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 <libgen.h>
  29
  30 #include "libbb.h"
  31 #include "gzip.h"
  32
  33 #define CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY 1
  34 #define CONFIG_FEATURE_TAR_GNU_EXTENSIONS
  35
  36 #ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
  37 static char *longname = NULL;
  38 static char *linkname = NULL;
  39 #endif
  40
  41 off_t archive_offset;
  42
  43 static ssize_t
  44 seek_forward(struct gzip_handle *zh, ssize_t len)
  45 {
  46         ssize_t slen = gzip_seek(zh, len);
  47
  48         if (slen == len)
  49                 archive_offset += len;
  50
  51         return slen;
  52 }
  53
  54
  55 /* Extract the data postioned at src_stream to either filesystem, stdout or
  56  * buffer depending on the value of 'function' which is defined in libbb.h
  57  *
  58  * prefix doesnt have to be just a directory, it may prefix the filename as well.
  59  *
  60  * e.g. '/var/lib/dpkg/info/dpkg.' will extract all files to the base bath
  61  * '/var/lib/dpkg/info/' and all files/dirs created in that dir will have
  62  * 'dpkg.' as their prefix
  63  *
  64  * For this reason if prefix does point to a dir then it must end with a
  65  * trailing '/' or else the last dir will be assumed to be the file prefix
  66  */
  67 static char *
  68 extract_archive(struct gzip_handle *src_stream, FILE *out_stream,
  69                 const file_header_t *file_entry, const int function,
  70                 const char *prefix,
  71                 int *err)
  72 {
  73         FILE *dst_stream = NULL;
  74         char *full_name = NULL;
  75         char *full_link_name = NULL;
  76         char *buffer = NULL;
  77         struct utimbuf t;
  78
  79         *err = 0;
  80
  81         /* prefix doesnt have to be a proper path it may prepend
  82          * the filename as well */
  83         if (prefix != NULL) {
  84                 /* strip leading '/' in filename to extract as prefix may not be dir */
  85                 /* Cant use concat_path_file here as prefix might not be a directory */
  86                 char *path = file_entry->name;
  87                 if (strncmp("./", path, 2) == 0) {
  88                         path += 2;
  89                         if (strlen(path) == 0)
  90                                 /* Do nothing, current dir already exists. */
  91                                 return NULL;
  92                 }
  93                 full_name = xmalloc(strlen(prefix) + strlen(path) + 1);
  94                 strcpy(full_name, prefix);
  95                 strcat(full_name, path);
  96                 if ( file_entry->link_name ){
  97                    full_link_name = xmalloc(strlen(prefix) + strlen(file_entry->link_name) + 1);
  98                    strcpy(full_link_name, prefix);
  99                    strcat(full_link_name, file_entry->link_name);
 100                 }
 101         } else {
 102                 full_name = xstrdup(file_entry->name);
 103                 if ( file_entry->link_name )
 104                    full_link_name = xstrdup(file_entry->link_name);
 105         }
 106
 107
 108         if (function & extract_to_stream) {
 109                 if (S_ISREG(file_entry->mode)) {
 110                         *err = gzip_copy(src_stream, out_stream, file_entry->size);
 111                         archive_offset += file_entry->size;
 112                 }
 113         }
 114         else if (function & extract_one_to_buffer) {
 115                 if (S_ISREG(file_entry->mode)) {
 116                         buffer = (char *) xmalloc(file_entry->size + 1);
 117                         gzip_read(src_stream, buffer, file_entry->size);
 118                         buffer[file_entry->size] = '\0';
 119                         archive_offset += file_entry->size;
 120                         goto cleanup;
 121                 }
 122         }
 123         else if (function & extract_all_to_fs) {
 124                 struct stat oldfile;
 125                 int stat_res;
 126                 stat_res = lstat (full_name, &oldfile);
 127                 if (stat_res == 0) { /* The file already exists */
 128                         if ((function & extract_unconditional) || (oldfile.st_mtime < file_entry->mtime)) {
 129                                 if (!S_ISDIR(oldfile.st_mode)) {
 130                                         unlink(full_name); /* Directories might not be empty etc */
 131                                 }
 132                         } else {
 133                                 if ((function & extract_quiet) != extract_quiet) {
 134                                         *err = -1;
 135                                         error_msg("%s not created: newer or same age file exists", file_entry->name);
 136                                 }
 137                                 seek_forward(src_stream, file_entry->size);
 138                                 goto cleanup;
 139                         }
 140                 }
 141                 if (function & extract_create_leading_dirs) { /* Create leading directories with default umask */
 142                         char *buf, *parent;
 143                         buf = xstrdup(full_name);
 144                         parent = dirname(buf);
 145                         if (make_directory (parent, -1, FILEUTILS_RECUR) != 0) {
