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 <stdlib.h>
  29 #include <sys/stat.h>
  30 #include <errno.h>
  31 #include <unistd.h>
  32
  33 #ifdef __WIN32__
  34 #  include "win32.h"
  35 #endif
  36
  37 #include "wimlib_internal.h"
  38 #include "dentry.h"
  39 #include "lookup_table.h"
  40 #include "xml.h"
  41
  42 #ifdef WITH_NTFS_3G
  43 #  include <ntfs-3g/volume.h>
  44 #endif
  45
  46 static int
  47 do_apply_op(struct wim_dentry *dentry, struct apply_args *args,
  48             int (*apply_dentry_func)(const tchar *, size_t,
  49                                      struct wim_dentry *, struct apply_args *))
  50 {
  51         tchar *p;
  52         const tchar *full_path;
  53         size_t full_path_nchars;
  54
  55         wimlib_assert(dentry->_full_path != NULL);
  56         full_path = dentry->_full_path + 1;
  57         full_path_nchars = dentry->full_path_nbytes / sizeof(tchar) - 1;
  58         tchar output_path[args->target_nchars + 1 +
  59                          (full_path_nchars - args->wim_source_path_nchars) + 1];
  60         p = output_path;
  61
  62         tmemcpy(p, args->target, args->target_nchars);
  63         p += args->target_nchars;
  64
  65         if (dentry != args->extract_root) {
  66                 *p++ = T('/');
  67                 tmemcpy(p, full_path + args->wim_source_path_nchars,
  68                         full_path_nchars - args->wim_source_path_nchars);
  69                 p += full_path_nchars - args->wim_source_path_nchars;
  70         }
  71         *p = T('\0');
  72         return (*apply_dentry_func)(output_path, p - output_path,
  73                                     dentry, args);
  74 }
  75
  76
  77 /* Extracts a file, directory, or symbolic link from the WIM archive. */
  78 static int
  79 apply_dentry_normal(struct wim_dentry *dentry, void *arg)
  80 {
  81 #ifdef __WIN32__
  82         return do_apply_op(dentry, arg, win32_do_apply_dentry);
  83 #else
  84         return do_apply_op(dentry, arg, unix_do_apply_dentry);
  85 #endif
  86 }
  87
  88
  89 /* Apply timestamps to an extracted file or directory */
  90 static int
  91 apply_dentry_timestamps_normal(struct wim_dentry *dentry, void *arg)
  92 {
  93 #ifdef __WIN32__
  94         return do_apply_op(dentry, arg, win32_do_apply_dentry_timestamps);
  95 #else
  96         return do_apply_op(dentry, arg, unix_do_apply_dentry_timestamps);
  97 #endif
  98 }
  99
 100 /* Extract a dentry if it hasn't already been extracted and either
 101  * WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified, or the dentry is a directory
 102  * and/or has no unnamed stream. */
 103 static int
 104 maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
 105 {
 106         struct apply_args *args = arg;
 107         int ret;
 108
 109         if (!dentry->needs_extraction)
 110                 return 0;
 111
 112         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS &&
 113             !dentry_is_directory(dentry) &&
 114             inode_unnamed_lte_resolved(dentry->d_inode) != NULL)
 115                 return 0;
 116
 117         if ((args->extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) &&
 118              args->progress_func) {
 119                 args->progress.extract.cur_path = dentry->_full_path;
 120                 args->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY,
 121                                     &args->progress);
 122         }
 123         ret = args->apply_dentry(dentry, args);
 124         if (ret == 0)
 125                 dentry->needs_extraction = 0;
 126         return ret;
 127 }
 128
 129 static void
 130 calculate_bytes_to_extract(struct list_head *stream_list,
 131                            int extract_flags,
 132                            union wimlib_progress_info *progress)
 133 {
 134         struct wim_lookup_table_entry *lte;
 135         u64 total_bytes = 0;
 136         u64 num_streams = 0;
 137
 138         /* For each stream to be extracted... */
 139         list_for_each_entry(lte, stream_list, extraction_list) {
 140                 if (extract_flags &
 141                     (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 142                 {
 143                         /* In the symlink or hard link extraction mode, each
 144                          * stream will be extracted one time regardless of how
 145                          * many dentries share the stream. */
 146                         wimlib_assert(!(extract_flags & WIMLIB_EXTRACT_FLAG_NTFS));
 147                         if (!lte->extracted_file) {
 148                                 num_streams++;
 149                                 total_bytes += wim_resource_size(lte);
