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 #ifndef __WIN32__
  61
  62 /* Returns the number of components of @path.  */
  63 static unsigned
  64 get_num_path_components(const char *path)
  65 {
  66         unsigned num_components = 0;
  67         while (*path) {
  68                 while (*path == '/')
  69                         path++;
  70                 if (*path)
  71                         num_components++;
  72                 while (*path && *path != '/')
  73                         path++;
  74         }
  75         return num_components;
  76 }
  77
  78 static const char *
  79 path_next_part(const char *path)
  80 {
  81         while (*path && *path != '/')
  82                 path++;
  83         while (*path && *path == '/')
  84                 path++;
  85         return path;
  86 }
  87
  88 static int
  89 extract_regular_file_linked(struct wim_dentry *dentry,
  90                             const char *output_path,
  91                             struct apply_args *args,
  92                             struct wim_lookup_table_entry *lte)
  93 {
  94         /* This mode overrides the normal hard-link extraction and
  95          * instead either symlinks or hardlinks *all* identical files in
  96          * the WIM, even if they are in a different image (in the case
  97          * of a multi-image extraction) */
  98
  99         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
 100                 if (link(lte->extracted_file, output_path) != 0) {
 101                         ERROR_WITH_ERRNO("Failed to hard link "
 102                                          "`%s' to `%s'",
 103                                          output_path, lte->extracted_file);
 104                         return WIMLIB_ERR_LINK;
 105                 }
 106         } else {
 107                 int num_path_components;
 108                 int num_output_dir_path_components;
 109                 size_t extracted_file_len;
 110                 char *p;
 111                 const char *p2;
 112                 size_t i;
 113
 114                 num_path_components = get_num_path_components(dentry->_full_path) - 1;
 115                 num_output_dir_path_components = get_num_path_components(args->target);
 116
 117                 if (args->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
 118                         num_path_components++;
 119                         num_output_dir_path_components--;
 120                 }
 121                 extracted_file_len = strlen(lte->extracted_file);
 122
 123                 char buf[extracted_file_len + 3 * num_path_components + 1];
 124                 p = &buf[0];
 125
 126                 for (i = 0; i < num_path_components; i++) {
 127                         *p++ = '.';
 128                         *p++ = '.';
 129                         *p++ = '/';
 130                 }
 131                 p2 = lte->extracted_file;
 132                 while (*p2 == '/')
 133                         p2++;
 134                 while (num_output_dir_path_components > 0) {
 135                         p2 = path_next_part(p2);
 136                         num_output_dir_path_components--;
 137                 }
 138                 strcpy(p, p2);
 139                 if (symlink(buf, output_path) != 0) {
 140                         ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 141                                          buf, lte->extracted_file);
 142                         return WIMLIB_ERR_LINK;
 143                 }
 144         }
 145         return 0;
 146 }
 147
 148 static int
 149 symlink_apply_unix_data(const char *link,
 150                         const struct wimlib_unix_data *unix_data)
 151 {
 152         if (lchown(link, unix_data->uid, unix_data->gid)) {
 153                 if (errno == EPERM) {
 154                         /* Ignore */
 155                         WARNING_WITH_ERRNO("failed to set symlink UNIX "
 156                                            "owner/group on \"%s\"", link);
 157                 } else {
 158                         ERROR_WITH_ERRNO("failed to set symlink UNIX "
 159                                          "owner/group on \"%s\"", link);
 160                         return WIMLIB_ERR_INVALID_DENTRY;
 161                 }
 162         }
 163         return 0;
 164 }
 165
 166 static int
 167 fd_apply_unix_data(int fd, const char *path,
 168                    const struct wimlib_unix_data *unix_data)
 169 {
 170         if (fchown(fd, unix_data->uid, unix_data->gid)) {
 171                 if (errno == EPERM) {
 172                         WARNING_WITH_ERRNO("failed to set file UNIX "
 173                                            "owner/group on \"%s\"", path);
 174                         /* Ignore? */
 175                 } else {
 176                         ERROR_WITH_ERRNO("failed to set file UNIX "
 177                                          "owner/group on \"%s\"", path);
 178                         return WIMLIB_ERR_INVALID_DENTRY;
 179                 }
 180         }
 181
 182         if (fchmod(fd, unix_data->mode)) {
 183                 if (errno == EPERM) {
 184                         WARNING_WITH_ERRNO("failed to set UNIX file mode "
 185                                            "on \"%s\"", path);
 186                         /* Ignore? */
 187                 } else {
 188                         ERROR_WITH_ERRNO("failed to set UNIX file mode "
 189                                          "on \"%s\"", path);
 190                         return WIMLIB_ERR_INVALID_DENTRY;
 191                 }
 192         }
 193         return 0;
 194 }
 195
 196 static int
 197 dir_apply_unix_data(const char *dir, const struct wimlib_unix_data *unix_data)
 198 {
 199         int dfd = open(dir, O_RDONLY);
