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         /*print_dentry(dentry, NULL);*/
  63         /*ERROR("%"TS" %"TS, args->target, dentry->_full_path);*/
  64         /*ERROR("");*/
  65
  66         tmemcpy(p, args->target, args->target_nchars);
  67         p += args->target_nchars;
  68
  69         if (dentry != args->extract_root) {
  70                 *p++ = T('/');
  71                 tmemcpy(p, full_path + args->wim_source_path_nchars,
  72                         full_path_nchars - args->wim_source_path_nchars);
  73                 p += full_path_nchars - args->wim_source_path_nchars;
  74         }
  75         *p = T('\0');
  76         return (*apply_dentry_func)(output_path, p - output_path,
  77                                     dentry, args);
  78 }
  79
  80
  81 /* Extracts a file, directory, or symbolic link from the WIM archive. */
  82 static int
  83 apply_dentry_normal(struct wim_dentry *dentry, void *arg)
  84 {
  85 #ifdef __WIN32__
  86         return do_apply_op(dentry, arg, win32_do_apply_dentry);
  87 #else
  88         return do_apply_op(dentry, arg, unix_do_apply_dentry);
  89 #endif
  90 }
  91
  92
  93 /* Apply timestamps to an extracted file or directory */
  94 static int
  95 apply_dentry_timestamps_normal(struct wim_dentry *dentry, void *arg)
  96 {
  97 #ifdef __WIN32__
  98         return do_apply_op(dentry, arg, win32_do_apply_dentry_timestamps);
  99 #else
 100         return do_apply_op(dentry, arg, unix_do_apply_dentry_timestamps);
 101 #endif
 102 }
 103
 104 static bool
 105 dentry_is_descendent(const struct wim_dentry *dentry,
 106                      const struct wim_dentry *ancestor)
 107 {
 108         for (;;) {
 109                 if (dentry == ancestor)
 110                         return true;
 111                 if (dentry_is_root(dentry))
 112                         return false;
 113                 dentry = dentry->parent;
 114         }
 115 }
 116
 117 /* Extract a dentry if it hasn't already been extracted and either
 118  * WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified, or the dentry is a directory
 119  * and/or has no unnamed stream. */
 120 static int
 121 maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
 122 {
 123         struct apply_args *args = arg;
 124         int ret;
 125
 126         if (dentry->is_extracted)
 127                 return 0;
 128
 129         if (!dentry_is_descendent(dentry, args->extract_root))
 130                 return 0;
 131
 132         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS &&
 133             !dentry_is_directory(dentry) &&
 134             inode_unnamed_lte_resolved(dentry->d_inode) != NULL)
 135                 return 0;
 136
 137         if ((args->extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) &&
 138              args->progress_func) {
 139                 args->progress.extract.cur_path = dentry->_full_path;
 140                 args->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY,
 141                                     &args->progress);
 142         }
 143         ret = args->apply_dentry(dentry, args);
 144         if (ret == 0)
 145                 dentry->is_extracted = 1;
 146         return ret;
 147 }
 148
 149 static void
 150 calculate_bytes_to_extract(struct list_head *stream_list,
 151                            int extract_flags,
 152                            union wimlib_progress_info *progress)
 153 {
 154         struct wim_lookup_table_entry *lte;
 155         u64 total_bytes = 0;
 156         u64 num_streams = 0;
 157
 158         /* For each stream to be extracted... */
 159         list_for_each_entry(lte, stream_list, extraction_list) {
 160                 if (extract_flags &
 161                     (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 162                 {
 163                         /* In the symlink or hard link extraction mode, each
 164                          * stream will be extracted one time regardless of how
 165                          * many dentries share the stream. */
 166                         wimlib_assert(!(extract_flags & WIMLIB_EXTRACT_FLAG_NTFS));
 167                         if (!lte->extracted_file) {
 168                                 num_streams++;
 169                                 total_bytes += wim_resource_size(lte);
 170                         }
 171                 } else {
 172                         num_streams += lte->out_refcnt;
 173                         total_bytes += lte->out_refcnt * wim_resource_size(lte);
 174                 }
 175         }
 176         progress->extract.num_streams = num_streams;
 177         progress->extract.total_bytes = total_bytes;
 178         progress->extract.completed_bytes = 0;
 179 }
 180
 181 static void
 182 maybe_add_stream_for_extraction(struct wim_lookup_table_entry *lte,
 183                                 struct list_head *stream_list)
 184 {
 185         if (++lte->out_refcnt == 1) {
 186                 INIT_LIST_HEAD(&lte->inode_list);
 187                 list_add_tail(&lte->extraction_list, stream_list);
 188         }
 189 }
 190
 191 static void
 192 inode_find_streams_for_extraction(struct wim_inode *inode,
 193                                   struct list_head *stream_list,
 194                                   int extract_flags)
 195 {
 196         struct wim_lookup_table_entry *lte;
 197         bool inode_added = false;
 198
 199         lte = inode_unnamed_lte_resolved(inode);
 200         if (lte) {
 201                 maybe_add_stream_for_extraction(lte, stream_list);
 202                 list_add_tail(&inode->i_lte_inode_list, &lte->inode_list);
 203                 inode_added = true;
 204         }
 205
 206         /* Determine whether to include alternate data stream entries or not.
