wimlib.net Git - wimlib/blob - src/extract_image.c

   1 /*
   2  * extract_image.c
   3  *
   4  * Support for extracting WIM files.
   5  */
   6
   7 /*
   8  * Copyright (C) 2012, 2013 Eric Biggers
   9  *
  10  * This file is part of wimlib, a library for working with WIM files.
  11  *
  12  * wimlib is free software; you can redistribute it and/or modify it under the
  13  * terms of the GNU General Public License as published by the Free
  14  * Software Foundation; either version 3 of the License, or (at your option)
  15  * any later version.
  16  *
  17  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  18  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  19  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  20  * details.
  21  *
  22  * You should have received a copy of the GNU General Public License
  23  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  24  */
  25
  26 #include "config.h"
  27
  28 #include <dirent.h>
  29
  30 #ifdef __WIN32__
  31 #  include "win32.h"
  32 #else
  33 #  ifdef HAVE_UTIME_H
  34 #    include <utime.h>
  35 #  endif
  36 #  include "timestamp.h"
  37 #  include <sys/time.h>
  38 #endif
  39
  40 #include <errno.h>
  41 #include <fcntl.h>
  42 #include <stdlib.h>
  43 #include <string.h>
  44 #include <sys/stat.h>
  45 #include <unistd.h>
  46
  47 #include "dentry.h"
  48 #include "lookup_table.h"
  49 #include "wimlib_internal.h"
  50 #include "xml.h"
  51
  52 #ifdef WITH_NTFS_3G
  53 #  include <ntfs-3g/volume.h>
  54 #endif
  55
  56 #ifdef HAVE_ALLOCA_H
  57 #  include <alloca.h>
  58 #endif
  59
  60
  61 #ifndef __WIN32__
  62
  63 /* Returns the number of components of @path.  */
  64 static unsigned
  65 get_num_path_components(const char *path)
  66 {
  67         unsigned num_components = 0;
  68         while (*path) {
  69                 while (*path == '/')
  70                         path++;
  71                 if (*path)
  72                         num_components++;
  73                 while (*path && *path != '/')
  74                         path++;
  75         }
  76         return num_components;
  77 }
  78
  79 static const char *
  80 path_next_part(const char *path)
  81 {
  82         while (*path && *path != '/')
  83                 path++;
  84         while (*path && *path == '/')
  85                 path++;
  86         return path;
  87 }
  88
  89 static int
  90 extract_regular_file_linked(struct wim_dentry *dentry,
  91                             const char *output_path,
  92                             struct apply_args *args,
  93                             struct wim_lookup_table_entry *lte)
  94 {
  95         /* This mode overrides the normal hard-link extraction and
  96          * instead either symlinks or hardlinks *all* identical files in
  97          * the WIM, even if they are in a different image (in the case
  98          * of a multi-image extraction) */
  99
 100         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
 101                 if (link(lte->extracted_file, output_path) != 0) {
 102                         ERROR_WITH_ERRNO("Failed to hard link "
 103                                          "`%s' to `%s'",
 104                                          output_path, lte->extracted_file);
 105                         return WIMLIB_ERR_LINK;
 106                 }
 107         } else {
 108                 int num_path_components;
 109                 int num_output_dir_path_components;
 110                 size_t extracted_file_len;
 111                 char *p;
 112                 const char *p2;
 113                 size_t i;
 114
 115                 num_path_components = get_num_path_components(dentry->_full_path) - 1;
 116                 num_output_dir_path_components = get_num_path_components(args->target);
 117
 118                 if (args->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
 119                         num_path_components++;
 120                         num_output_dir_path_components--;
 121                 }
 122                 extracted_file_len = strlen(lte->extracted_file);
 123
 124                 char buf[extracted_file_len + 3 * num_path_components + 1];
 125                 p = &buf[0];
 126
 127                 for (i = 0; i < num_path_components; i++) {
 128                         *p++ = '.';
 129                         *p++ = '.';
 130                         *p++ = '/';
 131                 }
 132                 p2 = lte->extracted_file;
 133                 while (*p2 == '/')
 134                         p2++;
 135                 while (num_output_dir_path_components > 0) {
 136                         p2 = path_next_part(p2);
 137                         num_output_dir_path_components--;
 138                 }
 139                 strcpy(p, p2);
 140                 if (symlink(buf, output_path) != 0) {
 141                         ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 142                                          buf, lte->extracted_file);
 143                         return WIMLIB_ERR_LINK;
 144                 }
 145         }
 146         return 0;
 147 }
 148
 149 static int
 150 symlink_apply_unix_data(const char *link,
 151                         const struct wimlib_unix_data *unix_data)
 152 {
 153         if (lchown(link, unix_data->uid, unix_data->gid)) {
 154                 if (errno == EPERM) {
 155                         /* Ignore */
 156                         WARNING_WITH_ERRNO("failed to set symlink UNIX "
 157                                            "owner/group on \"%s\"", link);
 158                 } else {
 159                         ERROR_WITH_ERRNO("failed to set symlink UNIX "
 160                                          "owner/group on \"%s\"", link);
 161                         return WIMLIB_ERR_INVALID_DENTRY;
 162                 }
 163         }
 164         return 0;
 165 }
 166
 167 static int
 168 fd_apply_unix_data(int fd, const char *path,
 169                    const struct wimlib_unix_data *unix_data,
 170                    int extract_flags)
 171 {
 172         if (extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS)
 173                 return 0;
 174
 175         if (fchown(fd, unix_data->uid, unix_data->gid)) {
 176                 if (errno == EPERM &&
 177                     !(extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
 178                 {
 179                         WARNING_WITH_ERRNO("failed to set file UNIX "
 180                                            "owner/group on \"%s\"", path);
 181                 } else {
 182                         ERROR_WITH_ERRNO("failed to set file UNIX "
 183                                          "owner/group on \"%s\"", path);
 184                         return (errno == EPERM) ? WIMLIB_ERR_INSUFFICIENT_PRIVILEGES_TO_EXTRACT :
 185                                 WIMLIB_ERR_WRITE;
 186                 }
 187         }
 188
 189         if (fchmod(fd, unix_data->mode)) {
 190                 if (errno == EPERM &&
