(src_stream);
  62                 }
  63         }
  64         return;
  65 }
  66 #endif
  67
  68
  69
  70 #ifdef L_extract_archive
  71 /* Extract the data postioned at src_stream to either filesystem, stdout or
  72  * buffer depending on the value of 'function' which is defined in libbb.h
  73  *
  74  * prefix doesnt have to be just a directory, it may prefix the filename as well.
  75  *
  76  * e.g. '/var/lib/dpkg/info/dpkg.' will extract all files to the base bath
  77  * '/var/lib/dpkg/info/' and all files/dirs created in that dir will have
  78  * 'dpkg.' as their prefix
  79  *
  80  * For this reason if prefix does point to a dir then it must end with a
  81  * trailing '/' or else the last dir will be assumed to be the file prefix
  82  */
  83 char *extract_archive(FILE *src_stream, FILE *out_stream, const file_header_t *file_entry,
  84  const int function, const char *prefix)
  85 {
  86         FILE *dst_stream = NULL;
  87         char *full_name = NULL;
  88         char *buffer = NULL;
  89         struct utimbuf t;
  90
  91         /* prefix doesnt have to be a proper path it may prepend
  92          * the filename as well */
  93         if (prefix != NULL) {
  94                 /* strip leading '/' in filename to extract as prefix may not be dir */
  95                 /* Cant use concat_path_file here as prefix might not be a directory */
  96                 char *path = file_entry->name;
  97                 if (*path == '/') {
  98                         path++;
  99                 }
 100                 full_name = xmalloc(strlen(prefix) + strlen(path) + 1);
 101                 strcpy(full_name, prefix);
 102                 strcat(full_name, path);
 103         } else {
 104                 full_name = file_entry->name;
 105         }
 106
 107         if (function & extract_to_stdout) {
 108                 if (S_ISREG(file_entry->mode)) {
 109                         copy_file_chunk(src_stream, out_stream, file_entry->size);
 110                         archive_offset += file_entry->size;
 111                 }
 112         }
 113         else if (function & extract_one_to_buffer) {
 114                 if (S_ISREG(file_entry->mode)) {
 115                         buffer = (char *) xmalloc(file_entry->size + 1);
 116                         fread(buffer, 1, file_entry->size, src_stream);
 117                         archive_offset += file_entry->size;
 118                         return(buffer);
 119                 }
 120         }
 121         else if (function & extract_all_to_fs) {
 122                 struct stat oldfile;
 123                 int stat_res;
 124                 stat_res = lstat (full_name, &oldfile);
 125                 if (stat_res == 0) { /* The file already exists */
 126                         if ((function & extract_unconditional) || (oldfile.st_mtime < file_entry->mtime)) {
 127                                 if (!S_ISDIR(oldfile.st_mode)) {
 128                                         unlink(full_name); /* Directories might not be empty etc */
 129                                 }
 130                         } else {
 131                                 error_msg("%s not created: newer or same age file exists", file_entry->name);
 132                                         seek_sub_file(src_stream, file_entry->size);
 133                                 return (NULL);
 134                         }
 135                 }
 136                 if (function & extract_create_leading_dirs) { /* Create leading directories with default umask */
 137                         char *parent = dirname(full_name);
 138                         if (make_directory (parent, -1, FILEUTILS_RECUR) != 0) {
 139                                 error_msg("couldn't create leading directories");
 140                         }
 141                         free (parent);
 142                 }
 143                 switch(file_entry->mode & S_IFMT) {
 144                         case S_IFREG:
 145                                 if (file_entry->link_name) { /* Found a cpio hard link */
 146                                         if (link(file_entry->link_name, full_name) != 0) {
 147                                                 perror_msg("Cannot link from %s to '%s'",
 148                                                         file_entry->name, file_entry->link_name);
 149                                         }
 150                                 } else {
 151                                         if ((dst_stream = wfopen(full_name, "w")) == NULL) {
 152                                                 seek_sub_file(src_stream, file_entry->size);
 153                                                 return NULL;
 154                                         }
 155                                         archive_offset += file_entry->size;
 156                                         copy_file_chunk(src_stream, dst_stream, file_entry->size);
 157                                         fclose(dst_stream);
 158                                 }
 159                                 break;
 160                         case S_IFDIR:
 161                                 if (stat_res != 0) {
 162                                         if (mkdir(full_name, file_entry->mode) < 0) {
 163                                                 perror_msg("extract_archive: ");
 164                                         }