 207          *
 208          * UNIX:  Include them if extracting using NTFS-3g.
 209          *
 210          * Windows: Include them undconditionally, although if the filesystem is
 211          * not NTFS we won't actually be able to extract them. */
 212 #if defined(WITH_NTFS_3G)
 213         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
 214 #elif defined(__WIN32__)
 215         if (1)
 216 #else
 217         if (0)
 218 #endif
 219         {
 220                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
 221                         if (inode->i_ads_entries[i].stream_name_nbytes != 0) {
 222                                 lte = inode->i_ads_entries[i].lte;
 223                                 if (lte) {
 224                                         maybe_add_stream_for_extraction(lte,
 225                                                                         stream_list);
 226                                         if (!inode_added) {
 227                                                 list_add_tail(&inode->i_lte_inode_list,
 228                                                               &lte->inode_list);
 229                                                 inode_added = true;
 230                                         }
 231                                 }
 232                         }
 233                 }
 234         }
 235 }
 236
 237 struct find_streams_ctx {
 238         struct list_head stream_list;
 239         int extract_flags;
 240 };
 241
 242 static int
 243 dentry_find_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
 244 {
 245         struct find_streams_ctx *ctx = _ctx;
 246         struct wim_inode *inode = dentry->d_inode;
 247
 248         dentry->is_extracted = 0;
 249         if (!inode->i_visited) {
 250                 inode_find_streams_for_extraction(inode, &ctx->stream_list,
 251                                                   ctx->extract_flags);
 252                 inode->i_visited = 1;
 253         }
 254         return 0;
 255 }
 256
 257 static int
 258 dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_lookup_table)
 259 {
 260         struct wim_inode *inode = dentry->d_inode;
 261         struct wim_lookup_table *lookup_table = _lookup_table;
 262         struct wim_lookup_table_entry *lte;
 263
 264         inode_resolve_ltes(inode, lookup_table);
 265         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 266                 lte = inode_stream_lte_resolved(inode, i);
 267                 if (lte)
 268                         lte->out_refcnt = 0;
 269         }
 270         return 0;
 271 }
 272
 273 static void
 274 find_streams_for_extraction(struct wim_dentry *root,
 275                             struct list_head *stream_list,
 276                             struct wim_lookup_table *lookup_table,
 277                             int extract_flags)
 278 {
 279         struct find_streams_ctx ctx;
 280
 281         INIT_LIST_HEAD(&ctx.stream_list);
 282         ctx.extract_flags = extract_flags;
 283         for_dentry_in_tree(root, dentry_resolve_and_zero_lte_refcnt, lookup_table);
 284         for_dentry_in_tree(root, dentry_find_streams_to_extract, &ctx);
 285         list_transfer(&ctx.stream_list, stream_list);
 286 }
 287
 288 static int
 289 dentry_mark_inode_unvisited(struct wim_dentry *dentry, void *_ignore)
 290 {
 291         dentry->d_inode->i_visited = 0;
 292         return 0;
 293 }
 294
 295 struct apply_operations {
 296         int (*apply_dentry)(struct wim_dentry *dentry, void *arg);
 297         int (*apply_dentry_timestamps)(struct wim_dentry *dentry, void *arg);
 298 };
 299
 300 static const struct apply_operations normal_apply_operations = {
 301         .apply_dentry = apply_dentry_normal,
 302         .apply_dentry_timestamps = apply_dentry_timestamps_normal,
 303 };
 304
 305 #ifdef WITH_NTFS_3G
 306 static const struct apply_operations ntfs_apply_operations = {
 307         .apply_dentry = apply_dentry_ntfs,
 308         .apply_dentry_timestamps = apply_dentry_timestamps_ntfs,
 309 };
 310 #endif
 311
 312 static int
 313 apply_stream_list(struct list_head *stream_list,
 314                   struct apply_args *args,
 315                   const struct apply_operations *ops,