 191                     !(extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
 192                 {
 193                         WARNING_WITH_ERRNO("failed to set UNIX file mode "
 194                                            "on \"%s\"", path);
 195                 } else {
 196                         ERROR_WITH_ERRNO("failed to set UNIX file mode "
 197                                          "on \"%s\"", path);
 198                         return (errno == EPERM) ? WIMLIB_ERR_INSUFFICIENT_PRIVILEGES_TO_EXTRACT :
 199                                 WIMLIB_ERR_WRITE;
 200                 }
 201         }
 202         return 0;
 203 }
 204
 205 static int
 206 dir_apply_unix_data(const char *dir, const struct wimlib_unix_data *unix_data,
 207                     int extract_flags)
 208 {
 209         int dfd = open(dir, O_RDONLY);
 210         int ret;
 211         if (dfd >= 0) {
 212                 ret = fd_apply_unix_data(dfd, dir, unix_data, extract_flags);
 213                 if (close(dfd) && ret == 0) {
 214                         ERROR_WITH_ERRNO("can't close directory `%s'", dir);
 215                         ret = WIMLIB_ERR_WRITE;
 216                 }
 217         } else {
 218                 ERROR_WITH_ERRNO("can't open directory `%s'", dir);
 219                 ret = WIMLIB_ERR_OPENDIR;
 220         }
 221         return ret;
 222 }
 223
 224 static int
 225 extract_regular_file_unlinked(struct wim_dentry *dentry,
 226                               struct apply_args *args,
 227                               const char *output_path,
 228                               struct wim_lookup_table_entry *lte)
 229 {
 230         /* Normal mode of extraction.  Regular files and hard links are
 231          * extracted in the way that they appear in the WIM. */
 232
 233         int out_fd;
 234         int ret;
 235         struct wim_inode *inode = dentry->d_inode;
 236
 237         if (!((args->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE)
 238                 && (args->extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 239                                      WIMLIB_EXTRACT_FLAG_HARDLINK))))
 240         {
 241                 /* If the dentry is part of a hard link set of at least 2
 242                  * dentries and one of the other dentries has already been
 243                  * extracted, make a hard link to the file corresponding to this
 244                  * already-extracted directory.  Otherwise, extract the file and
 245                  * set the inode->i_extracted_file field so that other dentries
 246                  * in the hard link group can link to it. */
 247                 if (inode->i_nlink > 1) {
 248                         if (inode->i_extracted_file) {
 249                                 DEBUG("Extracting hard link `%s' => `%s'",
 250                                       output_path, inode->i_extracted_file);
 251                                 if (link(inode->i_extracted_file, output_path) != 0) {
 252                                         ERROR_WITH_ERRNO("Failed to hard link "
 253                                                          "`%s' to `%s'",
 254                                                          output_path,
 255                                                          inode->i_extracted_file);
 256                                         return WIMLIB_ERR_LINK;
 257                                 }
 258                                 return 0;
 259                         }
 260                         FREE(inode->i_extracted_file);
 261                         inode->i_extracted_file = STRDUP(output_path);
 262                         if (!inode->i_extracted_file) {
 263                                 ERROR("Failed to allocate memory for filename");
 264                                 return WIMLIB_ERR_NOMEM;
 265                         }
 266                 }
 267         }
 268
 269         /* Extract the contents of the file to @output_path. */
 270
 271         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
 272         if (out_fd == -1) {
 273                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
 274                                  output_path);
 275                 return WIMLIB_ERR_OPEN;
 276         }
 277
 278         if (!lte) {
 279                 /* Empty file with no lookup table entry */
 280                 DEBUG("Empty file `%s'.", output_path);
 281                 ret = 0;
 282                 goto out_extract_unix_data;
 283         }
 284
 285         ret = extract_wim_resource_to_fd(lte, out_fd, wim_resource_size(lte));
 286         if (ret) {
 287                 ERROR("Failed to extract resource to `%s'", output_path);
 288                 goto out;
 289         }
 290
 291 out_extract_unix_data:
 292         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 293                 struct wimlib_unix_data unix_data;
 294                 ret = inode_get_unix_data(inode, &unix_data, NULL);
 295                 if (ret > 0)
 296                         ;
 297                 else if (ret < 0)
 298                         ret = 0;
 299                 else
 300                         ret = fd_apply_unix_data(out_fd, output_path, &unix_data,
 301                                                  args->extract_flags);
 302                 if (ret)
 303                         goto out;
 304         }
 305         if (lte)
 306                 args->progress.extract.completed_bytes += wim_resource_size(lte);
 307 out:
 308         if (close(out_fd) != 0) {
 309                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
 310                 if (ret == 0)
 311                         ret = WIMLIB_ERR_WRITE;
 312         }
 313         return ret;
 314 }
 315
 316 static int
 317 extract_regular_file(struct wim_dentry *dentry,
 318                      struct apply_args *args,
 319                      const char *output_path)
 320 {
 321         struct wim_lookup_table_entry *lte;
 322         const struct wim_inode *inode = dentry->d_inode;
 323
 324         lte = inode_unnamed_lte_resolved(inode);
 325
 326         if (lte && (args->extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 327                                            WIMLIB_EXTRACT_FLAG_HARDLINK)))
 328         {
 329                 if (lte->extracted_file) {
 330                         return extract_regular_file_linked(dentry, output_path, args, lte);
 331                 } else {
 332                         lte->extracted_file = STRDUP(output_path);
 333                         if (!lte->extracted_file)
 334                                 return WIMLIB_ERR_NOMEM;
 335                 }
 336         }
 337         return extract_regular_file_unlinked(dentry, args, output_path, lte);
 338 }
 339
 340 static int
 341 extract_symlink(struct wim_dentry *dentry,
 342                 struct apply_args *args,
 343                 const char *output_path)
 344 {
 345         char target[4096 + args->target_realpath_len];
 346         char *fixed_target;
 347         const struct wim_inode *inode = dentry->d_inode;