 165                                 }
 166                                 break;
 167                         case S_IFLNK:
 168                                 if (symlink(file_entry->link_name, full_name) < 0) {
 169                                         perror_msg("Cannot create symlink from %s to '%s'", file_entry->name, file_entry->link_name);
 170                                         return NULL;
 171                                 }
 172                                 break;
 173                         case S_IFSOCK:
 174                         case S_IFBLK:
 175                         case S_IFCHR:
 176                         case S_IFIFO:
 177                                 if (mknod(full_name, file_entry->mode, file_entry->device) == -1) {
 178                                         perror_msg("Cannot create node %s", file_entry->name);
 179                                         return NULL;
 180                                 }
 181                                 break;
 182                 }
 183
 184                 /* Changing a symlink's properties normally changes the properties of the
 185                  * file pointed to, so dont try and change the date or mode, lchown does
 186                  * does the right thing, but isnt available in older versions of libc */
 187                 if (S_ISLNK(file_entry->mode)) {
 188 #if (__GLIBC__ >= 2) && (__GLIBC_MINOR__ >= 1)
 189                         lchown(full_name, file_entry->uid, file_entry->gid);
 190 #endif
 191                 } else {
 192                         if (function & extract_preserve_date) {
 193                                 t.actime = file_entry->mtime;
 194                                 t.modtime = file_entry->mtime;
 195                                 utime(full_name, &t);
 196                         }
 197                         chmod(full_name, file_entry->mode);
 198                         chown(full_name, file_entry->uid, file_entry->gid);
 199                 }
 200         } else {
 201                 /* If we arent extracting data we have to skip it,
 202                  * if data size is 0 then then just do it anyway
 203                  * (saves testing for it) */
 204                 seek_sub_file(src_stream, file_entry->size);
 205         }
 206
 207         /* extract_list and extract_verbose_list can be used in conjunction
 208          * with one of the above four extraction functions, so do this seperately */
 209         if (function & extract_verbose_list) {
 210                 fprintf(out_stream, "%s %d/%d %8d %s ", mode_string(file_entry->mode),
 211                         file_entry->uid, file_entry->gid,
 212                         (int) file_entry->size, time_string(file_entry->mtime));
 213         }
 214         if ((function & extract_list) || (function & extract_verbose_list)){
 215                 /* fputs doesnt add a trailing \n, so use fprintf */
 216                 fprintf(out_stream, "%s\n", file_entry->name);
 217         }
 218
 219         free(full_name);
 220
 221         return(NULL); /* Maybe we should say if failed */
 222 }
 223 #endif
 224
 225 #ifdef L_unarchive
 226 char *unarchive(FILE *src_stream, void *(*get_headers)(FILE *),
 227         const int extract_function, const char *prefix, char **extract_names)
 228 {
 229         file_header_t *file_entry;
 230         int found;
 231         int i;
 232         char *buffer = NULL;
 233
 234         archive_offset = 0;
 235         while ((file_entry = (file_header_t *) get_headers(src_stream)) != NULL) {
 236                 found = FALSE;
 237                 if (extract_names[0] != NULL) {
 238                         for(i = 0; extract_names[i] != 0; i++) {
 239                                 if (strcmp(extract_names[i], file_entry->name) == 0) {
 240                                         found = TRUE;
 241                                 }
 242                         }
 243                         if (!found) {
 244                                 /* seek past the data entry */
 245                                 seek_sub_file(src_stream, file_entry->size);
 246                                 continue;
 247                         }
 248                 }
 249                 buffer = extract_archive(src_stream, stdout, file_entry, extract_function, prefix);
 250         }
 251         return(buffer);
 252 }
 253 #endif
 254
 255 #ifdef L_get_header_ar
 256 void *get_header_ar(FILE *src_stream)
 257 {
 258         file_header_t *typed;
 259         union {
 260                 char raw[60];
 261                 struct {
 262                         char name[16];
 263                         char date[12];
 264                         char uid[6];
 265                         char gid[6];
 266                         char mode[8];
 267                         char size[10];
 268                         char magic[2];
 269                 } formated;
 270         } ar;
 271         static char *ar_long_names;
 272
 273         if (fread(ar.raw, 1, 60, src_stream) != 60) {
 274                 free (ar_long_names);
 275                 return(NULL);
 276         }
 277         archive_offset += 60;
 278         /* align the headers based on the header magic */
 279         if ((ar.formated.magic[0] != '`') || (ar.formated.magic[1] != '\n')) {
 280                 /* some version of ar, have an extra '\n' after each data entry,