 316                   wimlib_progress_func_t progress_func)
 317 {
 318         uint64_t bytes_per_progress = args->progress.extract.total_bytes / 100;
 319         uint64_t next_progress = bytes_per_progress;
 320         struct wim_lookup_table_entry *lte;
 321         struct wim_inode *inode;
 322         struct wim_dentry *dentry;
 323         int ret;
 324
 325         /* This complicated loop is essentially looping through the dentries,
 326          * although dentries may be visited more than once (if a dentry contains
 327          * two different nonempty streams) or not at all (if a dentry contains
 328          * no non-empty streams).
 329          *
 330          * The outer loop is over the distinct streams to be extracted so that
 331          * sequential reading of the WIM can be implemented. */
 332
 333         /* For each distinct stream to be extracted */
 334         list_for_each_entry(lte, stream_list, extraction_list) {
 335                 /* For each inode that contains the stream */
 336                 list_for_each_entry(inode, &lte->inode_list, i_lte_inode_list) {
 337                         /* For each dentry that points to the inode */
 338                         inode_for_each_dentry(dentry, inode) {
 339                                 /* Extract the dentry if it was not already
 340                                  * extracted */
 341                                 ret = maybe_apply_dentry(dentry, args);
 342                                 if (ret)
 343                                         return ret;
 344                                 if (progress_func &&
 345                                     args->progress.extract.completed_bytes >= next_progress)
 346                                 {
 347                                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS,
 348                                                       &args->progress);
 349                                         if (args->progress.extract.completed_bytes >=
 350                                             args->progress.extract.total_bytes)
 351                                         {
 352                                                 next_progress = ~0ULL;
 353                                         } else {
 354                                                 next_progress =
 355                                                         min (args->progress.extract.completed_bytes +
 356                                                              bytes_per_progress,
 357                                                              args->progress.extract.total_bytes);
 358                                         }
 359                                 }
 360                         }
 361                 }
 362         }
 363         return 0;
 364 }
 365
 366 static int
 367 sort_stream_list_by_wim_position(struct list_head *stream_list)
 368 {
 369         struct list_head *cur;
 370         size_t num_streams;
 371         struct wim_lookup_table_entry **array;
 372         size_t i;
 373         size_t array_size;
 374
 375         num_streams = 0;
 376         list_for_each(cur, stream_list)
 377                 num_streams++;
 378         array_size = num_streams * sizeof(array[0]);
 379         array = MALLOC(array_size);
 380         if (!array) {
 381                 ERROR("Failed to allocate %zu bytes to sort stream entries",
 382                       array_size);
 383                 return WIMLIB_ERR_NOMEM;
 384         }
 385         cur = stream_list->next;
 386         for (i = 0; i < num_streams; i++) {
 387                 array[i] = container_of(cur, struct wim_lookup_table_entry, extraction_list);
 388                 cur = cur->next;
 389         }
 390
 391         qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
 392
 393         INIT_LIST_HEAD(stream_list);
 394         for (i = 0; i < num_streams; i++)
 395                 list_add_tail(&array[i]->extraction_list, stream_list);
 396         FREE(array);
 397         return 0;
 398 }
 399
 400 /*
 401  * Extract a dentry to standard output.
 402  *
 403  * This obviously doesn't make sense in all cases.  We return an error if the
 404  * dentry does not correspond to a regular file.  Otherwise we extract the
 405  * unnamed data stream only.