 200         int ret;
 201         if (dfd >= 0) {
 202                 ret = fd_apply_unix_data(dfd, dir, unix_data);
 203                 if (close(dfd)) {
 204                         ERROR_WITH_ERRNO("can't close directory `%s'", dir);
 205                         ret = WIMLIB_ERR_MKDIR;
 206                 }
 207         } else {
 208                 ERROR_WITH_ERRNO("can't open directory `%s'", dir);
 209                 ret = WIMLIB_ERR_MKDIR;
 210         }
 211         return ret;
 212 }
 213
 214 static int
 215 extract_regular_file_unlinked(struct wim_dentry *dentry,
 216                               struct apply_args *args,
 217                               const char *output_path,
 218                               struct wim_lookup_table_entry *lte)
 219 {
 220         /* Normal mode of extraction.  Regular files and hard links are
 221          * extracted in the way that they appear in the WIM. */
 222
 223         int out_fd;
 224         int ret;
 225         struct wim_inode *inode = dentry->d_inode;
 226
 227         if (!((args->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE)
 228                 && (args->extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 229                                      WIMLIB_EXTRACT_FLAG_HARDLINK))))
 230         {
 231                 /* If the dentry is part of a hard link set of at least 2
 232                  * dentries and one of the other dentries has already been
 233                  * extracted, make a hard link to the file corresponding to this
 234                  * already-extracted directory.  Otherwise, extract the file and
 235                  * set the inode->i_extracted_file field so that other dentries
 236                  * in the hard link group can link to it. */
 237                 if (inode->i_nlink > 1) {
 238                         if (inode->i_extracted_file) {
 239                                 DEBUG("Extracting hard link `%s' => `%s'",
 240                                       output_path, inode->i_extracted_file);
 241                                 if (link(inode->i_extracted_file, output_path) != 0) {
 242                                         ERROR_WITH_ERRNO("Failed to hard link "
 243                                                          "`%s' to `%s'",
 244                                                          output_path,
 245                                                          inode->i_extracted_file);
 246                                         return WIMLIB_ERR_LINK;
 247                                 }
 248                                 return 0;
 249                         }
 250                         FREE(inode->i_extracted_file);
 251                         inode->i_extracted_file = STRDUP(output_path);
 252                         if (!inode->i_extracted_file) {
 253                                 ERROR("Failed to allocate memory for filename");
 254                                 return WIMLIB_ERR_NOMEM;
 255                         }
 256                 }
 257         }
 258
 259         /* Extract the contents of the file to @output_path. */
 260
 261         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
 262         if (out_fd == -1) {
 263                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
 264                                  output_path);
 265                 return WIMLIB_ERR_OPEN;
 266         }
 267
 268         if (!lte) {
 269                 /* Empty file with no lookup table entry */
 270                 DEBUG("Empty file `%s'.", output_path);
 271                 ret = 0;
 272                 goto out_extract_unix_data;
 273         }
 274
 275         ret = extract_wim_resource_to_fd(lte, out_fd, wim_resource_size(lte));
 276         if (ret) {
 277                 ERROR("Failed to extract resource to `%s'", output_path);
 278                 goto out;
 279         }
 280
 281 out_extract_unix_data:
 282         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 283                 struct wimlib_unix_data unix_data;
 284                 ret = inode_get_unix_data(inode, &unix_data, NULL);
 285                 if (ret > 0)
 286                         ;
 287                 else if (ret < 0)
 288                         ret = 0;
 289                 else
 290                         ret = fd_apply_unix_data(out_fd, output_path, &unix_data);
 291                 if (ret)
 292                         goto out;
 293         }
 294         if (lte)
 295                 args->progress.extract.completed_bytes += wim_resource_size(lte);
 296 out:
 297         if (close(out_fd) != 0) {
 298                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
 299                 if (ret == 0)
 300                         ret = WIMLIB_ERR_WRITE;
 301         }
 302         return ret;
 303 }
 304
 305 static int
 306 extract_regular_file(struct wim_dentry *dentry,
 307                      struct apply_args *args,
 308                      const char *output_path)
 309 {
 310         struct wim_lookup_table_entry *lte;
 311         const struct wim_inode *inode = dentry->d_inode;
 312
 313         lte = inode_unnamed_lte_resolved(inode);
 314
 315         if (lte && (args->extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 316                                            WIMLIB_EXTRACT_FLAG_HARDLINK)))
 317         {
 318                 if (lte->extracted_file) {
 319                         return extract_regular_file_linked(dentry, output_path, args, lte);
 320                 } else {
 321                         lte->extracted_file = STRDUP(output_path);
 322                         if (!lte->extracted_file)
 323                                 return WIMLIB_ERR_NOMEM;
 324                 }
 325         }
 326         return extract_regular_file_unlinked(dentry, args, output_path, lte);