 146                                 if ((function & extract_quiet) != extract_quiet) {
 147                                         *err = -1;
 148                                         error_msg("couldn't create leading directories");
 149                                 }
 150                         }
 151                         free (buf);
 152                 }
 153                 switch(file_entry->mode & S_IFMT) {
 154                         case S_IFREG:
 155                                 if (file_entry->link_name) { /* Found a cpio hard link */
 156                                         if (link(full_link_name, full_name) != 0) {
 157                                                 if ((function & extract_quiet) != extract_quiet) {
 158                                                         *err = -1;
 159                                                         perror_msg("Cannot link from %s to '%s'",
 160                                                                 file_entry->name, file_entry->link_name);
 161                                                 }
 162                                         }
 163                                 } else {
 164                                         if ((dst_stream = wfopen(full_name, "w")) == NULL) {
 165                                                 *err = -1;
 166                                                 seek_forward(src_stream, file_entry->size);
 167                                                 goto cleanup;
 168                                         }
 169                                         archive_offset += file_entry->size;
 170                                         *err = gzip_copy(src_stream, dst_stream, file_entry->size);
 171                                         fclose(dst_stream);
 172                                 }
 173                                 break;
 174                         case S_IFDIR:
 175                                 if (stat_res != 0) {
 176                                         if (mkdir(full_name, file_entry->mode) < 0) {
 177                                                 if ((function & extract_quiet) != extract_quiet) {
 178                                                         *err = -1;
 179                                                         perror_msg("Cannot make dir %s", full_name);
 180                                                 }
 181                                         }
 182                                 }
 183                                 break;
 184                         case S_IFLNK:
 185                                 if (symlink(file_entry->link_name, full_name) < 0) {
 186                                         if ((function & extract_quiet) != extract_quiet) {
 187                                                 *err = -1;
 188                                                 perror_msg("Cannot create symlink from %s to '%s'", file_entry->name, file_entry->link_name);
 189                                         }
 190                                         goto cleanup;
 191                                 }
 192                                 break;
 193                         case S_IFSOCK:
 194                         case S_IFBLK:
 195                         case S_IFCHR:
 196                         case S_IFIFO:
 197                                 if (mknod(full_name, file_entry->mode, file_entry->device) == -1) {
 198                                         if ((function & extract_quiet) != extract_quiet) {
 199                                                 *err = -1;
 200                                                 perror_msg("Cannot create node %s", file_entry->name);
 201                                         }
 202                                         goto cleanup;
 203                                 }
 204                                 break;
 205                          default:
 206                                 *err = -1;
 207                                 perror_msg("Don't know how to handle %s", full_name);
 208
 209                 }
 210
 211                 /* Changing a symlink's properties normally changes the properties of the
 212                  * file pointed to, so dont try and change the date or mode, lchown does
 213                  * does the right thing, but isnt available in older versions of libc */
 214                 if (S_ISLNK(file_entry->mode)) {
 215 #if (__GLIBC__ > 2) && (__GLIBC_MINOR__ > 1)
 216                         lchown(full_name, file_entry->uid, file_entry->gid);
 217 #endif
 218                 } else {
 219                         if (function & extract_preserve_date) {
 220                                 t.actime = file_entry->mtime;
 221                                 t.modtime = file_entry->mtime;
 222                                 utime(full_name, &t);
 223                         }
 224                         chown(full_name, file_entry->uid, file_entry->gid);
 225                         chmod(full_name, file_entry->mode);
 226                 }
 227         } else {
 228                 /* If we arent extracting data we have to skip it,
 229                  * if data size is 0 then then just do it anyway
 230                  * (saves testing for it) */
 231                 seek_forward(src_stream, file_entry->size);
 232         }
 233