 406  */
 407 static int
 408 extract_dentry_to_stdout(struct wim_dentry *dentry)
 409 {
 410         int ret = 0;
 411         if (!dentry_is_regular_file(dentry)) {
 412                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 413                       "extracted to standard output", dentry->_full_path);
 414                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
 415         } else {
 416                 struct wim_lookup_table_entry *lte;
 417
 418                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
 419                 if (lte) {
 420                         ret = extract_wim_resource_to_fd(lte, STDOUT_FILENO,
 421                                                          wim_resource_size(lte));
 422                 }
 423         }
 424         return ret;
 425 }
 426
 427 /*
 428  * extract_tree - Extract a file or directory tree from the currently selected
 429  *                WIM image.
 430  *
 431  * @wim:        WIMStruct for the WIM file, with the desired image selected
 432  *              (as wim->current_image).
 433  * @wim_source_path:
 434  *              "Canonical" (i.e. no leading or trailing slashes, path
 435  *              separators forwald slashes) path inside the WIM image to
 436  *              extract.  An empty string means the full image.
 437  * @target:
 438  *              Filesystem path to extract the file or directory tree to.
 439  *
 440  * @extract_flags:
 441  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
 442  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
 443  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
 444  *              the image.
 445  *
 446  * @progress_func:
 447  *              If non-NULL, progress function for the extraction.  The messages
 448  *              we may in this function are:
 449  *
 450  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
 451  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
 452  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
 453  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
 454  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
 455  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
 456  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
 457  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
 458  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
 459  *
 460  * Returns 0 on success; nonzero on failure.
 461  */
 462 static int
 463 extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
 464              int extract_flags, wimlib_progress_func_t progress_func)
 465 {
 466         int ret;
 467         struct list_head stream_list;
 468         struct apply_args args;
 469         const struct apply_operations *ops;
 470         struct wim_dentry *root;
 471
 472         memset(&args, 0, sizeof(args));
 473
 474         args.w                      = wim;
 475         args.target                 = target;
 476         args.extract_flags          = extract_flags;
 477         args.progress_func          = progress_func;
 478         args.target_nchars          = tstrlen(target);
 479         args.wim_source_path_nchars = tstrlen(wim_source_path);
 480
 481         if (progress_func) {
 482                 args.progress.extract.wimfile_name = wim->filename;
 483                 args.progress.extract.image = wim->current_image;
 484                 args.progress.extract.extract_flags = (extract_flags &
 485                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
 486                 args.progress.extract.image_name = wimlib_get_image_name(wim,
 487                                                                          wim->current_image);
 488                 args.progress.extract.extract_root_wim_source_path = wim_source_path;
 489                 args.progress.extract.target = target;
 490         }
 491
 492 #ifdef WITH_NTFS_3G
 493         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 494                 args.vol = ntfs_mount(target, 0);
 495                 if (!args.vol) {
 496                         ERROR_WITH_ERRNO("Failed to mount NTFS volume `%"TS"'",
 497                                          target);
 498                         ret = WIMLIB_ERR_NTFS_3G;
 499                         goto out;
 500                 }
 501                 ops = &ntfs_apply_operations;
 502         } else
 503 #endif
 504                 ops = &normal_apply_operations;
 505
 506         root = get_dentry(wim, wim_source_path);
 507         if (!root) {
 508                 ERROR("Path \"%"TS"\" does not exist in WIM image %d",
 509                       wim_source_path, wim->current_image);
 510                 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
 511                 goto out_ntfs_umount;
 512         }
 513         args.extract_root = root;
 514
 515         ret = calculate_dentry_tree_full_paths(root);
 516         if (ret)
 517                 goto out_ntfs_umount;
 518