 281                  * this puts the next header out by 1 */
 282                 if (ar.formated.magic[1] != '`') {
 283                         error_msg("Invalid magic");
 284                         return(NULL);
 285                 }
 286                 /* read the next char out of what would be the data section,
 287                  * if its a '\n' then it is a valid header offset by 1*/
 288                 archive_offset++;
 289                 if (fgetc(src_stream) != '\n') {
 290                         error_msg("Invalid magic");
 291                         return(NULL);
 292                 }
 293                 /* fix up the header, we started reading 1 byte too early */
 294                 /* raw_header[60] wont be '\n' as it should, but it doesnt matter */
 295                 memmove(ar.raw, &ar.raw[1], 59);
 296         }
 297
 298         typed = (file_header_t *) xcalloc(1, sizeof(file_header_t));
 299
 300         typed->size = (size_t) atoi(ar.formated.size);
 301         /* long filenames have '/' as the first character */
 302         if (ar.formated.name[0] == '/') {
 303                 if (ar.formated.name[1] == '/') {
 304                         /* If the second char is a '/' then this entries data section
 305                          * stores long filename for multiple entries, they are stored
 306                          * in static variable long_names for use in future entries */
 307                         ar_long_names = (char *) xrealloc(ar_long_names, typed->size);
 308                         fread(ar_long_names, 1, typed->size, src_stream);
 309                         archive_offset += typed->size;
 310                         /* This ar entries data section only contained filenames for other records
 311                          * they are stored in the static ar_long_names for future reference */
 312                         return (get_header_ar(src_stream)); /* Return next header */
 313                 } else if (ar.formated.name[1] == ' ') {
 314                         /* This is the index of symbols in the file for compilers */
 315                         seek_sub_file(src_stream, typed->size);
 316                         return (get_header_ar(src_stream)); /* Return next header */
 317                 } else {
 318                         /* The number after the '/' indicates the offset in the ar data section
 319                         (saved in variable long_name) that conatains the real filename */
 320                         if (!ar_long_names) {
 321                                 error_msg("Cannot resolve long file name");
 322                                 return (NULL);
 323                         }
 324                         typed->name = xstrdup(ar_long_names + atoi(&ar.formated.name[1]));
 325                 }
 326         } else {
 327                 /* short filenames */
 328                 typed->name = xcalloc(1, 16);
 329                 strncpy(typed->name, ar.formated.name, 16);
 330         }
 331         typed->name[strcspn(typed->name, " /")]='\0';
 332
 333         /* convert the rest of the now valid char header to its typed struct */
 334         parse_mode(ar.formated.mode, &typed->mode);
 335         typed->mtime = atoi(ar.formated.date);
 336         typed->uid = atoi(ar.formated.uid);
 337         typed->gid = atoi(ar.formated.gid);
 338
 339         return(typed);
 340 }
 341 #endif
 342
 343 #ifdef L_get_header_cpio
 344 struct hardlinks {
 345         file_header_t *entry;
 346         int inode;
 347         struct hardlinks *next;
 348 };
 349
 350 void *get_header_cpio(FILE *src_stream)
 351 {
 352         file_header_t *cpio_entry = NULL;
 353         char cpio_header[110];
 354         int namesize;
 355         char dummy[16];
 356         int major, minor, nlink, inode;
 357         static struct hardlinks *saved_hardlinks = NULL;
 358         static int pending_hardlinks = 0;
 359
 360         if (pending_hardlinks) { /* Deal with any pending hardlinks */
 361                 struct hardlinks *tmp = saved_hardlinks, *oldtmp = NULL;
 362                 while (tmp) {
 363                         if (tmp->entry->link_name) { /* Found a hardlink ready to be extracted */
 364                                 cpio_entry = tmp->entry;
 365                                 if (oldtmp) oldtmp->next = tmp->next; /* Remove item from linked list */
 366                                 else saved_hardlinks = tmp->next;
 367                                 free(tmp);
 368                                 return (cpio_entry);
 369                         }
 370                         oldtmp = tmp;
 371                         tmp = tmp->next;
 372                 }
 373                 pending_hardlinks = 0; /* No more pending hardlinks, read next file entry */
 374         }
 375
 376         /* There can be padding before archive header */
 377         seek_sub_file(src_stream, (4 - (archive_offset % 4)) % 4);
 378         if (fread(cpio_header, 1, 110, src_stream) == 110) {
 379                 archive_offset += 110;
 380                 if (strncmp(cpio_header, "07070", 5) != 0) {
 381                         error_msg("Unsupported format or invalid magic");
 382                         return(NULL);
 383                 }
 384                 switch (cpio_header[5]) {
 385                         case '2': /* "crc" header format */