 150                         }
 151                 } else {
 152                         num_streams += lte->out_refcnt;
 153                         total_bytes += lte->out_refcnt * wim_resource_size(lte);
 154                 }
 155         }
 156         progress->extract.num_streams = num_streams;
 157         progress->extract.total_bytes = total_bytes;
 158         progress->extract.completed_bytes = 0;
 159 }
 160
 161 static void
 162 maybe_add_stream_for_extraction(struct wim_lookup_table_entry *lte,
 163                                 struct list_head *stream_list)
 164 {
 165         if (++lte->out_refcnt == 1) {
 166                 INIT_LIST_HEAD(&lte->lte_dentry_list);
 167                 list_add_tail(&lte->extraction_list, stream_list);
 168         }
 169 }
 170
 171 struct find_streams_ctx {
 172         struct list_head stream_list;
 173         int extract_flags;
 174 };
 175
 176 static int
 177 dentry_find_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
 178 {
 179         struct find_streams_ctx *ctx = _ctx;
 180         struct wim_inode *inode = dentry->d_inode;
 181         struct wim_lookup_table_entry *lte;
 182         bool dentry_added = false;
 183         struct list_head *stream_list = &ctx->stream_list;
 184         int extract_flags = ctx->extract_flags;
 185
 186         dentry->needs_extraction = 1;
 187
 188         lte = inode_unnamed_lte_resolved(inode);
 189         if (lte) {
 190                 maybe_add_stream_for_extraction(lte, stream_list);
 191                 list_add_tail(&dentry->tmp_list, &lte->lte_dentry_list);
 192                 dentry_added = true;
 193         }
 194
 195         /* Determine whether to include alternate data stream entries or not.
 196          *
 197          * UNIX:  Include them if extracting using NTFS-3g.
 198          *
 199          * Windows: Include them undconditionally, although if the filesystem is
 200          * not NTFS we won't actually be able to extract them. */
 201 #if defined(WITH_NTFS_3G)
 202         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
 203 #elif defined(__WIN32__)
 204         if (1)
 205 #else
 206         if (0)
 207 #endif
 208         {
 209                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
 210                         if (inode->i_ads_entries[i].stream_name_nbytes != 0) {
 211                                 lte = inode->i_ads_entries[i].lte;
 212                                 if (lte) {
 213                                         maybe_add_stream_for_extraction(lte,
 214                                                                         stream_list);
 215                                         if (!dentry_added) {
 216                                                 list_add_tail(&dentry->tmp_list,
 217                                                               &lte->lte_dentry_list);
 218                                                 dentry_added = true;
 219                                         }
 220                                 }
 221                         }
 222                 }
 223         }
 224         return 0;
 225 }
 226
 227 static int
 228 dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_lookup_table)
 229 {
 230         struct wim_inode *inode = dentry->d_inode;
 231         struct wim_lookup_table *lookup_table = _lookup_table;
 232         struct wim_lookup_table_entry *lte;
 233
 234         inode_resolve_ltes(inode, lookup_table);
 235         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 236                 lte = inode_stream_lte_resolved(inode, i);
 237                 if (lte)
 238                         lte->out_refcnt = 0;
 239         }
 240         return 0;
 241 }
 242
 243 static void
 244 find_streams_for_extraction(struct wim_dentry *root,
 245                             struct list_head *stream_list,
 246                             struct wim_lookup_table *lookup_table,
 247                             int extract_flags)
 248 {
 249         struct find_streams_ctx ctx;
 250
 251         INIT_LIST_HEAD(&ctx.stream_list);
 252         ctx.extract_flags = extract_flags;
 253         for_dentry_in_tree(root, dentry_resolve_and_zero_lte_refcnt, lookup_table);
 254         for_dentry_in_tree(root, dentry_find_streams_to_extract, &ctx);
 255         list_transfer(&ctx.stream_list, stream_list);
 256 }
 257
 258 static int
 259 dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
 260 {
 261         dentry->needs_extraction = 0;
 262         return 0;
 263 }
 264
 265 struct apply_operations {
 266         int (*apply_dentry)(struct wim_dentry *dentry, void *arg);