 234         /* extract_list and extract_verbose_list can be used in conjunction
 235          * with one of the above four extraction functions, so do this seperately */
 236         if (function & extract_verbose_list) {
 237                 fprintf(out_stream, "%s %d/%d %8d %s ", mode_string(file_entry->mode),
 238                         file_entry->uid, file_entry->gid,
 239                         (int) file_entry->size, time_string(file_entry->mtime));
 240         }
 241         if ((function & extract_list) || (function & extract_verbose_list)){
 242                 /* fputs doesnt add a trailing \n, so use fprintf */
 243                 fprintf(out_stream, "%s\n", file_entry->name);
 244         }
 245
 246 cleanup:
 247         free(full_name);
 248         if ( full_link_name )
 249             free(full_link_name);
 250
 251         return buffer;
 252 }
 253
 254 static char *
 255 unarchive(struct gzip_handle *src_stream, FILE *out_stream,
 256                 file_header_t *(*get_headers)(struct gzip_handle *),
 257                 void (*free_headers)(file_header_t *),
 258                 const int extract_function,
 259                 const char *prefix,
 260                 const char **extract_names,
 261                 int *err)
 262 {
 263         file_header_t *file_entry;
 264         int extract_flag;
 265         int i;
 266         char *buffer = NULL;
 267
 268         *err = 0;
 269
 270         archive_offset = 0;
 271         while ((file_entry = get_headers(src_stream)) != NULL) {
 272                 extract_flag = TRUE;
 273
 274                 if (extract_names != NULL) {
 275                         int found_flag = FALSE;
 276                         char *p = file_entry->name;
 277
 278                         if (p[0] == '.' && p[1] == '/')
 279                                 p += 2;
 280
 281                         for(i = 0; extract_names[i] != 0; i++) {
 282                                 if (strcmp(extract_names[i], p) == 0) {
 283                                         found_flag = TRUE;
 284                                         break;
 285                                 }
 286                         }
 287                         if (extract_function & extract_exclude_list) {
 288                                 if (found_flag == TRUE) {
 289                                         extract_flag = FALSE;
 290                                 }
 291                         } else {
 292                                 /* If its not found in the include list dont extract it */
 293                                 if (found_flag == FALSE) {
 294                                         extract_flag = FALSE;
 295                                 }
 296                         }
 297                 }
 298
 299                 if (extract_flag == TRUE) {
 300                         buffer = extract_archive(src_stream, out_stream,
 301                                         file_entry, extract_function,
 302                                         prefix, err);
 303                         *err = 0; /* XXX: ignore extraction errors */
 304                         if (*err) {
 305                                 free_headers(file_entry);
 306                                 break;
 307                         }
 308                 } else {
 309                         /* seek past the data entry */
 310                         seek_forward(src_stream, file_entry->size);
 311                 }
 312                 free_headers(file_entry);
 313         }
 314
 315         return buffer;
 316 }
 317
 318
 319 static file_header_t *
 320 get_header_tar(struct gzip_handle *tar_stream)
 321 {
 322         union {
 323                 unsigned char raw[512];
 324                 struct {
 325                         char name[100];         /*   0-99 */
 326                         char mode[8];           /* 100-107 */
 327                         char uid[8];            /* 108-115 */
 328                         char gid[8];            /* 116-123 */
 329                         char size[12];          /* 124-135 */
 330                         char mtime[12];         /* 136-147 */
 331                         char chksum[8];         /* 148-155 */
 332                         char typeflag;          /* 156-156 */
 333                         char linkname[100];     /* 157-256 */
 334                         char magic[6];          /* 257-262 */
 335                         char version[2];        /* 263-264 */
 336                         char uname[32];         /* 265-296 */
 337                         char gname[32];         /* 297-328 */
 338                         char devmajor[8];       /* 329-336 */
 339                         char devminor[8];       /* 337-344 */
 340                         char prefix[155];       /* 345-499 */
 341                         char padding[12];       /* 500-512 */
 342                 } formated;
 343         } tar;
 344         file_header_t *tar_entry = NULL;
 345         long i;
 346         long sum = 0;
 347
 348         if (archive_offset % 512 != 0) {
 349                 seek_forward(tar_stream, 512 - (archive_offset % 512));
 350         }
 351
 352         if (gzip_read(tar_stream, tar.raw, 512) != 512) {
 353                 /* Unfortunately its common for tar files to have all sorts of