 386                                 /* Doesnt do the crc check yet */
 387                         case '1': /* "newc" header format */
 388                                 cpio_entry = (file_header_t *) xcalloc(1, sizeof(file_header_t));
 389                                 sscanf(cpio_header, "%6c%8x%8x%8x%8x%8x%8lx%8lx%16c%8x%8x%8x%8c",
 390                                         dummy, &inode, &cpio_entry->mode, &cpio_entry->uid, &cpio_entry->gid,
 391                                         &nlink, &cpio_entry->mtime, &cpio_entry->size,
 392                                         dummy, &major, &minor, &namesize, dummy);
 393
 394                                 cpio_entry->name = (char *) xcalloc(1, namesize);
 395                                 fread(cpio_entry->name, 1, namesize, src_stream); /* Read in filename */
 396                                 archive_offset += namesize;
 397                                 /* Skip padding before file contents */
 398                                 seek_sub_file(src_stream, (4 - (archive_offset % 4)) % 4);
 399                                 if (strcmp(cpio_entry->name, "TRAILER!!!") == 0) {
 400                                         printf("%d blocks\n", (int) (archive_offset % 512 ? (archive_offset / 512) + 1 : archive_offset / 512)); /* Always round up */
 401                                         if (saved_hardlinks) { /* Bummer - we still have unresolved hardlinks */
 402                                                 struct hardlinks *tmp = saved_hardlinks, *oldtmp = NULL;
 403                                                 while (tmp) {
 404                                                         error_msg("%s not created: cannot resolve hardlink", tmp->entry->name);
 405                                                         oldtmp = tmp;
 406                                                         tmp = tmp->next;
 407                                                         free (oldtmp->entry->name);
 408                                                         free (oldtmp->entry);
 409                                                         free (oldtmp);
 410                                                 }
 411                                                 saved_hardlinks = NULL;
 412                                                 pending_hardlinks = 0;
 413                                         }
 414                                         return(NULL);
 415                                 }
 416
 417                                 if (S_ISLNK(cpio_entry->mode)) {
 418                                         cpio_entry->link_name = (char *) xcalloc(1, cpio_entry->size + 1);
 419                                         fread(cpio_entry->link_name, 1, cpio_entry->size, src_stream);
 420                                         archive_offset += cpio_entry->size;
 421                                         cpio_entry->size = 0; /* Stop possiable seeks in future */
 422                                 }
 423                                 if (nlink > 1 && !S_ISDIR(cpio_entry->mode)) {
 424                                         if (cpio_entry->size == 0) { /* Put file on a linked list for later */
 425                                                 struct hardlinks *new = xmalloc(sizeof(struct hardlinks));
 426                                                 new->next = saved_hardlinks;
 427                                                 new->inode = inode;
 428                                                 new->entry = cpio_entry;
 429                                                 saved_hardlinks = new;
 430                                         return(get_header_cpio(src_stream)); /* Recurse to next file */
 431                                         } else { /* Found the file with data in */
 432                                                 struct hardlinks *tmp = saved_hardlinks;
 433                                                 pending_hardlinks = 1;
 434                                                 while (tmp) {
 435                                                         if (tmp->inode == inode) {
 436                                                                 tmp->entry->link_name = xstrdup(cpio_entry->name);
 437                                                                 nlink--;
 438                                                         }
 439                                                         tmp = tmp->next;
 440                                                 }
 441                                                 if (nlink > 1) error_msg("error resolving hardlink: did you create the archive with GNU cpio 2.0-2.2?");
 442                                         }
 443                                 }
 444                                 cpio_entry->device = (major << 8) | minor;
 445                                 break;
 446                         default:
 447                                 error_msg("Unsupported format");
 448                                 return(NULL);
 449                 }
 450                 if (ferror(src_stream) || feof(src_stream)) {
 451                         perror_msg("Stream error");
 452                         return(NULL);
 453                 }
 454         }
 455         return(cpio_entry);
 456 }
 457 #endif
 458
 459 #ifdef L_get_header_tar