 267         int (*apply_dentry_timestamps)(struct wim_dentry *dentry, void *arg);
 268 };
 269
 270 static const struct apply_operations normal_apply_operations = {
 271         .apply_dentry = apply_dentry_normal,
 272         .apply_dentry_timestamps = apply_dentry_timestamps_normal,
 273 };
 274
 275 #ifdef WITH_NTFS_3G
 276 static const struct apply_operations ntfs_apply_operations = {
 277         .apply_dentry = apply_dentry_ntfs,
 278         .apply_dentry_timestamps = apply_dentry_timestamps_ntfs,
 279 };
 280 #endif
 281
 282 static int
 283 apply_stream_list(struct list_head *stream_list,
 284                   struct apply_args *args,
 285                   const struct apply_operations *ops,
 286                   wimlib_progress_func_t progress_func)
 287 {
 288         uint64_t bytes_per_progress = args->progress.extract.total_bytes / 100;
 289         uint64_t next_progress = bytes_per_progress;
 290         struct wim_lookup_table_entry *lte;
 291         struct wim_dentry *dentry;
 292         int ret;
 293
 294         /* This complicated loop is essentially looping through the dentries,
 295          * although dentries may be visited more than once (if a dentry contains
 296          * two different nonempty streams) or not at all (if a dentry contains
 297          * no non-empty streams).
 298          *
 299          * The outer loop is over the distinct streams to be extracted so that
 300          * sequential reading of the WIM can be implemented. */
 301
 302         /* For each distinct stream to be extracted */
 303         list_for_each_entry(lte, stream_list, extraction_list) {
 304                 /* For each dentry to be extracted that is a name for an inode
 305                  * containing the stream */
 306                 list_for_each_entry(dentry, &lte->lte_dentry_list, tmp_list) {
 307                         /* Extract the dentry if it was not already
 308                          * extracted */
 309                         ret = maybe_apply_dentry(dentry, args);
 310                         if (ret)
 311                                 return ret;
 312                         if (progress_func &&
 313                             args->progress.extract.completed_bytes >= next_progress)
 314                         {
 315                                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS,
 316                                               &args->progress);
 317                                 if (args->progress.extract.completed_bytes >=
 318                                     args->progress.extract.total_bytes)
 319                                 {
 320                                         next_progress = ~0ULL;
 321                                 } else {
 322                                         next_progress =
 323                                                 min (args->progress.extract.completed_bytes +
 324                                                      bytes_per_progress,
 325                                                      args->progress.extract.total_bytes);
 326                                 }
 327                         }
 328                 }
 329         }
 330         return 0;
 331 }
 332
 333 static int
 334 sort_stream_list_by_wim_position(struct list_head *stream_list)
 335 {
 336         struct list_head *cur;
 337         size_t num_streams;
 338         struct wim_lookup_table_entry **array;
 339         size_t i;
 340         size_t array_size;
 341
 342         num_streams = 0;
 343         list_for_each(cur, stream_list)
 344                 num_streams++;
 345         array_size = num_streams * sizeof(array[0]);
 346         array = MALLOC(array_size);
 347         if (!array) {
 348                 ERROR("Failed to allocate %zu bytes to sort stream entries",
 349                       array_size);
 350                 return WIMLIB_ERR_NOMEM;
 351         }
 352         cur = stream_list->next;
 353         for (i = 0; i < num_streams; i++) {
 354                 array[i] = container_of(cur, struct wim_lookup_table_entry, extraction_list);
 355                 cur = cur->next;
 356         }
 357
 358         qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
 359
 360         INIT_LIST_HEAD(stream_list);
 361         for (i = 0; i < num_streams; i++)
 362                 list_add_tail(&array[i]->extraction_list, stream_list);
 363         FREE(array);
 364         return 0;
 365 }
 366
 367 /*
 368  * Extract a dentry to standard output.
 369  *
 370  * This obviously doesn't make sense in all cases.  We return an error if the
 371  * dentry does not correspond to a regular file.  Otherwise we extract the
 372  * unnamed data stream only.