 348
 349         ssize_t ret = wim_inode_readlink(inode,
 350                                          target + args->target_realpath_len,
 351                                          sizeof(target) - args->target_realpath_len - 1);
 352         struct wim_lookup_table_entry *lte;
 353
 354         if (ret <= 0) {
 355                 ERROR("Could not read the symbolic link from dentry `%s'",
 356                       dentry->_full_path);
 357                 return WIMLIB_ERR_INVALID_DENTRY;
 358         }
 359         target[args->target_realpath_len + ret] = '\0';
 360         if (target[args->target_realpath_len] == '/' &&
 361             args->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)
 362         {
 363                 /* Fix absolute symbolic link target to point into the actual
 364                  * extraction destination */
 365                 memcpy(target, args->target_realpath,
 366                        args->target_realpath_len);
 367                 fixed_target = target;
 368         } else {
 369                 /* Keep same link target */
 370                 fixed_target = target + args->target_realpath_len;
 371         }
 372         ret = symlink(fixed_target, output_path);
 373         if (ret) {
 374                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 375                                  output_path, fixed_target);
 376                 return WIMLIB_ERR_LINK;
 377         }
 378         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 379                 struct wimlib_unix_data unix_data;
 380                 ret = inode_get_unix_data(inode, &unix_data, NULL);
 381                 if (ret > 0)
 382                         ;
 383                 else if (ret < 0)
 384                         ret = 0;
 385                 else
 386                         ret = symlink_apply_unix_data(output_path, &unix_data);
 387                 if (ret)
 388                         return ret;
 389         }
 390         lte = inode_unnamed_lte_resolved(inode);
 391         wimlib_assert(lte != NULL);
 392         args->progress.extract.completed_bytes += wim_resource_size(lte);
 393         return 0;
 394 }
 395
 396 #endif /* !__WIN32__ */
 397
 398 static int
 399 extract_directory(struct wim_dentry *dentry,
 400                   const tchar *output_path, bool is_root,
 401                   int extract_flags)
 402 {
 403         int ret;
 404         struct stat stbuf;
 405
 406         ret = tstat(output_path, &stbuf);
 407         if (ret == 0) {
 408                 if (S_ISDIR(stbuf.st_mode)) {
 409                         /*if (!is_root)*/
 410                                 /*WARNING("`%s' already exists", output_path);*/
 411                         goto dir_exists;
 412                 } else {
 413                         ERROR("`%"TS"' is not a directory", output_path);
 414                         return WIMLIB_ERR_MKDIR;
 415                 }
 416         } else {
 417                 if (errno != ENOENT) {
 418                         ERROR_WITH_ERRNO("Failed to stat `%"TS"'", output_path);
 419                         return WIMLIB_ERR_STAT;
 420                 }
 421         }
 422
 423         if (tmkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH))
 424         {
 425                 ERROR_WITH_ERRNO("Cannot create directory `%"TS"'", output_path);
 426                 return WIMLIB_ERR_MKDIR;
 427         }
 428 dir_exists:
 429         ret = 0;
 430 #ifndef __WIN32__
 431         if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 432                 struct wimlib_unix_data unix_data;
 433                 ret = inode_get_unix_data(dentry->d_inode, &unix_data, NULL);
 434                 if (ret > 0)
 435                         ;
 436                 else if (ret < 0)
 437                         ret = 0;
 438                 else
 439                         ret = dir_apply_unix_data(output_path, &unix_data,
 440                                                   extract_flags);
 441         }
 442 #endif
 443         return ret;
 444 }
 445
 446 #ifndef __WIN32__
 447 static int
 448 unix_do_apply_dentry(const char *output_path, size_t output_path_len,
 449                      struct wim_dentry *dentry, struct apply_args *args)
 450 {
 451         const struct wim_inode *inode = dentry->d_inode;
 452
 453         if (inode_is_symlink(inode))
 454                 return extract_symlink(dentry, args, output_path);
 455         else if (inode_is_directory(inode))
 456                 return extract_directory((args->extract_flags &
 457                                            WIMLIB_EXTRACT_FLAG_UNIX_DATA) ? dentry : NULL,
 458                                          output_path, false, args->extract_flags);
 459         else
 460                 return extract_regular_file(dentry, args, output_path);
 461 }
 462
 463 static int
 464 unix_do_apply_dentry_timestamps(const char *output_path,
 465                                 size_t output_path_len,
 466                                 struct wim_dentry *dentry,
 467                                 struct apply_args *args)
 468 {
 469         int ret;
 470         const struct wim_inode *inode = dentry->d_inode;
 471
 472 #ifdef HAVE_UTIMENSAT
 473         /* Convert the WIM timestamps, which are accurate to 100 nanoseconds,
 474          * into `struct timespec's for passing to utimensat(), which is accurate
 475          * to 1 nanosecond. */
 476
 477         struct timespec ts[2];
 478         ts[0] = wim_timestamp_to_timespec(inode->i_last_access_time);
 479         ts[1] = wim_timestamp_to_timespec(inode->i_last_write_time);
 480         ret = utimensat(AT_FDCWD, output_path, ts, AT_SYMLINK_NOFOLLOW);
 481         if (ret)
 482                 ret = errno;
 483 #else
 484         ret = ENOSYS;
 485 #endif
 486
 487         if (ret == ENOSYS) {
 488                 /* utimensat() not implemented or not available */
 489         #ifdef HAVE_LUTIMES
 490                 /* Convert the WIM timestamps, which are accurate to 100
 491                  * nanoseconds, into `struct timeval's for passing to lutimes(),
 492                  * which is accurate to 1 microsecond. */
 493                 struct timeval tv[2];
 494                 tv[0] = wim_timestamp_to_timeval(inode->i_last_access_time);
 495                 tv[1] = wim_timestamp_to_timeval(inode->i_last_write_time);
 496                 ret = lutimes(output_path, tv);
 497                 if (ret)
 498                         ret = errno;
 499         #endif
 500         }
 501
 502         if (ret == ENOSYS) {
 503                 /* utimensat() and lutimes() both not implemented or not
 504                  * available */
 505         #ifdef HAVE_UTIME
 506                 /* Convert the WIM timestamps, which are accurate to 100
 507                  * nanoseconds, into a `struct utimbuf's for passing to
 508                  * utime(), which is accurate to 1 second. */
 509                 struct utimbuf buf;