 327 }
 328
 329 static int
 330 extract_symlink(struct wim_dentry *dentry,
 331                 struct apply_args *args,
 332                 const char *output_path)
 333 {
 334         char target[4096 + args->target_realpath_len];
 335         char *fixed_target;
 336         const struct wim_inode *inode = dentry->d_inode;
 337
 338         ssize_t ret = wim_inode_readlink(inode,
 339                                          target + args->target_realpath_len,
 340                                          sizeof(target) - args->target_realpath_len - 1);
 341         struct wim_lookup_table_entry *lte;
 342
 343         if (ret <= 0) {
 344                 ERROR("Could not read the symbolic link from dentry `%s'",
 345                       dentry->_full_path);
 346                 return WIMLIB_ERR_INVALID_DENTRY;
 347         }
 348         target[args->target_realpath_len + ret] = '\0';
 349         if (target[args->target_realpath_len] == '/' &&
 350             args->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)
 351         {
 352                 /* Fix absolute symbolic link target to point into the actual
 353                  * extraction destination */
 354                 memcpy(target, args->target_realpath,
 355                        args->target_realpath_len);
 356                 fixed_target = target;
 357         } else {
 358                 /* Keep same link target */
 359                 fixed_target = target + args->target_realpath_len;
 360         }
 361         ret = symlink(fixed_target, output_path);
 362         if (ret) {
 363                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 364                                  output_path, fixed_target);
 365                 return WIMLIB_ERR_LINK;
 366         }
 367         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 368                 struct wimlib_unix_data unix_data;
 369                 ret = inode_get_unix_data(inode, &unix_data, NULL);
 370                 if (ret > 0)
 371                         ;
 372                 else if (ret < 0)
 373                         ret = 0;
 374                 else
 375                         ret = symlink_apply_unix_data(output_path, &unix_data);
 376                 if (ret)
 377                         return ret;
 378         }
 379         lte = inode_unnamed_lte_resolved(inode);
 380         wimlib_assert(lte != NULL);
 381         args->progress.extract.completed_bytes += wim_resource_size(lte);
 382         return 0;
 383 }
 384
 385 #endif /* !__WIN32__ */
 386
 387 static int
 388 extract_directory(struct wim_dentry *dentry,
 389                   const tchar *output_path, bool is_root)
 390 {
 391         int ret;
 392         struct stat stbuf;
 393
 394         ret = tstat(output_path, &stbuf);
 395         if (ret == 0) {
 396                 if (S_ISDIR(stbuf.st_mode)) {
 397                         /*if (!is_root)*/
 398                                 /*WARNING("`%s' already exists", output_path);*/
 399                         goto dir_exists;
 400                 } else {
 401                         ERROR("`%"TS"' is not a directory", output_path);
 402                         return WIMLIB_ERR_MKDIR;
 403                 }
 404         } else {
 405                 if (errno != ENOENT) {
 406                         ERROR_WITH_ERRNO("Failed to stat `%"TS"'", output_path);
 407                         return WIMLIB_ERR_STAT;
 408                 }
 409         }
 410
 411         if (tmkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH))
 412         {
 413                 ERROR_WITH_ERRNO("Cannot create directory `%"TS"'", output_path);
 414                 return WIMLIB_ERR_MKDIR;
 415         }
 416 dir_exists:
 417         ret = 0;
 418 #ifndef __WIN32__
 419         if (dentry) {
 420                 struct wimlib_unix_data unix_data;
 421                 ret = inode_get_unix_data(dentry->d_inode, &unix_data, NULL);
 422                 if (ret > 0)
 423                         ;
 424                 else if (ret < 0)
 425                         ret = 0;
 426                 else
 427                         ret = dir_apply_unix_data(output_path, &unix_data);
 428         }
 429 #endif
 430         return ret;
 431 }
 432
 433 #ifndef __WIN32__
 434 static int
 435 unix_do_apply_dentry(const char *output_path, size_t output_path_len,
 436                      struct wim_dentry *dentry, struct apply_args *args)
 437 {
 438         const struct wim_inode *inode = dentry->d_inode;
 439
 440         if (inode_is_symlink(inode))
 441                 return extract_symlink(dentry, args, output_path);
 442         else if (inode_is_directory(inode))
 443                 return extract_directory((args->extract_flags &
 444                                            WIMLIB_EXTRACT_FLAG_UNIX_DATA) ? dentry : NULL,
 445                                          output_path, false);
 446         else
 447                 return extract_regular_file(dentry, args, output_path);
 448 }
 449
 450 static int
 451 unix_do_apply_dentry_timestamps(const char *output_path,
 452                                 size_t output_path_len,
 453                                 const struct wim_dentry *dentry,
 454                                 struct apply_args *args)
 455 {
 456         int ret;
 457         const struct wim_inode *inode = dentry->d_inode;
 458
 459 #ifdef HAVE_UTIMENSAT
 460         /* Convert the WIM timestamps, which are accurate to 100 nanoseconds,
 461          * into `struct timespec's for passing to utimensat(), which is accurate
 462          * to 1 nanosecond. */
 463
 464         struct timespec ts[2];
 465         ts[0] = wim_timestamp_to_timespec(inode->i_last_access_time);
 466         ts[1] = wim_timestamp_to_timespec(inode->i_last_write_time);