 373  */
 374 static int
 375 extract_dentry_to_stdout(struct wim_dentry *dentry)
 376 {
 377         int ret = 0;
 378         if (!dentry_is_regular_file(dentry)) {
 379                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 380                       "extracted to standard output", dentry->_full_path);
 381                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
 382         } else {
 383                 struct wim_lookup_table_entry *lte;
 384
 385                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
 386                 if (lte) {
 387                         ret = extract_wim_resource_to_fd(lte, STDOUT_FILENO,
 388                                                          wim_resource_size(lte));
 389                 }
 390         }
 391         return ret;
 392 }
 393
 394 /*
 395  * extract_tree - Extract a file or directory tree from the currently selected
 396  *                WIM image.
 397  *
 398  * @wim:        WIMStruct for the WIM file, with the desired image selected
 399  *              (as wim->current_image).
 400  * @wim_source_path:
 401  *              "Canonical" (i.e. no leading or trailing slashes, path
 402  *              separators forwald slashes) path inside the WIM image to
 403  *              extract.  An empty string means the full image.
 404  * @target:
 405  *              Filesystem path to extract the file or directory tree to.
 406  *
 407  * @extract_flags:
 408  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
 409  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
 410  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
 411  *              the image.
 412  *
 413  * @progress_func:
 414  *              If non-NULL, progress function for the extraction.  The messages
 415  *              we may in this function are:
 416  *
 417  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
 418  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
 419  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
 420  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
 421  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
 422  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
 423  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
 424  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
 425  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
 426  *
 427  * Returns 0 on success; nonzero on failure.
 428  */
 429 static int
 430 extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
 431              int extract_flags, wimlib_progress_func_t progress_func)
 432 {
 433         int ret;
 434         struct list_head stream_list;
 435         struct apply_args args;
 436         const struct apply_operations *ops;
 437         struct wim_dentry *root;
 438
 439         memset(&args, 0, sizeof(args));
 440
 441         args.w                      = wim;
 442         args.target                 = target;
 443         args.extract_flags          = extract_flags;
 444         args.progress_func          = progress_func;
 445         args.target_nchars          = tstrlen(target);
 446         args.wim_source_path_nchars = tstrlen(wim_source_path);
 447
 448         if (progress_func) {
 449                 args.progress.extract.wimfile_name = wim->filename;
 450                 args.progress.extract.image = wim->current_image;
 451                 args.progress.extract.extract_flags = (extract_flags &
 452                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
 453                 args.progress.extract.image_name = wimlib_get_image_name(wim,
 454                                                                          wim->current_image);
 455                 args.progress.extract.extract_root_wim_source_path = wim_source_path;
 456                 args.progress.extract.target = target;
 457         }
 458
 459 #ifdef WITH_NTFS_3G
 460         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 461                 args.vol = ntfs_mount(target, 0);
 462                 if (!args.vol) {
 463                         ERROR_WITH_ERRNO("Failed to mount NTFS volume `%"TS"'",
 464                                          target);
 465                         ret = WIMLIB_ERR_NTFS_3G;
 466                         goto out;
 467                 }
 468                 ops = &ntfs_apply_operations;
 469         } else
 470 #endif
 471                 ops = &normal_apply_operations;
 472
 473         root = get_dentry(wim, wim_source_path);
 474         if (!root) {
 475                 ERROR("Path \"%"TS"\" does not exist in WIM image %d",
 476                       wim_source_path, wim->current_image);
 477                 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
 478                 goto out_ntfs_umount;
 479         }
 480         args.extract_root = root;
 481
 482         ret = calculate_dentry_tree_full_paths(root);
 483         if (ret)
 484                 goto out_ntfs_umount;
 485