 519
 520         /* Build a list of the streams that need to be extracted */
 521         find_streams_for_extraction(root,
 522                                     &stream_list,
 523                                     wim->lookup_table, extract_flags);
 524
 525         /* Calculate the number of bytes of data that will be extracted */
 526         calculate_bytes_to_extract(&stream_list, extract_flags,
 527                                    &args.progress);
 528
 529         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
 530                 ret = extract_dentry_to_stdout(root);
 531                 goto out_mark_inodes_unvisited;
 532         }
 533
 534         if (progress_func) {
 535                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
 536                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
 537                               &args.progress);
 538         }
 539
 540         /* If a sequential extraction was specified, sort the streams to be
 541          * extracted by their position in the WIM file, so that the WIM file can
 542          * be read sequentially. */
 543         if (extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL) {
 544                 ret = sort_stream_list_by_wim_position(&stream_list);
 545                 if (ret != 0) {
 546                         WARNING("Falling back to non-sequential extraction");
 547                         extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
 548                 }
 549         }
 550
 551         if (progress_func) {
 552                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
 553                               &args.progress);
 554         }
 555
 556         /* Make the directory structure and extract empty files */
 557         args.extract_flags |= WIMLIB_EXTRACT_FLAG_NO_STREAMS;
 558         args.apply_dentry = ops->apply_dentry;
 559         ret = for_dentry_in_tree(root, maybe_apply_dentry, &args);
 560         args.extract_flags &= ~WIMLIB_EXTRACT_FLAG_NO_STREAMS;
 561         if (ret)
 562                 goto out_mark_inodes_unvisited;
 563
 564         if (progress_func) {
 565                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
 566                               &args.progress);
 567         }
 568
 569         if (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) {
 570                 args.target_realpath = realpath(target, NULL);
 571                 if (!args.target_realpath) {
 572                         ret = WIMLIB_ERR_NOMEM;
 573                         goto out_mark_inodes_unvisited;
 574                 }
 575                 args.target_realpath_len = tstrlen(args.target_realpath);
 576         }
 577
 578         /* Extract non-empty files */
 579         ret = apply_stream_list(&stream_list, &args, ops, progress_func);
 580         if (ret)
 581                 goto out_free_target_realpath;
 582
 583         if (progress_func) {
 584                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
 585                               &args.progress);
 586         }
 587
 588         /* Apply timestamps */
 589         ret = for_dentry_in_tree_depth(root,
 590                                        ops->apply_dentry_timestamps, &args);
 591         if (ret)
 592                 goto out_free_target_realpath;
 593
 594         if (progress_func) {
 595                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
 596                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
 597                               &args.progress);
 598         }
 599 out_free_target_realpath:
 600         FREE(args.target_realpath);
 601 out_mark_inodes_unvisited:
 602         for_dentry_in_tree(root, dentry_mark_inode_unvisited, NULL);
 603 out_ntfs_umount:
 604 #ifdef WITH_NTFS_3G
 605         /* Unmount the NTFS volume */
 606         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 607                 if (ntfs_umount(args.vol, FALSE) != 0) {
 608                         ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%"TS"'",
 609                                          args.target);
 610                         if (ret == 0)
 611                                 ret = WIMLIB_ERR_NTFS_3G;
 612                 }
 613         }
 614 #endif
 615 out:
 616         return ret;
 617 }
 618
 619 /* Validates a single wimlib_extract_command, mostly checking to make sure the
 620  * extract flags make sense. */
 621 static int
 622 check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
 623 {
 624         int extract_flags;
 625         bool is_entire_image = (cmd->wim_source_path[0] == T('\0'));
 626
 627         /* Empty destination path? */
 628         if (cmd->fs_dest_path[0] == T('\0'))
 629                 return WIMLIB_ERR_INVALID_PARAM;
 630
 631         extract_flags = cmd->extract_flags;
 632
 633         /* Specified both symlink and hardlink modes? */
 634         if ((extract_flags &