 510                 buf.actime = wim_timestamp_to_unix(inode->i_last_access_time);
 511                 buf.modtime = wim_timestamp_to_unix(inode->i_last_write_time);
 512                 ret = utime(output_path, &buf);
 513         #endif
 514         }
 515         if (ret && args->num_utime_warnings < 10) {
 516                 WARNING_WITH_ERRNO("Failed to set timestamp on file `%s'",
 517                                     output_path);
 518                 args->num_utime_warnings++;
 519         }
 520         return 0;
 521 }
 522 #endif /* !__WIN32__ */
 523
 524 static int
 525 do_apply_op(struct wim_dentry *dentry, struct apply_args *args,
 526             int (*apply_dentry_func)(const tchar *, size_t,
 527                                      struct wim_dentry *, struct apply_args *))
 528 {
 529         tchar *p;
 530         const tchar *full_path;
 531         size_t full_path_nchars;
 532
 533         wimlib_assert(dentry->_full_path != NULL);
 534         full_path = dentry->_full_path + 1;
 535         full_path_nchars = dentry->full_path_nbytes / sizeof(tchar) - 1;
 536         tchar output_path[args->target_nchars + 1 +
 537                          (full_path_nchars - args->wim_source_path_nchars) + 1];
 538         p = output_path;
 539
 540         /*print_dentry(dentry, NULL);*/
 541         /*ERROR("%"TS" %"TS, args->target, dentry->_full_path);*/
 542         /*ERROR("");*/
 543
 544         tmemcpy(p, args->target, args->target_nchars);
 545         p += args->target_nchars;
 546
 547         if (dentry != args->extract_root) {
 548                 *p++ = T('/');
 549                 tmemcpy(p, full_path + args->wim_source_path_nchars,
 550                         full_path_nchars - args->wim_source_path_nchars);
 551                 p += full_path_nchars - args->wim_source_path_nchars;
 552         }
 553         *p = T('\0');
 554         return (*apply_dentry_func)(output_path, p - output_path,
 555                                     dentry, args);
 556 }
 557
 558
 559 /* Extracts a file, directory, or symbolic link from the WIM archive. */
 560 static int
 561 apply_dentry_normal(struct wim_dentry *dentry, void *arg)
 562 {
 563 #ifdef __WIN32__
 564         return do_apply_op(dentry, arg, win32_do_apply_dentry);
 565 #else
 566         return do_apply_op(dentry, arg, unix_do_apply_dentry);
 567 #endif
 568 }
 569
 570
 571 /* Apply timestamps to an extracted file or directory */
 572 static int
 573 apply_dentry_timestamps_normal(struct wim_dentry *dentry, void *arg)
 574 {
 575 #ifdef __WIN32__
 576         return do_apply_op(dentry, arg, win32_do_apply_dentry_timestamps);
 577 #else
 578         return do_apply_op(dentry, arg, unix_do_apply_dentry_timestamps);
 579 #endif
 580 }
 581
 582 static bool
 583 dentry_is_descendent(const struct wim_dentry *dentry,
 584                      const struct wim_dentry *ancestor)
 585 {
 586         for (;;) {
 587                 if (dentry == ancestor)
 588                         return true;
 589                 if (dentry_is_root(dentry))
 590                         return false;
 591                 dentry = dentry->parent;
 592         }
 593 }
 594
 595 /* Extract a dentry if it hasn't already been extracted and either
 596  * WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified, or the dentry is a directory
 597  * and/or has no unnamed stream. */
 598 static int
 599 maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
 600 {
 601         struct apply_args *args = arg;
 602         int ret;
 603
 604         if (dentry->is_extracted)
 605                 return 0;
 606
 607         if (!dentry_is_descendent(dentry, args->extract_root))
 608                 return 0;
 609
 610         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS &&
 611             !dentry_is_directory(dentry) &&
 612             inode_unnamed_lte_resolved(dentry->d_inode) != NULL)
 613                 return 0;
 614
 615         if ((args->extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) &&
 616              args->progress_func) {
 617                 args->progress.extract.cur_path = dentry->_full_path;
 618                 args->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY,
 619                                     &args->progress);
 620         }
 621         ret = args->apply_dentry(dentry, args);
 622         if (ret == 0)
 623                 dentry->is_extracted = 1;
 624         return ret;
 625 }
 626
 627 static void
 628 calculate_bytes_to_extract(struct list_head *stream_list,
 629                            int extract_flags,
 630                            union wimlib_progress_info *progress)
 631 {
 632         struct wim_lookup_table_entry *lte;
 633         u64 total_bytes = 0;
 634         u64 num_streams = 0;
 635
 636         /* For each stream to be extracted... */
 637         list_for_each_entry(lte, stream_list, extraction_list) {
 638                 if (extract_flags &
 639                     (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 640                 {
 641                         /* In the symlink or hard link extraction mode, each
 642                          * stream will be extracted one time regardless of how
 643                          * many dentries share the stream. */
 644                         wimlib_assert(!(extract_flags & WIMLIB_EXTRACT_FLAG_NTFS));
 645                         if (!lte->extracted_file) {
 646                                 num_streams++;
 647                                 total_bytes += wim_resource_size(lte);
 648                         }
 649                 } else {
 650                         num_streams += lte->out_refcnt;
 651                         total_bytes += lte->out_refcnt * wim_resource_size(lte);
 652                 }
 653         }
 654         progress->extract.num_streams = num_streams;
 655         progress->extract.total_bytes = total_bytes;
 656         progress->extract.completed_bytes = 0;
 657 }
 658
 659 static void
 660 maybe_add_stream_for_extraction(struct wim_lookup_table_entry *lte,
 661                                 struct list_head *stream_list)
 662 {
 663         if (++lte->out_refcnt == 1) {
 664                 INIT_LIST_HEAD(&lte->inode_list);
 665                 list_add_tail(&lte->extraction_list, stream_list);
 666         }
 667 }
 668
 669 static void
 670 inode_find_streams_for_extraction(struct wim_inode *inode,
 671                                   struct list_head *stream_list,
 672                                   int extract_flags)
 673 {
 674         struct wim_lookup_table_entry *lte;
 675         bool inode_added = false;
 676
 677         lte = inode_unnamed_lte_resolved(inode);
 678         if (lte) {
 679                 maybe_add_stream_for_extraction(lte, stream_list);
 680                 list_add_tail(&inode->i_lte_inode_list, &lte->inode_list);
 681                 inode_added = true;
 682         }
 683
 684         /* Determine whether to include alternate data stream entries or not.
 685          *
 686          * UNIX:  Include them if extracting using NTFS-3g.