 486         /* Build a list of the streams that need to be extracted */
 487         find_streams_for_extraction(root,
 488                                     &stream_list,
 489                                     wim->lookup_table, extract_flags);
 490
 491         /* Calculate the number of bytes of data that will be extracted */
 492         calculate_bytes_to_extract(&stream_list, extract_flags,
 493                                    &args.progress);
 494
 495         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
 496                 ret = extract_dentry_to_stdout(root);
 497                 goto out_dentry_reset_needs_extraction;
 498         }
 499
 500         if (progress_func) {
 501                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
 502                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
 503                               &args.progress);
 504         }
 505
 506         /* If a sequential extraction was specified, sort the streams to be
 507          * extracted by their position in the WIM file, so that the WIM file can
 508          * be read sequentially. */
 509         if (extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL) {
 510                 ret = sort_stream_list_by_wim_position(&stream_list);
 511                 if (ret != 0) {
 512                         WARNING("Falling back to non-sequential extraction");
 513                         extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
 514                 }
 515         }
 516
 517         if (progress_func) {
 518                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
 519                               &args.progress);
 520         }
 521
 522         /* Make the directory structure and extract empty files */
 523         args.extract_flags |= WIMLIB_EXTRACT_FLAG_NO_STREAMS;
 524         args.apply_dentry = ops->apply_dentry;
 525         ret = for_dentry_in_tree(root, maybe_apply_dentry, &args);
 526         args.extract_flags &= ~WIMLIB_EXTRACT_FLAG_NO_STREAMS;
 527         if (ret)
 528                 goto out_dentry_reset_needs_extraction;
 529
 530         if (progress_func) {
 531                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
 532                               &args.progress);
 533         }
 534
 535         if (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) {
 536                 args.target_realpath = realpath(target, NULL);
 537                 if (!args.target_realpath) {
 538                         ret = WIMLIB_ERR_NOMEM;
 539                         goto out_dentry_reset_needs_extraction;
 540                 }
 541                 args.target_realpath_len = tstrlen(args.target_realpath);
 542         }
 543
 544         /* Extract non-empty files */
 545         ret = apply_stream_list(&stream_list, &args, ops, progress_func);
 546         if (ret)
 547                 goto out_free_target_realpath;
 548
 549         if (progress_func) {
 550                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
 551                               &args.progress);
 552         }
 553
 554         /* Apply timestamps */
 555         ret = for_dentry_in_tree_depth(root,
 556                                        ops->apply_dentry_timestamps, &args);
 557         if (ret)
 558                 goto out_free_target_realpath;
 559
 560         if (progress_func) {
 561                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
 562                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
 563                               &args.progress);
 564         }
 565 out_free_target_realpath:
 566         FREE(args.target_realpath);
 567 out_dentry_reset_needs_extraction:
 568         for_dentry_in_tree(root, dentry_reset_needs_extraction, NULL);
 569 out_ntfs_umount:
 570 #ifdef WITH_NTFS_3G
 571         /* Unmount the NTFS volume */
 572         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 573                 if (ntfs_umount(args.vol, FALSE) != 0) {
 574                         ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%"TS"'",
 575                                          args.target);
 576                         if (ret == 0)
 577                                 ret = WIMLIB_ERR_NTFS_3G;
 578                 }
 579         }
 580 #endif
 581 out:
 582         return ret;
 583 }
 584
 585 /* Validates a single wimlib_extract_command, mostly checking to make sure the
 586  * extract flags make sense. */
 587 static int
 588 check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
 589 {
 590         int extract_flags;
 591         bool is_entire_image = (cmd->wim_source_path[0] == T('\0'));
 592
 593         /* Empty destination path? */
 594         if (cmd->fs_dest_path[0] == T('\0'))
 595                 return WIMLIB_ERR_INVALID_PARAM;
 596
 597         extract_flags = cmd->extract_flags;
 598
 599         /* Specified both symlink and hardlink modes? */
 600         if ((extract_flags &