 467         ret = utimensat(AT_FDCWD, output_path, ts, AT_SYMLINK_NOFOLLOW);
 468         if (ret)
 469                 ret = errno;
 470 #else
 471         ret = ENOSYS;
 472 #endif
 473
 474         if (ret == ENOSYS) {
 475                 /* utimensat() not implemented or not available */
 476         #ifdef HAVE_LUTIMES
 477                 /* Convert the WIM timestamps, which are accurate to 100
 478                  * nanoseconds, into `struct timeval's for passing to lutimes(),
 479                  * which is accurate to 1 microsecond. */
 480                 struct timeval tv[2];
 481                 tv[0] = wim_timestamp_to_timeval(inode->i_last_access_time);
 482                 tv[1] = wim_timestamp_to_timeval(inode->i_last_write_time);
 483                 ret = lutimes(output_path, tv);
 484                 if (ret)
 485                         ret = errno;
 486         #endif
 487         }
 488
 489         if (ret == ENOSYS) {
 490                 /* utimensat() and lutimes() both not implemented or not
 491                  * available */
 492         #ifdef HAVE_UTIME
 493                 /* Convert the WIM timestamps, which are accurate to 100
 494                  * nanoseconds, into a `struct utimbuf's for passing to
 495                  * utime(), which is accurate to 1 second. */
 496                 struct utimbuf buf;
 497                 buf.actime = wim_timestamp_to_unix(inode->i_last_access_time);
 498                 buf.modtime = wim_timestamp_to_unix(inode->i_last_write_time);
 499                 ret = utime(output_path, &buf);
 500         #endif
 501         }
 502         if (ret && args->num_utime_warnings < 10) {
 503                 WARNING_WITH_ERRNO("Failed to set timestamp on file `%s'",
 504                                     output_path);
 505                 args->num_utime_warnings++;
 506         }
 507         return 0;
 508 }
 509 #endif /* !__WIN32__ */
 510
 511 /* Extracts a file, directory, or symbolic link from the WIM archive. */
 512 static int
 513 apply_dentry_normal(struct wim_dentry *dentry, void *arg)
 514 {
 515         struct apply_args *args = arg;
 516         size_t len;
 517         tchar *output_path;
 518
 519         len = tstrlen(args->target);
 520         if (dentry_is_root(dentry)) {
 521                 output_path = (tchar*)args->target;
 522         } else {
 523                 output_path = alloca(len * sizeof(tchar) + dentry->full_path_nbytes +
 524                                      sizeof(tchar));
 525                 memcpy(output_path, args->target, len * sizeof(tchar));
 526                 memcpy(output_path + len, dentry->_full_path, dentry->full_path_nbytes);
 527                 len += dentry->full_path_nbytes / sizeof(tchar);
 528                 output_path[len] = T('\0');
 529         }
 530 #ifdef __WIN32__
 531         return win32_do_apply_dentry(output_path, len, dentry, args);
 532 #else
 533         return unix_do_apply_dentry(output_path, len, dentry, args);
 534 #endif
 535 }
 536
 537
 538 /* Apply timestamps to an extracted file or directory */
 539 static int
 540 apply_dentry_timestamps_normal(struct wim_dentry *dentry, void *arg)
 541 {
 542         struct apply_args *args = arg;
 543         size_t len;
 544         tchar *output_path;
 545
 546         len = tstrlen(args->target);
 547         if (dentry_is_root(dentry)) {
 548                 output_path = (tchar*)args->target;
 549         } else {
 550                 output_path = alloca(len * sizeof(tchar) + dentry->full_path_nbytes +
 551                                      sizeof(tchar));
 552                 memcpy(output_path, args->target, len * sizeof(tchar));
 553                 memcpy(output_path + len, dentry->_full_path, dentry->full_path_nbytes);
 554                 len += dentry->full_path_nbytes / sizeof(tchar);
 555                 output_path[len] = T('\0');
 556         }
 557
 558 #ifdef __WIN32__
 559         return win32_do_apply_dentry_timestamps(output_path, len, dentry, args);
 560 #else
 561         return unix_do_apply_dentry_timestamps(output_path, len, dentry, args);
 562 #endif
 563 }
 564
 565 /* Extract a dentry if it hasn't already been extracted and either
 566  * WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified, or the dentry is a directory
 567  * and/or has no unnamed stream. */
 568 static int
 569 maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
 570 {
 571         struct apply_args *args = arg;
 572         int ret;
 573
 574         if (dentry->is_extracted)
 575                 return 0;
 576
 577         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS &&
 578             !dentry_is_directory(dentry) &&
 579             inode_unnamed_lte_resolved(dentry->d_inode) != NULL)
 580                 return 0;
 581
 582         if ((args->extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) &&
 583              args->progress_func) {
 584                 args->progress.extract.cur_path = dentry->_full_path;
 585                 args->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY,
 586                                     &args->progress);
 587         }
 588         ret = args->apply_dentry(dentry, args);
 589         if (ret == 0)
 590                 dentry->is_extracted = 1;
 591         return ret;
 592 }
 593
 594 static void
 595 calculate_bytes_to_extract(struct list_head *stream_list,
 596                            int extract_flags,
 597                            union wimlib_progress_info *progress)
 598 {
 599         struct wim_lookup_table_entry *lte;
 600         u64 total_bytes = 0;
 601         u64 num_streams = 0;
 602
 603         /* For each stream to be extracted... */
 604         list_for_each_entry(lte, stream_list, extraction_list) {
 605                 if (extract_flags &