 687          *
 688          * Windows: Include them undconditionally, although if the filesystem is
 689          * not NTFS we won't actually be able to extract them. */
 690 #if defined(WITH_NTFS_3G)
 691         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
 692 #elif defined(__WIN32__)
 693         if (1)
 694 #else
 695         if (0)
 696 #endif
 697         {
 698                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
 699                         if (inode->i_ads_entries[i].stream_name_nbytes != 0) {
 700                                 lte = inode->i_ads_entries[i].lte;
 701                                 if (lte) {
 702                                         maybe_add_stream_for_extraction(lte,
 703                                                                         stream_list);
 704                                         if (!inode_added) {
 705                                                 list_add_tail(&inode->i_lte_inode_list,
 706                                                               &lte->inode_list);
 707                                                 inode_added = true;
 708                                         }
 709                                 }
 710                         }
 711                 }
 712         }
 713 }
 714
 715 struct find_streams_ctx {
 716         struct list_head stream_list;
 717         int extract_flags;
 718 };
 719
 720 static int
 721 dentry_find_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
 722 {
 723         struct find_streams_ctx *ctx = _ctx;
 724         struct wim_inode *inode = dentry->d_inode;
 725
 726         dentry->is_extracted = 0;
 727         if (!inode->i_visited) {
 728                 inode_find_streams_for_extraction(inode, &ctx->stream_list,
 729                                                   ctx->extract_flags);
 730                 inode->i_visited = 1;
 731         }
 732         return 0;
 733 }
 734
 735 static int
 736 dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_lookup_table)
 737 {
 738         struct wim_inode *inode = dentry->d_inode;
 739         struct wim_lookup_table *lookup_table = _lookup_table;
 740         struct wim_lookup_table_entry *lte;
 741
 742         inode_resolve_ltes(inode, lookup_table);
 743         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 744                 lte = inode_stream_lte_resolved(inode, i);
 745                 if (lte)
 746                         lte->out_refcnt = 0;
 747         }
 748         return 0;
 749 }
 750
 751 static void
 752 find_streams_for_extraction(struct wim_dentry *root,
 753                             struct list_head *stream_list,
 754                             struct wim_lookup_table *lookup_table,
 755                             int extract_flags)
 756 {
 757         struct find_streams_ctx ctx;
 758
 759         INIT_LIST_HEAD(&ctx.stream_list);
 760         ctx.extract_flags = extract_flags;
 761         for_dentry_in_tree(root, dentry_resolve_and_zero_lte_refcnt, lookup_table);
 762         for_dentry_in_tree(root, dentry_find_streams_to_extract, &ctx);
 763         list_transfer(&ctx.stream_list, stream_list);
 764 }
 765
 766 static int
 767 dentry_mark_inode_unvisited(struct wim_dentry *dentry, void *_ignore)
 768 {
 769         dentry->d_inode->i_visited = 0;
 770         return 0;
 771 }
 772
 773 struct apply_operations {
 774         int (*apply_dentry)(struct wim_dentry *dentry, void *arg);
 775         int (*apply_dentry_timestamps)(struct wim_dentry *dentry, void *arg);
 776 };
 777
 778 static const struct apply_operations normal_apply_operations = {
 779         .apply_dentry = apply_dentry_normal,
 780         .apply_dentry_timestamps = apply_dentry_timestamps_normal,
 781 };
 782
 783 #ifdef WITH_NTFS_3G
 784 static const struct apply_operations ntfs_apply_operations = {
 785         .apply_dentry = apply_dentry_ntfs,
 786         .apply_dentry_timestamps = apply_dentry_timestamps_ntfs,
 787 };
 788 #endif
 789
 790 static int
 791 apply_stream_list(struct list_head *stream_list,
 792                   struct apply_args *args,
 793                   const struct apply_operations *ops,
 794                   wimlib_progress_func_t progress_func)
 795 {
 796         uint64_t bytes_per_progress = args->progress.extract.total_bytes / 100;
 797         uint64_t next_progress = bytes_per_progress;
 798         struct wim_lookup_table_entry *lte;
 799         struct wim_inode *inode;
 800         struct wim_dentry *dentry;
 801         int ret;
 802
 803         /* This complicated loop is essentially looping through the dentries,
 804          * although dentries may be visited more than once (if a dentry contains
 805          * two different nonempty streams) or not at all (if a dentry contains
 806          * no non-empty streams).
 807          *
 808          * The outer loop is over the distinct streams to be extracted so that
 809          * sequential reading of the WIM can be implemented. */
 810
 811         /* For each distinct stream to be extracted */
 812         list_for_each_entry(lte, stream_list, extraction_list) {
 813                 /* For each inode that contains the stream */
 814                 list_for_each_entry(inode, &lte->inode_list, i_lte_inode_list) {
 815                         /* For each dentry that points to the inode */
 816                         inode_for_each_dentry(dentry, inode) {
 817                                 /* Extract the dentry if it was not already
 818                                  * extracted */
 819                                 ret = maybe_apply_dentry(dentry, args);
 820                                 if (ret)
 821                                         return ret;
 822                                 if (progress_func &&
 823                                     args->progress.extract.completed_bytes >= next_progress)
 824                                 {
 825                                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS,
 826                                                       &args->progress);
 827                                         if (args->progress.extract.completed_bytes >=
 828                                             args->progress.extract.total_bytes)
 829                                         {
 830                                                 next_progress = ~0ULL;
 831                                         } else {
 832                                                 next_progress =
 833                                                         min (args->progress.extract.completed_bytes +
 834                                                              bytes_per_progress,
 835                                                              args->progress.extract.total_bytes);
 836                                         }
 837                                 }
 838                         }
 839                 }
 840         }
 841         return 0;
 842 }
 843
 844 static int
 845 sort_stream_list_by_wim_position(struct list_head *stream_list)
 846 {
 847         struct list_head *cur;
 848         size_t num_streams;
 849         struct wim_lookup_table_entry **array;
 850         size_t i;
 851         size_t array_size;
 852
 853         num_streams = 0;
 854         list_for_each(cur, stream_list)
 855                 num_streams++;
 856         array_size = num_streams * sizeof(array[0]);
 857         array = MALLOC(array_size);
 858         if (!array) {
 859                 ERROR("Failed to allocate %zu bytes to sort stream entries",
 860                       array_size);
 861                 return WIMLIB_ERR_NOMEM;
 862         }
 863         cur = stream_list->next;
 864         for (i = 0; i < num_streams; i++) {
 865                 array[i] = container_of(cur, struct wim_lookup_table_entry, extraction_list);
 866                 cur = cur->next;
 867         }
 868
 869         qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
 870
 871         INIT_LIST_HEAD(stream_list);
 872         for (i = 0; i < num_streams; i++)
 873                 list_add_tail(&array[i]->extraction_list, stream_list);
 874         FREE(array);
 875         return 0;
 876 }
 877
 878 /*
 879  * Extract a dentry to standard output.
 880  *
 881  * This obviously doesn't make sense in all cases.  We return an error if the
 882  * dentry does not correspond to a regular file.  Otherwise we extract the
 883  * unnamed data stream only.
 884  */
 885 static int
 886 extract_dentry_to_stdout(struct wim_dentry *dentry)
 887 {
 888         int ret = 0;
 889         if (!dentry_is_regular_file(dentry)) {
 890                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 891                       "extracted to standard output", dentry->_full_path);
 892                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
 893         } else {
 894                 struct wim_lookup_table_entry *lte;
 895
 896                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
 897                 if (lte) {
 898                         ret = extract_wim_resource_to_fd(lte, STDOUT_FILENO,
 899                                                          wim_resource_size(lte));
 900                 }
 901         }
 902         return ret;
 903 }
 904
 905 /*
 906  * extract_tree - Extract a file or directory tree from the currently selected
 907  *                WIM image.