 354                  * trailing garbage, fail silently */
 355 //              error_msg("Couldnt read header");
 356                 return(NULL);
 357         }
 358         archive_offset += 512;
 359
 360         /* Check header has valid magic, unfortunately some tar files
 361          * have empty (0'ed) tar entries at the end, which will
 362          * cause this to fail, so fail silently for now
 363          */
 364         if (strncmp(tar.formated.magic, "ustar", 5) != 0) {
 365 #ifdef CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY
 366                 if (strncmp(tar.formated.magic, "\0\0\0\0\0", 5) != 0)
 367 #endif
 368                 return(NULL);
 369         }
 370
 371         /* Do checksum on headers */
 372         for (i =  0; i < 148 ; i++) {
 373                 sum += tar.raw[i];
 374         }
 375         sum += ' ' * 8;
 376         for (i =  156; i < 512 ; i++) {
 377                 sum += tar.raw[i];
 378         }
 379         if (sum != strtol(tar.formated.chksum, NULL, 8)) {
 380                 if ( strtol(tar.formated.chksum,NULL,8) != 0 )
 381                         error_msg("Invalid tar header checksum");
 382                 return(NULL);
 383         }
 384
 385         /* convert to type'ed variables */
 386         tar_entry = xcalloc(1, sizeof(file_header_t));
 387
 388
 389
 390         // tar_entry->name = xstrdup(tar.formated.name);
 391
 392 /*
 393         parse_mode(tar.formated.mode, &tar_entry->mode);
 394 */
 395         tar_entry->mode = 07777 & strtol(tar.formated.mode, NULL, 8);
 396
 397         tar_entry->uid   = strtol(tar.formated.uid, NULL, 8);
 398         tar_entry->gid   = strtol(tar.formated.gid, NULL, 8);
 399         tar_entry->size  = strtol(tar.formated.size, NULL, 8);
 400         tar_entry->mtime = strtol(tar.formated.mtime, NULL, 8);
 401
 402         tar_entry->device = (strtol(tar.formated.devmajor, NULL, 8) << 8) +
 403                 strtol(tar.formated.devminor, NULL, 8);
 404
 405         /* Fix mode, used by the old format */
 406         switch (tar.formated.typeflag) {
 407         /* hard links are detected as regular files with 0 size and a link name */
 408         case '1':
 409                 tar_entry->mode |= S_IFREG ;
 410                 break;
 411         case 0:
 412         case '0':
 413
 414 # ifdef CONFIG_FEATURE_TAR_OLDGNU_COMPATABILITY
 415                 if (last_char_is(tar_entry->name, '/')) {
 416                         tar_entry->mode |= S_IFDIR;
 417                 } else
 418 # endif
 419                         tar_entry->mode |= S_IFREG;
 420                 break;
 421         case '2':
 422                 tar_entry->mode |= S_IFLNK;
 423                 break;
 424         case '3':
 425                 tar_entry->mode |= S_IFCHR;
 426                 break;
 427         case '4':
 428                 tar_entry->mode |= S_IFBLK;
 429                 break;
 430         case '5':
 431                 tar_entry->mode |= S_IFDIR;
 432                 break;
 433         case '6':
 434                 tar_entry->mode |= S_IFIFO;
 435                 break;
 436 # ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
 437         case 'L': {
 438                         longname = xmalloc(tar_entry->size + 1);
 439                         if(gzip_read(tar_stream, longname, tar_entry->size) != tar_entry->size)
 440                                 return NULL;
 441                         longname[tar_entry->size] = '\0';
 442                         archive_offset += tar_entry->size;
 443
 444                         return(get_header_tar(tar_stream));
 445                 }
 446         case 'K': {
 447                         linkname = xmalloc(tar_entry->size + 1);
 448                         if(gzip_read(tar_stream, linkname, tar_entry->size) != tar_entry->size)
 449                                 return NULL;
 450                         linkname[tar_entry->size] = '\0';
 451                         archive_offset += tar_entry->size;
 452
 453                         return(get_header_tar(tar_stream));
 454                 }
 455         case 'D':
 456         case 'M':
 457         case 'N':
 458         case 'S':
 459         case 'V':
 460                 perror_msg("Ignoring GNU extension type %c", tar.formated.typeflag);
 461 # endif
 462         default:
 463                 perror_msg("Unknown typeflag: 0x%x", tar.formated.typeflag);
 464                 break;
 465
 466         }
 467
 468
 469 #ifdef CONFIG_FEATURE_TAR_GNU_EXTENSIONS
 470         if (longname) {
 471                 tar_entry->name = longname;
 472                 longname = NULL;
 473         } else
 474 #endif
 475         {
 476                 tar_entry->name = xstrndup(tar.formated.name, 100);
 477
 478                 if (tar.formated.prefix[0]) {
 479                         char *temp = tar_entry->name;
 480                         char *prefixTemp = xstrndup(tar.formated.prefix, 155);
 481                         tar_entry->name = concat_path_file(prefixTemp, temp);
 482                         free(temp);
 483                         free(prefixTemp);
 484                 }
 485         }
 486
 487         if (linkname) {
 488                 tar_entry->link_name = linkname;
 489                 linkname = NULL;
 490         } else
 491         {
 492                 tar_entry->link_name = *tar.formated.linkname != '\0' ?