 606                     (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 607                 {
 608                         /* In the symlink or hard link extraction mode, each
 609                          * stream will be extracted one time regardless of how
 610                          * many dentries share the stream. */
 611                         wimlib_assert(!(extract_flags & WIMLIB_EXTRACT_FLAG_NTFS));
 612                         if (!lte->extracted_file) {
 613                                 num_streams++;
 614                                 total_bytes += wim_resource_size(lte);
 615                         }
 616                 } else {
 617                         num_streams += lte->out_refcnt;
 618                         total_bytes += lte->out_refcnt * wim_resource_size(lte);
 619                 }
 620         }
 621         progress->extract.num_streams = num_streams;
 622         progress->extract.total_bytes = total_bytes;
 623         progress->extract.completed_bytes = 0;
 624 }
 625
 626 static void
 627 maybe_add_stream_for_extraction(struct wim_lookup_table_entry *lte,
 628                                 struct list_head *stream_list)
 629 {
 630         if (++lte->out_refcnt == 1) {
 631                 INIT_LIST_HEAD(&lte->inode_list);
 632                 list_add_tail(&lte->extraction_list, stream_list);
 633         }
 634 }
 635
 636 static void
 637 inode_find_streams_for_extraction(struct wim_inode *inode,
 638                                   struct list_head *stream_list,
 639                                   int extract_flags)
 640 {
 641         struct wim_lookup_table_entry *lte;
 642         bool inode_added = false;
 643
 644         lte = inode_unnamed_lte_resolved(inode);
 645         if (lte) {
 646                 maybe_add_stream_for_extraction(lte, stream_list);
 647                 list_add_tail(&inode->i_lte_inode_list, &lte->inode_list);
 648                 inode_added = true;
 649         }
 650
 651         /* Determine whether to include alternate data stream entries or not.
 652          *
 653          * UNIX:  Include them if extracting using NTFS-3g.
 654          *
 655          * Windows: Include them undconditionally, although if the filesystem is
 656          * not NTFS we won't actually be able to extract them. */
 657 #if defined(WITH_NTFS_3G)
 658         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
 659 #elif defined(__WIN32__)
 660         if (1)
 661 #else
 662         if (0)
 663 #endif
 664         {
 665                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
 666                         if (inode->i_ads_entries[i].stream_name_nbytes != 0) {
 667                                 lte = inode->i_ads_entries[i].lte;
 668                                 if (lte) {
 669                                         maybe_add_stream_for_extraction(lte,
 670                                                                         stream_list);
 671                                         if (!inode_added) {
 672                                                 list_add_tail(&inode->i_lte_inode_list,
 673                                                               &lte->inode_list);
 674                                                 inode_added = true;
 675                                         }
 676                                 }
 677                         }
 678                 }
 679         }
 680 }
 681
 682 static void
 683 find_streams_for_extraction(struct wim_image_metadata *imd,
 684                             struct list_head *stream_list,
 685                             struct wim_lookup_table *lookup_table,
 686                             int extract_flags)
 687 {
 688         struct wim_inode *inode;
 689         struct wim_dentry *dentry;
 690
 691         for_lookup_table_entry(lookup_table, lte_zero_out_refcnt, NULL);
 692         INIT_LIST_HEAD(stream_list);
 693         image_for_each_inode(inode, imd) {
 694                 if (!inode->i_resolved)
 695                         inode_resolve_ltes(inode, lookup_table);
 696                 inode_for_each_dentry(dentry, inode)
 697                         dentry->is_extracted = 0;
 698                 inode_find_streams_for_extraction(inode, stream_list,
 699                                                   extract_flags);
 700         }
 701 }
 702
 703 struct apply_operations {
 704         int (*apply_dentry)(struct wim_dentry *dentry, void *arg);
 705         int (*apply_dentry_timestamps)(struct wim_dentry *dentry, void *arg);
 706 };
 707
 708 static const struct apply_operations normal_apply_operations = {
 709         .apply_dentry = apply_dentry_normal,
 710         .apply_dentry_timestamps = apply_dentry_timestamps_normal,
 711 };
 712
 713 #ifdef WITH_NTFS_3G
 714 static const struct apply_operations ntfs_apply_operations = {
 715         .apply_dentry = apply_dentry_ntfs,
 716         .apply_dentry_timestamps = apply_dentry_timestamps_ntfs,
 717 };
 718 #endif
 719
 720 static int
 721 apply_stream_list(struct list_head *stream_list,
 722                   struct apply_args *args,
 723                   const struct apply_operations *ops,
 724                   wimlib_progress_func_t progress_func)
 725 {
 726         uint64_t bytes_per_progress = args->progress.extract.total_bytes / 100;
 727         uint64_t next_progress = bytes_per_progress;
 728         struct wim_lookup_table_entry *lte;
 729         struct wim_inode *inode;
 730         struct wim_dentry *dentry;
 731         int ret;
 732
 733         /* This complicated loop is essentially looping through the dentries,
 734          * although dentries may be visited more than once (if a dentry contains
 735          * two different nonempty streams) or not at all (if a dentry contains
 736          * no non-empty streams).