 908  *
 909  * @wim:        WIMStruct for the WIM file, with the desired image selected
 910  *              (as wim->current_image).
 911  * @wim_source_path:
 912  *              "Canonical" (i.e. no leading or trailing slashes, path
 913  *              separators forwald slashes) path inside the WIM image to
 914  *              extract.  An empty string means the full image.
 915  * @target:
 916  *              Filesystem path to extract the file or directory tree to.
 917  *
 918  * @extract_flags:
 919  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
 920  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
 921  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
 922  *              the image.
 923  *
 924  * @progress_func:
 925  *              If non-NULL, progress function for the extraction.  The messages
 926  *              we may in this function are:
 927  *
 928  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
 929  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
 930  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
 931  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
 932  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
 933  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
 934  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
 935  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
 936  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
 937  *
 938  * Returns 0 on success; nonzero on failure.
 939  */
 940 static int
 941 extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
 942              int extract_flags, wimlib_progress_func_t progress_func)
 943 {
 944         int ret;
 945         struct list_head stream_list;
 946         struct apply_args args;
 947         const struct apply_operations *ops;
 948         struct wim_dentry *root;
 949
 950         memset(&args, 0, sizeof(args));
 951
 952         args.w                      = wim;
 953         args.target                 = target;
 954         args.extract_flags          = extract_flags;
 955         args.progress_func          = progress_func;
 956         args.target_nchars          = tstrlen(target);
 957         args.wim_source_path_nchars = tstrlen(wim_source_path);
 958
 959         if (progress_func) {
 960                 args.progress.extract.wimfile_name = wim->filename;
 961                 args.progress.extract.image = wim->current_image;
 962                 args.progress.extract.extract_flags = (extract_flags &
 963                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
 964                 args.progress.extract.image_name = wimlib_get_image_name(wim,
 965                                                                          wim->current_image);
 966                 args.progress.extract.extract_root_wim_source_path = wim_source_path;
 967                 args.progress.extract.target = target;
 968         }
 969
 970 #ifdef WITH_NTFS_3G
 971         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 972                 args.vol = ntfs_mount(target, 0);
 973                 if (!args.vol) {
 974                         ERROR_WITH_ERRNO("Failed to mount NTFS volume `%"TS"'",
 975                                          target);
 976                         ret = WIMLIB_ERR_NTFS_3G;
 977                         goto out;
 978                 }
 979                 ops = &ntfs_apply_operations;
 980         } else
 981 #endif
 982                 ops = &normal_apply_operations;
 983
 984         root = get_dentry(wim, wim_source_path);
 985         if (!root) {
 986                 ERROR("Path \"%"TS"\" does not exist in WIM image %d",
 987                       wim_source_path, wim->current_image);
 988                 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
 989                 goto out_ntfs_umount;
 990         }
 991         args.extract_root = root;
 992
 993         ret = calculate_dentry_tree_full_paths(root);
 994         if (ret)
 995                 goto out_ntfs_umount;
 996
 997
 998         /* Build a list of the streams that need to be extracted */
 999         find_streams_for_extraction(root,
1000                                     &stream_list,
1001                                     wim->lookup_table, extract_flags);
1002
1003         /* Calculate the number of bytes of data that will be extracted */
1004         calculate_bytes_to_extract(&stream_list, extract_flags,
1005                                    &args.progress);
1006
1007         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1008                 ret = extract_dentry_to_stdout(root);
1009                 goto out_mark_inodes_unvisited;
1010         }
1011
1012         if (progress_func) {
1013                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
1014                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
1015                               &args.progress);
1016         }
1017
1018         /* If a sequential extraction was specified, sort the streams to be
1019          * extracted by their position in the WIM file, so that the WIM file can
1020          * be read sequentially. */
1021         if (extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL) {
1022                 ret = sort_stream_list_by_wim_position(&stream_list);
1023                 if (ret != 0) {
1024                         WARNING("Falling back to non-sequential extraction");
1025                         extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
1026                 }
1027         }
1028
1029         if (progress_func) {
1030                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
1031                               &args.progress);
1032         }
1033
1034         /* Make the directory structure and extract empty files */
1035         args.extract_flags |= WIMLIB_EXTRACT_FLAG_NO_STREAMS;
1036         args.apply_dentry = ops->apply_dentry;
1037         ret = for_dentry_in_tree(root, maybe_apply_dentry, &args);
1038         args.extract_flags &= ~WIMLIB_EXTRACT_FLAG_NO_STREAMS;
1039         if (ret)
1040                 goto out_mark_inodes_unvisited;
1041
1042         if (progress_func) {
1043                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
1044                               &args.progress);
1045         }
1046
1047         if (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) {
1048                 args.target_realpath = realpath(target, NULL);
1049                 if (!args.target_realpath) {
1050                         ret = WIMLIB_ERR_NOMEM;
1051                         goto out_mark_inodes_unvisited;
1052                 }
1053                 args.target_realpath_len = tstrlen(args.target_realpath);
1054         }
1055
1056         /* Extract non-empty files */
1057         ret = apply_stream_list(&stream_list, &args, ops, progress_func);
1058         if (ret)
1059                 goto out_free_target_realpath;
1060
1061         if (progress_func) {
1062                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
1063                               &args.progress);
1064         }
1065
1066         /* Apply timestamps */
1067         ret = for_dentry_in_tree_depth(root,
1068                                        ops->apply_dentry_timestamps, &args);
1069         if (ret)
1070                 goto out_free_target_realpath;
1071
1072         if (progress_func) {
1073                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
1074                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
1075                               &args.progress);
1076         }
1077 out_free_target_realpath:
1078         FREE(args.target_realpath);
1079 out_mark_inodes_unvisited:
1080         for_dentry_in_tree(root, dentry_mark_inode_unvisited, NULL);
1081 out_ntfs_umount:
1082 #ifdef WITH_NTFS_3G
1083         /* Unmount the NTFS volume */
1084         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1085                 if (ntfs_umount(args.vol, FALSE) != 0) {
1086                         ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%"TS"'",
1087                                          args.target);
1088                         if (ret == 0)
1089                                 ret = WIMLIB_ERR_NTFS_3G;
1090                 }
1091         }
1092 #endif
1093 out:
1094         return ret;
1095 }
1096
1097 /* Validates a single wimlib_extract_command, mostly checking to make sure the
1098  * extract flags make sense. */
1099 static int
1100 check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
1101 {
1102         int extract_flags;
1103         bool is_entire_image = (cmd->wim_source_path[0] == T('\0'));
1104
1105         /* Empty destination path? */
1106         if (cmd->fs_dest_path[0] == T('\0'))
1107                 return WIMLIB_ERR_INVALID_PARAM;
1108
1109         extract_flags = cmd->extract_flags;
1110
1111         /* Specified both symlink and hardlink modes? */
1112         if ((extract_flags &
1113              (WIMLIB_EXTRACT_FLAG_SYMLINK |
1114               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
1115                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
1116                 return WIMLIB_ERR_INVALID_PARAM;
1117