 493                         xstrndup(tar.formated.linkname, 100) : NULL;
 494         }
 495
 496         return(tar_entry);
 497 }
 498
 499 static void
 500 free_header_tar(file_header_t *tar_entry)
 501 {
 502         if (tar_entry == NULL)
 503                 return;
 504
 505         free(tar_entry->name);
 506         if (tar_entry->link_name)
 507                 free(tar_entry->link_name);
 508
 509         free(tar_entry);
 510 }
 511
 512 char *
 513 deb_extract(const char *package_filename, FILE *out_stream,
 514         const int extract_function, const char *prefix,
 515         const char *filename, int *err)
 516 {
 517         FILE *deb_stream = NULL;
 518         file_header_t *ar_header = NULL;
 519         const char **file_list = NULL;
 520         char *output_buffer = NULL;
 521         char *ared_file = NULL;
 522         char ar_magic[8];
 523         int gz_err;
 524         struct gzip_handle tar_outer, tar_inner;
 525         file_header_t *tar_header;
 526         ssize_t len;
 527
 528         *err = 0;
 529
 530         if (filename != NULL) {
 531                 file_list = xmalloc(sizeof(char *) * 2);
 532                 file_list[0] = filename;
 533                 file_list[1] = NULL;
 534         }
 535
 536         if (extract_function & extract_control_tar_gz) {
 537                 ared_file = "control.tar.gz";
 538         }
 539         else if (extract_function & extract_data_tar_gz) {
 540                 ared_file = "data.tar.gz";
 541         } else {
 542                 opkg_msg(ERROR, "Internal error: extract_function=%x\n",
 543                                 extract_function);
 544                 *err = -1;
 545                 goto cleanup;
 546         }
 547
 548         /* open the debian package to be worked on */
 549         deb_stream = wfopen(package_filename, "r");
 550         if (deb_stream == NULL) {
 551                 *err = -1;
 552                 goto cleanup;
 553         }
 554         /* set the buffer size */
 555         setvbuf(deb_stream, NULL, _IOFBF, 0x8000);
 556
 557         memset(&tar_outer, 0, sizeof(tar_outer));
 558         tar_outer.file = deb_stream;
 559         gzip_exec(&tar_outer, NULL);
 560
 561         /* walk through outer tar file to find ared_file */
 562         while ((tar_header = get_header_tar(&tar_outer)) != NULL) {
 563                     int name_offset = 0;
 564                     if (strncmp(tar_header->name, "./", 2) == 0)
 565                             name_offset = 2;
 566
 567                 if (strcmp(ared_file, tar_header->name+name_offset) == 0) {
 568                         memset(&tar_inner, 0, sizeof(tar_inner));
 569                         tar_inner.gzip = &tar_outer;
 570                         gzip_exec(&tar_inner, NULL);
 571
 572                         archive_offset = 0;
 573
 574                         output_buffer = unarchive(&tar_inner,
 575                                                   out_stream,
 576                                                   get_header_tar,
 577                                                   free_header_tar,
 578                                                   extract_function,
 579                                                   prefix,
 580                                                   file_list,
 581                                                   err);
 582
 583                         free_header_tar(tar_header);
 584                         gzip_close(&tar_inner);
 585                         break;
 586                 }
 587
 588                 seek_forward(&tar_outer, tar_header->size);
 589                 free_header_tar(tar_header);
 590         }
 591
 592 cleanup:
 593         gzip_close(&tar_outer);
 594
 595         if (file_list)
 596                 free(file_list);
 597
 598         return output_buffer;
 599 }