 737          *
 738          * The outer loop is over the distinct streams to be extracted so that
 739          * sequential reading of the WIM can be implemented. */
 740
 741         /* For each distinct stream to be extracted */
 742         list_for_each_entry(lte, stream_list, extraction_list) {
 743                 /* For each inode that contains the stream */
 744                 list_for_each_entry(inode, &lte->inode_list, i_lte_inode_list) {
 745                         /* For each dentry that points to the inode */
 746                         inode_for_each_dentry(dentry, inode) {
 747                                 /* Extract the dentry if it was not already
 748                                  * extracted */
 749                                 ret = maybe_apply_dentry(dentry, args);
 750                                 if (ret)
 751                                         return ret;
 752                                 if (progress_func &&
 753                                     args->progress.extract.completed_bytes >= next_progress)
 754                                 {
 755                                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS,
 756                                                       &args->progress);
 757                                         if (args->progress.extract.completed_bytes >=
 758                                             args->progress.extract.total_bytes)
 759                                         {
 760                                                 next_progress = ~0ULL;
 761                                         } else {
 762                                                 next_progress =
 763                                                         min (args->progress.extract.completed_bytes +
 764                                                              bytes_per_progress,
 765                                                              args->progress.extract.total_bytes);
 766                                         }
 767                                 }
 768                         }
 769                 }
 770         }
 771         return 0;
 772 }
 773
 774 static int
 775 sort_stream_list_by_wim_position(struct list_head *stream_list)
 776 {
 777         struct list_head *cur;
 778         size_t num_streams;
 779         struct wim_lookup_table_entry **array;
 780         size_t i;
 781         size_t array_size;
 782
 783         num_streams = 0;
 784         list_for_each(cur, stream_list)
 785                 num_streams++;
 786         array_size = num_streams * sizeof(array[0]);
 787         array = MALLOC(array_size);
 788         if (!array) {
 789                 ERROR("Failed to allocate %zu bytes to sort stream entries",
 790                       array_size);
 791                 return WIMLIB_ERR_NOMEM;
 792         }
 793         cur = stream_list->next;
 794         for (i = 0; i < num_streams; i++) {
 795                 array[i] = container_of(cur, struct wim_lookup_table_entry, extraction_list);
 796                 cur = cur->next;
 797         }
 798
 799         qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
 800
 801         INIT_LIST_HEAD(stream_list);
 802         for (i = 0; i < num_streams; i++)
 803                 list_add_tail(&array[i]->extraction_list, stream_list);
 804         FREE(array);
 805         return 0;
 806 }
 807
 808
 809 /* Extracts the image @image from the WIM @w to the directory or NTFS volume
 810  * @target. */
 811 static int
 812 extract_single_image(WIMStruct *w, int image,
 813                      const tchar *target, int extract_flags,
 814                      wimlib_progress_func_t progress_func)
 815 {
 816         int ret;
 817         struct list_head stream_list;
 818
 819         struct apply_args args;
 820         const struct apply_operations *ops;
 821
 822         memset(&args, 0, sizeof(args));
 823
 824         args.w                  = w;
 825         args.target             = target;
 826         args.extract_flags      = extract_flags;
 827         args.progress_func      = progress_func;
 828
 829         if (progress_func) {
 830                 args.progress.extract.wimfile_name = w->filename;
 831                 args.progress.extract.image = image;
 832                 args.progress.extract.extract_flags = (extract_flags &
 833                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
 834                 args.progress.extract.image_name = wimlib_get_image_name(w, image);
 835                 args.progress.extract.target = target;
 836         }
 837
 838 #ifdef WITH_NTFS_3G
 839         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 840                 args.vol = ntfs_mount(target, 0);
 841                 if (!args.vol) {
 842                         ERROR_WITH_ERRNO("Failed to mount NTFS volume `%"TS"'",
 843                                          target);
 844                         return WIMLIB_ERR_NTFS_3G;
 845                 }
 846                 ops = &ntfs_apply_operations;
 847         } else
 848 #endif
 849                 ops = &normal_apply_operations;
 850
 851         ret = select_wim_image(w, image);
 852         if (ret)
 853                 goto out;
 854
 855         /* Build a list of the streams that need to be extracted */
 856         find_streams_for_extraction(wim_get_current_image_metadata(w),
 857                                     &stream_list,