1118 #ifdef __WIN32__
1119         /* Wanted UNIX data on Win32? */
1120         if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
1121                 ERROR("Extracting UNIX data is not supported on Windows");
1122                 return WIMLIB_ERR_INVALID_PARAM;
1123         }
1124         /* Wanted linked extraction on Windows?  (XXX This is possible, just not
1125          * implemented yet.) */
1126         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
1127                              WIMLIB_EXTRACT_FLAG_HARDLINK))
1128         {
1129                 ERROR("Linked extraction modes are not supported on Windows");
1130                 return WIMLIB_ERR_INVALID_PARAM;
1131         }
1132 #endif
1133
1134         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1135                 /* NTFS-3g extraction mode requested */
1136 #ifdef WITH_NTFS_3G
1137                 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
1138                                       WIMLIB_EXTRACT_FLAG_HARDLINK))) {
1139                         ERROR("Cannot specify symlink or hardlink flags when applying\n"
1140                               "        directly to a NTFS volume");
1141                         return WIMLIB_ERR_INVALID_PARAM;
1142                 }
1143                 if (!is_entire_image &&
1144                     (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS))
1145                 {
1146                         ERROR("When applying directly to a NTFS volume you can "
1147                               "only extract a full image, not part of one");
1148                         return WIMLIB_ERR_INVALID_PARAM;
1149                 }
1150                 if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
1151                         ERROR("Cannot restore UNIX-specific data in "
1152                               "the NTFS extraction mode");
1153                         return WIMLIB_ERR_INVALID_PARAM;
1154                 }
1155 #else
1156                 ERROR("wimlib was compiled without support for NTFS-3g, so");
1157                 ERROR("we cannot apply a WIM image directly to a NTFS volume");
1158                 return WIMLIB_ERR_UNSUPPORTED;
1159 #endif
1160         }
1161
1162         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1163                               WIMLIB_EXTRACT_FLAG_NORPFIX)) ==
1164                 (WIMLIB_EXTRACT_FLAG_RPFIX | WIMLIB_EXTRACT_FLAG_NORPFIX))
1165         {
1166                 ERROR("Cannot specify RPFIX and NORPFIX flags at the same time!");
1167                 return WIMLIB_ERR_INVALID_PARAM;
1168         }
1169
1170         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1171                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
1172         {
1173                 /* Do reparse point fixups by default if the WIM header says
1174                  * they are enabled and we are extracting a full image. */
1175                 if ((wim_header_flags & WIM_HDR_FLAG_RP_FIX) && is_entire_image)
1176                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1177         }
1178
1179         if (!is_entire_image && (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)) {
1180                 ERROR("Cannot specify --rpfix when not extracting entire image");
1181                 return WIMLIB_ERR_INVALID_PARAM;
1182         }
1183
1184         cmd->extract_flags = extract_flags;
1185         return 0;
1186 }
1187
1188
1189 /* Internal function to execute extraction commands for a WIM image. */
1190 static int
1191 do_wimlib_extract_files(WIMStruct *wim,
1192                         int image,
1193                         struct wimlib_extract_command *cmds,
1194                         size_t num_cmds,
1195                         wimlib_progress_func_t progress_func)
1196 {
1197         int ret;
1198         bool found_link_cmd = false;
1199         bool found_nolink_cmd = false;
1200
1201         /* Select the image from which we are extracting files */
1202         ret = select_wim_image(wim, image);
1203         if (ret)
1204                 return ret;
1205
1206         /* Make sure there are no streams in the WIM that have not been
1207          * checksummed yet. */
1208         ret = wim_checksum_unhashed_streams(wim);
1209         if (ret)
1210                 return ret;
1211
1212         /* Check for problems with the extraction commands */
1213         for (size_t i = 0; i < num_cmds; i++) {
1214                 ret = check_extract_command(&cmds[i], wim->hdr.flags);
1215                 if (ret)
1216                         return ret;
1217                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
1218                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
1219                         found_link_cmd = true;
1220                 } else {
1221                         found_nolink_cmd = true;
1222                 }
1223                 if (found_link_cmd && found_nolink_cmd) {
1224                         ERROR("Symlink or hardlink extraction mode must "
1225                               "be set on all extraction commands");
1226                         return WIMLIB_ERR_INVALID_PARAM;
1227                 }
1228         }
1229
1230         /* Execute the extraction commands */
1231         for (size_t i = 0; i < num_cmds; i++) {
1232                 ret = extract_tree(wim,
1233                                    cmds[i].wim_source_path,
1234                                    cmds[i].fs_dest_path,
1235                                    cmds[i].extract_flags,
1236                                    progress_func);
1237                 if (ret)
1238                         return ret;
1239         }
1240         return 0;
1241 }
1242
1243 /* Extract files or directories from a WIM image. */
1244 WIMLIBAPI int
1245 wimlib_extract_files(WIMStruct *wim,
1246                      int image,
1247                      int default_extract_flags,
1248                      const struct wimlib_extract_command *cmds,
1249                      size_t num_cmds,
1250                      WIMStruct **additional_swms,
1251                      unsigned num_additional_swms,
1252                      wimlib_progress_func_t progress_func)
1253 {
1254         int ret;
1255         struct wimlib_extract_command *cmds_copy;
1256         struct wim_lookup_table *wim_tab_save, *joined_tab;
1257         int all_flags = 0;
1258
1259         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
1260
1261         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
1262         if (ret)
1263                 goto out;
1264
1265         if (num_cmds == 0)
1266                 goto out;
1267
1268         if (num_additional_swms) {
1269                 ret = new_joined_lookup_table(wim, additional_swms,
1270                                               num_additional_swms,
1271                                               &joined_tab);
1272                 if (ret)
1273                         goto out;
1274                 wim_tab_save = wim->lookup_table;
1275                 wim->lookup_table = joined_tab;
1276         }
1277
1278         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
1279         if (!cmds_copy) {
1280                 ret = WIMLIB_ERR_NOMEM;
1281                 goto out_restore_lookup_table;
1282         }
1283
1284         for (size_t i = 0; i < num_cmds; i++) {
1285                 cmds_copy[i].extract_flags = (default_extract_flags |
1286                                                  cmds[i].extract_flags)
1287                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
1288                 all_flags |= cmds_copy[i].extract_flags;
1289
1290                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
1291                 if (!cmds_copy[i].wim_source_path) {
1292                         ret = WIMLIB_ERR_NOMEM;
1293                         goto out_free_cmds_copy;
1294                 }
1295
1296                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
1297                 if (!cmds_copy[i].fs_dest_path) {
1298                         ret = WIMLIB_ERR_NOMEM;
1299                         goto out_free_cmds_copy;
1300                 }
1301
1302         }
1303         ret = do_wimlib_extract_files(wim, image,
1304                                       cmds_copy, num_cmds,
1305                                       progress_func);
1306
1307         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
1308                          WIMLIB_EXTRACT_FLAG_HARDLINK))
1309         {
1310                 for_lookup_table_entry(wim->lookup_table,
1311                                        lte_free_extracted_file, NULL);
1312         }
1313 out_free_cmds_copy:
1314         for (size_t i = 0; i < num_cmds; i++) {
1315                 FREE(cmds_copy[i].wim_source_path);
1316                 FREE(cmds_copy[i].fs_dest_path);
1317         }
1318         FREE(cmds_copy);
1319 out_restore_lookup_table:
1320         if (num_additional_swms) {
1321                 free_lookup_table(wim->lookup_table);
1322                 wim->lookup_table = wim_tab_save;
1323         }
1324 out:
1325         return ret;
1326 }
1327
1328 /*
1329  * Extracts an image from a WIM file.