 601              (WIMLIB_EXTRACT_FLAG_SYMLINK |
 602               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
 603                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
 604                 return WIMLIB_ERR_INVALID_PARAM;
 605
 606 #ifdef __WIN32__
 607         /* Wanted UNIX data on Windows? */
 608         if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 609                 ERROR("Extracting UNIX data is not supported on Windows");
 610                 return WIMLIB_ERR_INVALID_PARAM;
 611         }
 612         /* Wanted linked extraction on Windows?  (XXX This is possible, just not
 613          * implemented yet.) */
 614         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 615                              WIMLIB_EXTRACT_FLAG_HARDLINK))
 616         {
 617                 ERROR("Linked extraction modes are not supported on Windows");
 618                 return WIMLIB_ERR_INVALID_PARAM;
 619         }
 620 #endif
 621
 622         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 623                 /* NTFS-3g extraction mode requested */
 624 #ifdef WITH_NTFS_3G
 625                 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 626                                       WIMLIB_EXTRACT_FLAG_HARDLINK))) {
 627                         ERROR("Cannot specify symlink or hardlink flags when applying\n"
 628                               "        directly to a NTFS volume");
 629                         return WIMLIB_ERR_INVALID_PARAM;
 630                 }
 631                 if (!is_entire_image &&
 632                     (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS))
 633                 {
 634                         ERROR("When applying directly to a NTFS volume you can "
 635                               "only extract a full image, not part of one");
 636                         return WIMLIB_ERR_INVALID_PARAM;
 637                 }
 638                 if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 639                         ERROR("Cannot restore UNIX-specific data in "
 640                               "the NTFS extraction mode");
 641                         return WIMLIB_ERR_INVALID_PARAM;
 642                 }
 643 #else
 644                 ERROR("wimlib was compiled without support for NTFS-3g, so");
 645                 ERROR("we cannot apply a WIM image directly to a NTFS volume");
 646                 return WIMLIB_ERR_UNSUPPORTED;
 647 #endif
 648         }
 649
 650         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
 651                               WIMLIB_EXTRACT_FLAG_NORPFIX)) ==
 652                 (WIMLIB_EXTRACT_FLAG_RPFIX | WIMLIB_EXTRACT_FLAG_NORPFIX))
 653         {
 654                 ERROR("Cannot specify RPFIX and NORPFIX flags at the same time!");
 655                 return WIMLIB_ERR_INVALID_PARAM;
 656         }
 657
 658         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
 659                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
 660         {
 661                 /* Do reparse point fixups by default if the WIM header says
 662                  * they are enabled and we are extracting a full image. */
 663                 if ((wim_header_flags & WIM_HDR_FLAG_RP_FIX) && is_entire_image)
 664                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
 665         }
 666
 667         if (!is_entire_image && (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)) {
 668                 ERROR("Cannot specify --rpfix when not extracting entire image");
 669                 return WIMLIB_ERR_INVALID_PARAM;
 670         }
 671
 672         cmd->extract_flags = extract_flags;
 673         return 0;
 674 }
 675
 676
 677 /* Internal function to execute extraction commands for a WIM image. */
 678 static int
 679 do_wimlib_extract_files(WIMStruct *wim,
 680                         int image,
 681                         struct wimlib_extract_command *cmds,
 682                         size_t num_cmds,
 683                         wimlib_progress_func_t progress_func)
 684 {
 685         int ret;
 686         bool found_link_cmd = false;
 687         bool found_nolink_cmd = false;
 688
 689         /* Select the image from which we are extracting files */
 690         ret = select_wim_image(wim, image);
 691         if (ret)
 692                 return ret;
 693
 694         /* Make sure there are no streams in the WIM that have not been
 695          * checksummed yet. */
 696         ret = wim_checksum_unhashed_streams(wim);
 697         if (ret)
 698                 return ret;
 699
 700         /* Check for problems with the extraction commands */
 701         for (size_t i = 0; i < num_cmds; i++) {
 702                 ret = check_extract_command(&cmds[i], wim->hdr.flags);
 703                 if (ret)
 704                         return ret;
 705                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 706                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
