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