1330  *
1331  * @wim:                WIMStruct for the WIM file.
1332  *
1333  * @image:              Number of the single image to extract.
1334  *
1335  * @target:             Directory or NTFS volume to extract the image to.
1336  *
1337  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
1338  *
1339  * @progress_func:      If non-NULL, a progress function to be called
1340  *                      periodically.
1341  *
1342  * Returns 0 on success; nonzero on failure.
1343  */
1344 static int
1345 extract_single_image(WIMStruct *wim, int image,
1346                      const tchar *target, int extract_flags,
1347                      wimlib_progress_func_t progress_func)
1348 {
1349         int ret;
1350         tchar *target_copy = canonicalize_fs_path(target);
1351         if (!target_copy)
1352                 return WIMLIB_ERR_NOMEM;
1353         struct wimlib_extract_command cmd = {
1354                 .wim_source_path = T(""),
1355                 .fs_dest_path = target_copy,
1356                 .extract_flags = extract_flags,
1357         };
1358         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
1359         FREE(target_copy);
1360         return ret;
1361 }
1362
1363 static const tchar * const filename_forbidden_chars =
1364 T(
1365 #ifdef __WIN32__
1366 "<>:\"/\\|?*"
1367 #else
1368 "/"
1369 #endif
1370 );
1371
1372 /* This function checks if it is okay to use a WIM image's name as a directory
1373  * name.  */
1374 static bool
1375 image_name_ok_as_dir(const tchar *image_name)
1376 {
1377         return image_name && *image_name &&
1378                 !tstrpbrk(image_name, filename_forbidden_chars);
1379 }
1380
1381 /* Extracts all images from the WIM to the directory @target, with the images
1382  * placed in subdirectories named by their image names. */
1383 static int
1384 extract_all_images(WIMStruct *wim,
1385                    const tchar *target,
1386                    int extract_flags,
1387                    wimlib_progress_func_t progress_func)
1388 {
1389         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
1390         size_t output_path_len = tstrlen(target);
1391         tchar buf[output_path_len + 1 + image_name_max_len + 1];
1392         int ret;
1393         int image;
1394         const tchar *image_name;
1395
1396         ret = extract_directory(NULL, target, true, 0);
1397         if (ret)
1398                 return ret;
1399
1400         tmemcpy(buf, target, output_path_len);
1401         buf[output_path_len] = T('/');
1402         for (image = 1; image <= wim->hdr.image_count; image++) {
1403                 image_name = wimlib_get_image_name(wim, image);
1404                 if (image_name_ok_as_dir(image_name)) {
1405                         tstrcpy(buf + output_path_len + 1, image_name);
1406                 } else {
1407                         /* Image name is empty or contains forbidden characters.
1408                          * Use image number instead. */
1409                         tsprintf(buf + output_path_len + 1, T("%d"), image);
1410                 }
1411                 ret = extract_single_image(wim, image, buf, extract_flags,
1412                                            progress_func);
1413                 if (ret)
1414                         return ret;
1415         }
1416         return 0;
1417 }
1418
1419 /* Extracts a single image or all images from a WIM file to a directory or NTFS
1420  * volume. */
1421 WIMLIBAPI int
1422 wimlib_extract_image(WIMStruct *wim,
1423                      int image,
1424                      const tchar *target,
1425                      int extract_flags,
1426                      WIMStruct **additional_swms,
1427                      unsigned num_additional_swms,
1428                      wimlib_progress_func_t progress_func)
1429 {
1430         struct wim_lookup_table *joined_tab, *wim_tab_save;
1431         int ret;
1432
1433         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
1434
1435         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
1436         if (ret)
1437                 return ret;
1438
1439         if (num_additional_swms) {
1440                 ret = new_joined_lookup_table(wim, additional_swms,
1441                                               num_additional_swms, &joined_tab);
1442                 if (ret)
1443                         return ret;
1444                 wim_tab_save = wim->lookup_table;
1445                 wim->lookup_table = joined_tab;
1446         }
1447
1448         if (image == WIMLIB_ALL_IMAGES) {
1449                 extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
1450                 ret = extract_all_images(wim, target, extract_flags,
1451                                          progress_func);
1452         } else {
1453                 extract_flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
1454                 ret = extract_single_image(wim, image, target, extract_flags,
1455                                            progress_func);
1456         }
1457
1458         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
1459                              WIMLIB_EXTRACT_FLAG_HARDLINK))
1460         {
1461                 for_lookup_table_entry(wim->lookup_table,
1462                                        lte_free_extracted_file,
1463                                        NULL);
1464         }
1465         if (num_additional_swms) {
1466                 free_lookup_table(wim->lookup_table);
1467                 wim->lookup_table = wim_tab_save;
1468         }
1469         return ret;
1470 }