 707                         found_link_cmd = true;
 708                 } else {
 709                         found_nolink_cmd = true;
 710                 }
 711                 if (found_link_cmd && found_nolink_cmd) {
 712                         ERROR("Symlink or hardlink extraction mode must "
 713                               "be set on all extraction commands");
 714                         return WIMLIB_ERR_INVALID_PARAM;
 715                 }
 716         }
 717
 718         /* Execute the extraction commands */
 719         for (size_t i = 0; i < num_cmds; i++) {
 720                 ret = extract_tree(wim,
 721                                    cmds[i].wim_source_path,
 722                                    cmds[i].fs_dest_path,
 723                                    cmds[i].extract_flags,
 724                                    progress_func);
 725                 if (ret)
 726                         return ret;
 727         }
 728         return 0;
 729 }
 730
 731 /* Extract files or directories from a WIM image. */
 732 WIMLIBAPI int
 733 wimlib_extract_files(WIMStruct *wim,
 734                      int image,
 735                      const struct wimlib_extract_command *cmds,
 736                      size_t num_cmds,
 737                      int default_extract_flags,
 738                      WIMStruct **additional_swms,
 739                      unsigned num_additional_swms,
 740                      wimlib_progress_func_t progress_func)
 741 {
 742         int ret;
 743         struct wimlib_extract_command *cmds_copy;
 744         struct wim_lookup_table *wim_tab_save, *joined_tab;
 745         int all_flags = 0;
 746
 747         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
 748
 749         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
 750         if (ret)
 751                 goto out;
 752
 753         if (num_cmds == 0)
 754                 goto out;
 755
 756         if (num_additional_swms) {
 757                 ret = new_joined_lookup_table(wim, additional_swms,
 758                                               num_additional_swms,
 759                                               &joined_tab);
 760                 if (ret)
 761                         goto out;
 762                 wim_tab_save = wim->lookup_table;
 763                 wim->lookup_table = joined_tab;
 764         }
 765
 766         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
 767         if (!cmds_copy) {
 768                 ret = WIMLIB_ERR_NOMEM;
 769                 goto out_restore_lookup_table;
 770         }
 771
 772         for (size_t i = 0; i < num_cmds; i++) {
 773                 cmds_copy[i].extract_flags = (default_extract_flags |
 774                                                  cmds[i].extract_flags)
 775                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
 776                 all_flags |= cmds_copy[i].extract_flags;
 777
 778                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
 779                 if (!cmds_copy[i].wim_source_path) {
 780                         ret = WIMLIB_ERR_NOMEM;
 781                         goto out_free_cmds_copy;
 782                 }
 783
 784                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
 785                 if (!cmds_copy[i].fs_dest_path) {
 786                         ret = WIMLIB_ERR_NOMEM;
 787                         goto out_free_cmds_copy;
 788                 }
 789
 790         }
 791         ret = do_wimlib_extract_files(wim, image,
 792                                       cmds_copy, num_cmds,
 793                                       progress_func);
 794
 795         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 796                          WIMLIB_EXTRACT_FLAG_HARDLINK))
 797         {
 798                 for_lookup_table_entry(wim->lookup_table,
 799                                        lte_free_extracted_file, NULL);
 800         }
 801 out_free_cmds_copy:
 802         for (size_t i = 0; i < num_cmds; i++) {
 803                 FREE(cmds_copy[i].wim_source_path);
 804                 FREE(cmds_copy[i].fs_dest_path);
 805         }
 806         FREE(cmds_copy);
 807 out_restore_lookup_table:
 808         if (num_additional_swms) {
 809                 free_lookup_table(wim->lookup_table);
 810                 wim->lookup_table = wim_tab_save;
 811         }
 812 out:
 813         return ret;
 814 }
 815
 816 /*
 817  * Extracts an image from a WIM file.
 818  *
 819  * @wim:                WIMStruct for the WIM file.
 820  *
 821  * @image:              Number of the single image to extract.
 822  *
 823  * @target:             Directory or NTFS volume to extract the image to.
 824  *
 825  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
 826  *
 827  * @progress_func:      If non-NULL, a progress function to be called
 828  *                      periodically.
 829  *
 830  * Returns 0 on success; nonzero on failure.
