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