 635              (WIMLIB_EXTRACT_FLAG_SYMLINK |
 636               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
 637                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
 638                 return WIMLIB_ERR_INVALID_PARAM;
 639
 640 #ifdef __WIN32__
 641         /* Wanted UNIX data on Win32? */
 642         if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 643                 ERROR("Extracting UNIX data is not supported on Windows");
 644                 return WIMLIB_ERR_INVALID_PARAM;
 645         }
 646         /* Wanted linked extraction on Windows?  (XXX This is possible, just not
 647          * implemented yet.) */
 648         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 649                              WIMLIB_EXTRACT_FLAG_HARDLINK))
 650         {
 651                 ERROR("Linked extraction modes are not supported on Windows");
 652                 return WIMLIB_ERR_INVALID_PARAM;
 653         }
 654 #endif
 655
 656         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 657                 /* NTFS-3g extraction mode requested */
 658 #ifdef WITH_NTFS_3G
 659                 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 660                                       WIMLIB_EXTRACT_FLAG_HARDLINK))) {
 661                         ERROR("Cannot specify symlink or hardlink flags when applying\n"
 662                               "        directly to a NTFS volume");
 663                         return WIMLIB_ERR_INVALID_PARAM;
 664                 }
 665                 if (!is_entire_image &&
 666                     (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS))
 667                 {
 668                         ERROR("When applying directly to a NTFS volume you can "
 669                               "only extract a full image, not part of one");
 670                         return WIMLIB_ERR_INVALID_PARAM;
 671                 }
 672                 if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 673                         ERROR("Cannot restore UNIX-specific data in "
 674                               "the NTFS extraction mode");
 675                         return WIMLIB_ERR_INVALID_PARAM;
 676                 }
 677 #else
 678                 ERROR("wimlib was compiled without support for NTFS-3g, so");
 679                 ERROR("we cannot apply a WIM image directly to a NTFS volume");
 680                 return WIMLIB_ERR_UNSUPPORTED;
 681 #endif
 682         }
 683
 684         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
 685                               WIMLIB_EXTRACT_FLAG_NORPFIX)) ==
 686                 (WIMLIB_EXTRACT_FLAG_RPFIX | WIMLIB_EXTRACT_FLAG_NORPFIX))
 687         {
 688                 ERROR("Cannot specify RPFIX and NORPFIX flags at the same time!");
 689                 return WIMLIB_ERR_INVALID_PARAM;
 690         }
 691
 692         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
 693                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
 694         {
 695                 /* Do reparse point fixups by default if the WIM header says
 696                  * they are enabled and we are extracting a full image. */
 697                 if ((wim_header_flags & WIM_HDR_FLAG_RP_FIX) && is_entire_image)
 698                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
 699         }
 700
 701         if (!is_entire_image && (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)) {
 702                 ERROR("Cannot specify --rpfix when not extracting entire image");
 703                 return WIMLIB_ERR_INVALID_PARAM;
 704         }
 705
 706         cmd->extract_flags = extract_flags;
 707         return 0;
 708 }
 709
 710
 711 /* Internal function to execute extraction commands for a WIM image. */
 712 static int
 713 do_wimlib_extract_files(WIMStruct *wim,
 714                         int image,
 715                         struct wimlib_extract_command *cmds,
 716                         size_t num_cmds,
 717                         wimlib_progress_func_t progress_func)
 718 {
 719         int ret;
 720         bool found_link_cmd = false;
 721         bool found_nolink_cmd = false;
 722
 723         /* Select the image from which we are extracting files */
 724         ret = select_wim_image(wim, image);
 725         if (ret)
 726                 return ret;
 727
 728         /* Make sure there are no streams in the WIM that have not been
 729          * checksummed yet. */
 730         ret = wim_checksum_unhashed_streams(wim);
 731         if (ret)
 732                 return ret;
 733
 734         /* Check for problems with the extraction commands */
 735         for (size_t i = 0; i < num_cmds; i++) {
 736                 ret = check_extract_command(&cmds[i], wim->hdr.flags);
 737                 if (ret)
 738                         return ret;
 739                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 740                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