 831  */
 832 static int
 833 extract_single_image(WIMStruct *wim, int image,
 834                      const tchar *target, int extract_flags,
 835                      wimlib_progress_func_t progress_func)
 836 {
 837         int ret;
 838         tchar *target_copy = canonicalize_fs_path(target);
 839         if (!target_copy)
 840                 return WIMLIB_ERR_NOMEM;
 841         struct wimlib_extract_command cmd = {
 842                 .wim_source_path = T(""),
 843                 .fs_dest_path = target_copy,
 844                 .extract_flags = extract_flags,
 845         };
 846         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
 847         FREE(target_copy);
 848         return ret;
 849 }
 850
 851 static const tchar * const filename_forbidden_chars =
 852 T(
 853 #ifdef __WIN32__
 854 "<>:\"/\\|?*"
 855 #else
 856 "/"
 857 #endif
 858 );
 859
 860 /* This function checks if it is okay to use a WIM image's name as a directory
 861  * name.  */
 862 static bool
 863 image_name_ok_as_dir(const tchar *image_name)
 864 {
 865         return image_name && *image_name &&
 866                 !tstrpbrk(image_name, filename_forbidden_chars) &&
 867                 tstrcmp(image_name, T(".")) &&
 868                 tstrcmp(image_name, T(".."));
 869 }
 870
 871 /* Extracts all images from the WIM to the directory @target, with the images
 872  * placed in subdirectories named by their image names. */
 873 static int
 874 extract_all_images(WIMStruct *wim,
 875                    const tchar *target,
 876                    int extract_flags,
 877                    wimlib_progress_func_t progress_func)
 878 {
 879         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
 880         size_t output_path_len = tstrlen(target);
 881         tchar buf[output_path_len + 1 + image_name_max_len + 1];
 882         int ret;
 883         int image;
 884         const tchar *image_name;
 885         struct stat stbuf;
 886
 887         if (tstat(target, &stbuf)) {
 888                 if (errno == ENOENT)
 889                 {
 890                         if (tmkdir(target, S_IRWXU | S_IRGRP | S_IXGRP |
 891                                            S_IROTH | S_IXOTH))
 892                         {
 893                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
 894                                 return WIMLIB_ERR_MKDIR;
 895                         }
 896                 } else {
 897                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
 898                         return WIMLIB_ERR_STAT;
 899                 }
 900         } else if (!S_ISDIR(stbuf.st_mode)) {
 901                 ERROR("\"%"TS"\" is not a directory", target);
 902                 return WIMLIB_ERR_NOTDIR;
 903         }
 904
 905         tmemcpy(buf, target, output_path_len);
 906         buf[output_path_len] = T('/');
 907         for (image = 1; image <= wim->hdr.image_count; image++) {
 908                 image_name = wimlib_get_image_name(wim, image);
 909                 if (image_name_ok_as_dir(image_name)) {
 910                         tstrcpy(buf + output_path_len + 1, image_name);
 911                 } else {
 912                         /* Image name is empty or contains forbidden characters.
 913                          * Use image number instead. */
 914                         tsprintf(buf + output_path_len + 1, T("%d"), image);
 915                 }
 916                 ret = extract_single_image(wim, image, buf, extract_flags,
 917                                            progress_func);
 918                 if (ret)
 919                         return ret;
 920         }
 921         return 0;
 922 }
 923
 924 /* Extracts a single image or all images from a WIM file to a directory or NTFS
 925  * volume. */
 926 WIMLIBAPI int
 927 wimlib_extract_image(WIMStruct *wim,
 928                      int image,
 929                      const tchar *target,
 930                      int extract_flags,
 931                      WIMStruct **additional_swms,
 932                      unsigned num_additional_swms,
 933                      wimlib_progress_func_t progress_func)
 934 {
 935         struct wim_lookup_table *joined_tab, *wim_tab_save;
 936         int ret;
 937
 938         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
 939
 940         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
 941         if (ret)
 942                 return ret;
 943
 944         if (num_additional_swms) {
 945                 ret = new_joined_lookup_table(wim, additional_swms,
 946                                               num_additional_swms, &joined_tab);
 947                 if (ret)
 948                         return ret;
 949                 wim_tab_save = wim->lookup_table;
 950                 wim->lookup_table = joined_tab;
 951         }
 952
 953         if (image == WIMLIB_ALL_IMAGES) {
 954                 ret = extract_all_images(wim, target,
 955                                          extract_flags | WIMLIB_EXTRACT_FLAG_MULTI_IMAGE,
 956                                          progress_func);
 957         } else {
 958                 ret = extract_single_image(wim, image, target, extract_flags,
 959                                            progress_func);
 960         }
 961
 962         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 963                              WIMLIB_EXTRACT_FLAG_HARDLINK))
 964         {
 965                 for_lookup_table_entry(wim->lookup_table,
 966                                        lte_free_extracted_file,
 967                                        NULL);
 968         }
 969         if (num_additional_swms) {
 970                 free_lookup_table(wim->lookup_table);
 971                 wim->lookup_table = wim_tab_save;
 972         }
 973         return ret;
 974 }