 460 void *get_header_tar(FILE *tar_stream)
 461 {
 462         union {
 463                 unsigned char raw[512];
 464                 struct {
 465                         char name[100];         /*   0-99 */
 466                         char mode[8];           /* 100-107 */
 467                         char uid[8];            /* 108-115 */
 468                         char gid[8];            /* 116-123 */
 469                         char size[12];          /* 124-135 */
 470                         char mtime[12];         /* 136-147 */
 471                         char chksum[8];         /* 148-155 */
 472                         char typeflag;          /* 156-156 */
 473                         char linkname[100];     /* 157-256 */
 474                         char magic[6];          /* 257-262 */
 475                         char version[2];        /* 263-264 */
 476                         char uname[32];         /* 265-296 */
 477                         char gname[32];         /* 297-328 */
 478                         char devmajor[8];       /* 329-336 */
 479                         char devminor[8];       /* 337-344 */
 480                         char prefix[155];       /* 345-499 */
 481                         char padding[12];       /* 500-512 */
 482                 } formated;
 483         } tar;
 484         file_header_t *tar_entry = NULL;
 485         long i;
 486         long sum = 0;
 487
 488         if (archive_offset % 512 != 0) {
 489                 seek_sub_file(tar_stream, 512 - (archive_offset % 512));
 490         }
 491
 492         if (fread(tar.raw, 1, 512, tar_stream) != 512) {
 493                 error_msg("Couldnt read header");
 494                 return(NULL);
 495         }
 496         archive_offset += 512;
 497
 498         /* Check header has valid magic, unfortunately some tar files
 499          * have empty (0'ed) tar entries at the end, which will
 500          * cause this to fail, so fail silently for now
 501          */
 502         if (strncmp(tar.formated.magic, "ustar", 5) != 0) {
 503                 return(NULL);
 504         }
 505
 506         /* Do checksum on headers */
 507         for (i =  0; i < 148 ; i++) {
 508                 sum += tar.raw[i];
 509         }
 510         sum += ' ' * 8;
 511         for (i =  156; i < 512 ; i++) {
 512                 sum += tar.raw[i];
 513         }
 514         if (sum != strtol(tar.formated.chksum, NULL, 8)) {
 515                 error_msg("Invalid tar header checksum");
 516                 return(NULL);
 517         }
 518
 519         /* convert to type'ed variables */
 520         tar_entry = xcalloc(1, sizeof(file_header_t));
 521         tar_entry->name = xstrdup(tar.formated.name);
 522
 523         parse_mode(tar.formated.mode, &tar_entry->mode);
 524         tar_entry->uid   = strtol(tar.formated.uid, NULL, 8);
 525         tar_entry->gid   = strtol(tar.formated.gid, NULL, 8);
 526         tar_entry->size  = strtol(tar.formated.size, NULL, 8);
 527         tar_entry->mtime = strtol(tar.formated.mtime, NULL, 8);
 528         tar_entry->link_name  = strlen(tar.formated.linkname) ? xstrdup(tar.formated.linkname) : NULL;
 529         tar_entry->device = (strtol(tar.formated.devmajor, NULL, 8) << 8) +
 530                 strtol(tar.formated.devminor, NULL, 8);
 531
 532         return(tar_entry);
 533 }
 534 #endif
 535
 536 #ifdef L_deb_extract
 537 char *deb_extract(const char *package_filename, FILE *out_stream, const int extract_function,
 538         const char *prefix, const char *filename)
 539 {
 540         FILE *deb_stream;
 541         FILE *uncompressed_stream = NULL;
 542         file_header_t *ar_header = NULL;
 543         char *output_buffer = NULL;
 544         char *ared_file = NULL;
 545         char ar_magic[8];
 546         char **file_list;
 547         int gunzip_pid;
 548
 549         file_list = malloc(sizeof(char *));
 550         file_list[0] = xstrdup(filename);
 551         file_list[1] = NULL;
 552
 553         if (extract_function & extract_control_tar_gz) {
 554                 ared_file = xstrdup("control.tar.gz");
 555         }
 556         else if (extract_function & extract_data_tar_gz) {
 557                 ared_file = xstrdup("data.tar.gz");
 558         }
 559
 560         /* open the debian package to be worked on */
 561         deb_stream = wfopen(package_filename, "r");
 562
 563         /* check ar magic */
 564         fread(ar_magic, 1, 8, deb_stream);
 565         if (strncmp(ar_magic,"!<arch>",7) != 0) {
 566                 error_msg_and_die("invalid magic");
 567         }
 568         archive_offset = 8;
 569
 570         while ((ar_header = get_header_ar(deb_stream)) != NULL) {
 571                 if (strcmp(ared_file, ar_header->name) == 0) {
 572                         /* open a stream of decompressed data */
 573                         uncompressed_stream = gz_open(deb_stream, &gunzip_pid);
 574                         archive_offset = 0;
 575                         output_buffer = unarchive(uncompressed_stream, get_header_tar, extract_function, prefix, file_list);
 576                 }
 577                 seek_sub_file(deb_stream, ar_header->size);
 578         }
 579         gz_close(gunzip_pid);
 580         fclose(deb_stream);
 581         fclose(uncompressed_stream);
 582         free(ared_file);
 583         return(output_buffer);
 584 }
 585 #endif