 858                                     w->lookup_table, extract_flags);
 859
 860         /* Calculate the number of bytes of data that will be extracted */
 861         calculate_bytes_to_extract(&stream_list, extract_flags,
 862                                    &args.progress);
 863
 864         if (progress_func) {
 865                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
 866                               &args.progress);
 867         }
 868
 869         /* If a sequential extraction was specified, sort the streams to be
 870          * extracted by their position in the WIM file, so that the WIM file can
 871          * be read sequentially. */
 872         if (extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL) {
 873                 ret = sort_stream_list_by_wim_position(&stream_list);
 874                 if (ret != 0) {
 875                         WARNING("Falling back to non-sequential extraction");
 876                         extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
 877                 }
 878         }
 879
 880         if (progress_func) {
 881                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
 882                               &args.progress);
 883         }
 884
 885         ret = calculate_dentry_tree_full_paths(wim_root_dentry(w));
 886         if (ret)
 887                 goto out;
 888
 889         /* Make the directory structure and extract empty files */
 890         args.extract_flags |= WIMLIB_EXTRACT_FLAG_NO_STREAMS;
 891         args.apply_dentry = ops->apply_dentry;
 892         ret = for_dentry_in_tree(wim_root_dentry(w), maybe_apply_dentry, &args);
 893         args.extract_flags &= ~WIMLIB_EXTRACT_FLAG_NO_STREAMS;
 894         if (ret)
 895                 goto out;
 896
 897         if (progress_func) {
 898                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
 899                               &args.progress);
 900         }
 901
 902         if (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) {
 903                 args.target_realpath = realpath(target, NULL);
 904                 if (!args.target_realpath)
 905                         return WIMLIB_ERR_NOMEM;
 906                 args.target_realpath_len = tstrlen(args.target_realpath);
 907         }
 908
 909         /* Extract non-empty files */
 910         ret = apply_stream_list(&stream_list, &args, ops, progress_func);
 911         if (ret)
 912                 goto out_free_target_realpath;
 913
 914         if (progress_func) {
 915                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
 916                               &args.progress);
 917         }
 918
 919         /* Apply timestamps */
 920         ret = for_dentry_in_tree_depth(wim_root_dentry(w),
 921                                        ops->apply_dentry_timestamps, &args);
 922         if (ret)
 923                 goto out_free_target_realpath;
 924
 925         if (progress_func) {
 926                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
 927                               &args.progress);
 928         }
 929 out_free_target_realpath:
 930         FREE(args.target_realpath);
 931 out:
 932 #ifdef WITH_NTFS_3G
 933         /* Unmount the NTFS volume */
 934         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 935                 if (ntfs_umount(args.vol, FALSE) != 0) {
 936                         ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%"TS"'",
 937                                          args.target);
 938                         if (ret == 0)
 939                                 ret = WIMLIB_ERR_NTFS_3G;
 940                 }
 941         }
 942 #endif
 943         return ret;
 944 }
 945
 946 static const tchar * const filename_forbidden_chars =
 947 T(
 948 #ifdef __WIN32__
 949 "<>:\"/\\|?*"
 950 #else
 951 "/"
 952 #endif
 953 );
 954
 955 /* This function checks if it is okay to use a WIM image's name as a directory
 956  * name.  */
 957 static bool
 958 image_name_ok_as_dir(const tchar *image_name)
 959 {
 960         return image_name && *image_name &&
 961                 !tstrpbrk(image_name, filename_forbidden_chars);
 962 }
 963
 964 /* Extracts all images from the WIM to the directory @target, with the images
 965  * placed in subdirectories named by their image names. */
 966 static int
 967 extract_all_images(WIMStruct *w,
 968                    const tchar *target,
 969                    int extract_flags,
 970                    wimlib_progress_func_t progress_func)
 971 {
 972         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
 973         size_t output_path_len = tstrlen(target);
 974         tchar buf[output_path_len + 1 + image_name_max_len + 1];
 975         int ret;
 976         int image;
 977         const tchar *image_name;
 978
 979         ret = extract_directory(NULL, target, true);
 980         if (ret)
 981                 return ret;
 982
 983         tmemcpy(buf, target, output_path_len);
 984         buf[output_path_len] = T('/');
 985         for (image = 1; image <= w->hdr.image_count; image++) {
 986                 image_name = wimlib_get_image_name(w, image);
 987                 if (image_name_ok_as_dir(image_name)) {
 988                         tstrcpy(buf + output_path_len + 1, image_name);
 989                 } else {
 990                         /* Image name is empty, or contains forbidden