 741                         found_link_cmd = true;
 742                 } else {
 743                         found_nolink_cmd = true;
 744                 }
 745                 if (found_link_cmd && found_nolink_cmd) {
 746                         ERROR("Symlink or hardlink extraction mode must "
 747                               "be set on all extraction commands");
 748                         return WIMLIB_ERR_INVALID_PARAM;
 749                 }
 750         }
 751
 752         /* Execute the extraction commands */
 753         for (size_t i = 0; i < num_cmds; i++) {
 754                 ret = extract_tree(wim,
 755                                    cmds[i].wim_source_path,
 756                                    cmds[i].fs_dest_path,
 757                                    cmds[i].extract_flags,
 758                                    progress_func);
 759                 if (ret)
 760                         return ret;
 761         }
 762         return 0;
 763 }
 764
 765 /* Extract files or directories from a WIM image. */
 766 WIMLIBAPI int
 767 wimlib_extract_files(WIMStruct *wim,
 768                      int image,
 769                      int default_extract_flags,
 770                      const struct wimlib_extract_command *cmds,
 771                      size_t num_cmds,
 772                      WIMStruct **additional_swms,
 773                      unsigned num_additional_swms,
 774                      wimlib_progress_func_t progress_func)
 775 {
 776         int ret;
 777         struct wimlib_extract_command *cmds_copy;
 778         struct wim_lookup_table *wim_tab_save, *joined_tab;
 779         int all_flags = 0;
 780
 781         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
 782
 783         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
 784         if (ret)
 785                 goto out;
 786
 787         if (num_cmds == 0)
 788                 goto out;
 789
 790         if (num_additional_swms) {
 791                 ret = new_joined_lookup_table(wim, additional_swms,
 792                                               num_additional_swms,
 793                                               &joined_tab);
 794                 if (ret)
 795                         goto out;
 796                 wim_tab_save = wim->lookup_table;
 797                 wim->lookup_table = joined_tab;
 798         }
 799
 800         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
 801         if (!cmds_copy) {
 802                 ret = WIMLIB_ERR_NOMEM;
 803                 goto out_restore_lookup_table;
 804         }
 805
 806         for (size_t i = 0; i < num_cmds; i++) {
 807                 cmds_copy[i].extract_flags = (default_extract_flags |
 808                                                  cmds[i].extract_flags)
 809                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
 810                 all_flags |= cmds_copy[i].extract_flags;
 811
 812                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
 813                 if (!cmds_copy[i].wim_source_path) {
 814                         ret = WIMLIB_ERR_NOMEM;
 815                         goto out_free_cmds_copy;
 816                 }
 817
 818                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
 819                 if (!cmds_copy[i].fs_dest_path) {
 820                         ret = WIMLIB_ERR_NOMEM;
 821                         goto out_free_cmds_copy;
 822                 }
 823
 824         }
 825         ret = do_wimlib_extract_files(wim, image,
 826                                       cmds_copy, num_cmds,
 827                                       progress_func);
 828
 829         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 830                          WIMLIB_EXTRACT_FLAG_HARDLINK))
 831         {
 832                 for_lookup_table_entry(wim->lookup_table,
 833                                        lte_free_extracted_file, NULL);
 834         }
 835 out_free_cmds_copy:
 836         for (size_t i = 0; i < num_cmds; i++) {
 837                 FREE(cmds_copy[i].wim_source_path);
 838                 FREE(cmds_copy[i].fs_dest_path);
 839         }
 840         FREE(cmds_copy);
 841 out_restore_lookup_table:
 842         if (num_additional_swms) {
 843                 free_lookup_table(wim->lookup_table);
 844                 wim->lookup_table = wim_tab_save;
 845         }
 846 out:
 847         return ret;
 848 }
 849
 850 /*
 851  * Extracts an image from a WIM file.
 852  *
 853  * @wim:                WIMStruct for the WIM file.
 854  *
 855  * @image:              Number of the single image to extract.
 856  *
 857  * @target:             Directory or NTFS volume to extract the image to.
 858  *
 859  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
 860  *
 861  * @progress_func:      If non-NULL, a progress function to be called
 862  *                      periodically.
 863  *
 864  * Returns 0 on success; nonzero on failure.