 991                          * characters. */
 992                         tsprintf(buf + output_path_len + 1, T("%d"), image);
 993                 }
 994                 ret = extract_single_image(w, image, buf, extract_flags,
 995                                            progress_func);
 996                 if (ret != 0)
 997                         return ret;
 998         }
 999         return 0;
1000 }
1001
1002 /* Extracts a single image or all images from a WIM file to a directory or NTFS
1003  * volume. */
1004 WIMLIBAPI int
1005 wimlib_extract_image(WIMStruct *w,
1006                      int image,
1007                      const tchar *target,
1008                      int extract_flags,
1009                      WIMStruct **additional_swms,
1010                      unsigned num_additional_swms,
1011                      wimlib_progress_func_t progress_func)
1012 {
1013         struct wim_lookup_table *joined_tab, *w_tab_save;
1014         int ret;
1015
1016         if (!target)
1017                 return WIMLIB_ERR_INVALID_PARAM;
1018
1019         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
1020
1021         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
1022                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
1023                 return WIMLIB_ERR_INVALID_PARAM;
1024
1025 #ifdef __WIN32__
1026         if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
1027                 ERROR("Extracting UNIX data is not supported on Windows");
1028                 return WIMLIB_ERR_INVALID_PARAM;
1029         }
1030         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK)) {
1031                 ERROR("Linked extraction modes are not supported on Windows");
1032                 return WIMLIB_ERR_INVALID_PARAM;
1033         }
1034 #endif
1035
1036         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1037 #ifdef WITH_NTFS_3G
1038                 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))) {
1039                         ERROR("Cannot specify symlink or hardlink flags when applying\n"
1040                               "        directly to a NTFS volume");
1041                         return WIMLIB_ERR_INVALID_PARAM;
1042                 }
1043                 if (image == WIMLIB_ALL_IMAGES) {
1044                         ERROR("Can only apply a single image when applying "
1045                               "directly to a NTFS volume");
1046                         return WIMLIB_ERR_INVALID_PARAM;
1047                 }
1048                 if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
1049                         ERROR("Cannot restore UNIX-specific data in the NTFS extraction mode");
1050                         return WIMLIB_ERR_INVALID_PARAM;
1051                 }
1052 #else
1053                 ERROR("wimlib was compiled without support for NTFS-3g, so");
1054                 ERROR("we cannot apply a WIM image directly to a NTFS volume");
1055                 return WIMLIB_ERR_UNSUPPORTED;
1056 #endif
1057         }
1058
1059         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1060                               WIMLIB_EXTRACT_FLAG_RPFIX)) ==
1061                 (WIMLIB_EXTRACT_FLAG_RPFIX | WIMLIB_EXTRACT_FLAG_NORPFIX))
1062         {
1063                 ERROR("Cannot specify RPFIX and NORPFIX flags at the same time!");
1064                 return WIMLIB_ERR_INVALID_PARAM;
1065         }
1066
1067         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1068                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
1069                 if (w->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1070                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1071
1072         ret = verify_swm_set(w, additional_swms, num_additional_swms);
1073         if (ret)
1074                 return ret;
1075
1076         ret = wim_checksum_unhashed_streams(w);
1077         if (ret)
1078                 return ret;
1079
1080         if (num_additional_swms) {
1081                 ret = new_joined_lookup_table(w, additional_swms,
1082                                               num_additional_swms, &joined_tab);
1083                 if (ret)
1084                         return ret;
1085                 w_tab_save = w->lookup_table;
1086                 w->lookup_table = joined_tab;
1087         }
1088
1089         if (image == WIMLIB_ALL_IMAGES) {
1090                 extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
1091                 ret = extract_all_images(w, target, extract_flags,
1092                                          progress_func);
1093         } else {
1094                 extract_flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
1095                 ret = extract_single_image(w, image, target, extract_flags,
1096                                            progress_func);
1097         }
1098
1099         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
1100                              WIMLIB_EXTRACT_FLAG_HARDLINK))
1101         {
1102                 for_lookup_table_entry(w->lookup_table,
1103                                        lte_free_extracted_file,
1104                                        NULL);
1105         }
1106
1107         if (num_additional_swms) {
1108                 free_lookup_table(w->lookup_table);
1109                 w->lookup_table = w_tab_save;
1110         }
1111         return ret;
1112 }