 865  */
 866 static int
 867 extract_single_image(WIMStruct *wim, int image,
 868                      const tchar *target, int extract_flags,
 869                      wimlib_progress_func_t progress_func)
 870 {
 871         int ret;
 872         tchar *target_copy = canonicalize_fs_path(target);
 873         if (!target_copy)
 874                 return WIMLIB_ERR_NOMEM;
 875         struct wimlib_extract_command cmd = {
 876                 .wim_source_path = T(""),
 877                 .fs_dest_path = target_copy,
 878                 .extract_flags = extract_flags,
 879         };
 880         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
 881         FREE(target_copy);
 882         return ret;
 883 }
 884
 885 static const tchar * const filename_forbidden_chars =
 886 T(
 887 #ifdef __WIN32__
 888 "<>:\"/\\|?*"
 889 #else
 890 "/"
 891 #endif
 892 );
 893
 894 /* This function checks if it is okay to use a WIM image's name as a directory
 895  * name.  */
 896 static bool
 897 image_name_ok_as_dir(const tchar *image_name)
 898 {
 899         return image_name && *image_name &&
 900                 !tstrpbrk(image_name, filename_forbidden_chars);
 901 }
 902
 903 /* Extracts all images from the WIM to the directory @target, with the images
 904  * placed in subdirectories named by their image names. */
 905 static int
 906 extract_all_images(WIMStruct *wim,
 907                    const tchar *target,
 908                    int extract_flags,
 909                    wimlib_progress_func_t progress_func)
 910 {
 911         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
 912         size_t output_path_len = tstrlen(target);
 913         tchar buf[output_path_len + 1 + image_name_max_len + 1];
 914         int ret;
 915         int image;
 916         const tchar *image_name;
 917         struct stat stbuf;
 918
 919         if (tstat(target, &stbuf)) {
 920                 if (errno == ENOENT)
 921                 {
 922                         if (tmkdir(target, S_IRWXU | S_IRGRP | S_IXGRP |
 923                                            S_IROTH | S_IXOTH))
 924                         {
 925                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
 926                                 return WIMLIB_ERR_MKDIR;
 927                         }
 928                 } else {
 929                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
 930                         return WIMLIB_ERR_STAT;
 931                 }
 932         } else if (!S_ISDIR(stbuf.st_mode)) {
 933                 ERROR("\"%"TS"\" is not a directory", target);
 934                 return WIMLIB_ERR_NOTDIR;
 935         }
 936
 937         tmemcpy(buf, target, output_path_len);
 938         buf[output_path_len] = T('/');
 939         for (image = 1; image <= wim->hdr.image_count; image++) {
 940                 image_name = wimlib_get_image_name(wim, image);
 941                 if (image_name_ok_as_dir(image_name)) {
 942                         tstrcpy(buf + output_path_len + 1, image_name);
 943                 } else {
 944                         /* Image name is empty or contains forbidden characters.
 945                          * Use image number instead. */
 946                         tsprintf(buf + output_path_len + 1, T("%d"), image);
 947                 }
 948                 ret = extract_single_image(wim, image, buf, extract_flags,
 949                                            progress_func);
 950                 if (ret)
 951                         return ret;
 952         }
 953         return 0;
 954 }
 955
 956 /* Extracts a single image or all images from a WIM file to a directory or NTFS
 957  * volume. */
 958 WIMLIBAPI int
 959 wimlib_extract_image(WIMStruct *wim,
 960                      int image,
 961                      const tchar *target,
 962                      int extract_flags,
 963                      WIMStruct **additional_swms,
 964                      unsigned num_additional_swms,
 965                      wimlib_progress_func_t progress_func)
 966 {
 967         struct wim_lookup_table *joined_tab, *wim_tab_save;
 968         int ret;
 969
 970         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
 971
 972         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
 973         if (ret)
 974                 return ret;
 975
 976         if (num_additional_swms) {
 977                 ret = new_joined_lookup_table(wim, additional_swms,
 978                                               num_additional_swms, &joined_tab);
 979                 if (ret)
 980                         return ret;
 981                 wim_tab_save = wim->lookup_table;
 982                 wim->lookup_table = joined_tab;
 983         }
 984
 985         if (image == WIMLIB_ALL_IMAGES) {
 986                 extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 987                 ret = extract_all_images(wim, target, extract_flags,
 988                                          progress_func);
 989         } else {
 990                 extract_flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 991                 ret = extract_single_image(wim, image, target, extract_flags,
 992                                            progress_func);
 993         }
 994
 995         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 996                              WIMLIB_EXTRACT_FLAG_HARDLINK))
 997         {
 998                 for_lookup_table_entry(wim->lookup_table,
 999                                        lte_free_extracted_file,
1000                                        NULL);
1001         }
1002         if (num_additional_swms) {
1003                 free_lookup_table(wim->lookup_table);
1004                 wim->lookup_table = wim_tab_save;
1005         }
1006         return ret;
1007 }