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

   1 /*
   2  * extract.c
   3  *
   4  * Support for extracting WIM images, or files or directories contained in a WIM
   5  * image.
   6  */
   7
   8 /*
   9  * Copyright (C) 2012, 2013, 2014, 2015 Eric Biggers
  10  *
  11  * This file is free software; you can redistribute it and/or modify it under
  12  * the terms of the GNU Lesser General Public License as published by the Free
  13  * Software Foundation; either version 3 of the License, or (at your option) any
  14  * later version.
  15  *
  16  * This file is distributed in the hope that it will be useful, but WITHOUT
  17  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  18  * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  19  * details.
  20  *
  21  * You should have received a copy of the GNU Lesser General Public License
  22  * along with this file; if not, see http://www.gnu.org/licenses/.
  23  */
  24
  25 /*
  26  * This file provides the API functions wimlib_extract_image(),
  27  * wimlib_extract_image_from_pipe(), wimlib_extract_paths(), and
  28  * wimlib_extract_pathlist().  Internally, all end up calling
  29  * do_wimlib_extract_paths() and extract_trees().
  30  *
  31  * Although wimlib supports multiple extraction modes/backends (NTFS-3g, UNIX,
  32  * Win32), this file does not itself have code to extract files or directories
  33  * to any specific target; instead, it handles generic functionality and relies
  34  * on lower-level callback functions declared in `struct apply_operations' to do
  35  * the actual extraction.
  36  */
  37
  38 #ifdef HAVE_CONFIG_H
  39 #  include "config.h"
  40 #endif
  41
  42 #include <errno.h>
  43 #include <fcntl.h>
  44 #include <sys/stat.h>
  45 #include <unistd.h>
  46
  47 #include "wimlib/apply.h"
  48 #include "wimlib/assert.h"
  49 #include "wimlib/blob_table.h"
  50 #include "wimlib/dentry.h"
  51 #include "wimlib/encoding.h"
  52 #include "wimlib/endianness.h"
  53 #include "wimlib/error.h"
  54 #include "wimlib/metadata.h"
  55 #include "wimlib/pathlist.h"
  56 #include "wimlib/paths.h"
  57 #include "wimlib/pattern.h"
  58 #include "wimlib/reparse.h"
  59 #include "wimlib/resource.h"
  60 #include "wimlib/security.h"
  61 #include "wimlib/unix_data.h"
  62 #include "wimlib/wim.h"
  63 #include "wimlib/win32.h" /* for realpath() equivalent */
  64 #include "wimlib/xml.h"
  65
  66 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x80000000
  67 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x40000000
  68
  69 /* Keep in sync with wimlib.h  */
  70 #define WIMLIB_EXTRACT_MASK_PUBLIC                              \
  71         (WIMLIB_EXTRACT_FLAG_NTFS                       |       \
  72          WIMLIB_EXTRACT_FLAG_UNIX_DATA                  |       \
  73          WIMLIB_EXTRACT_FLAG_NO_ACLS                    |       \
  74          WIMLIB_EXTRACT_FLAG_STRICT_ACLS                |       \
  75          WIMLIB_EXTRACT_FLAG_RPFIX                      |       \
  76          WIMLIB_EXTRACT_FLAG_NORPFIX                    |       \
  77          WIMLIB_EXTRACT_FLAG_TO_STDOUT                  |       \
  78          WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES  |       \
  79          WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS         |       \
  80          WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS          |       \
  81          WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES         |       \
  82          WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS            |       \
  83          WIMLIB_EXTRACT_FLAG_GLOB_PATHS                 |       \
  84          WIMLIB_EXTRACT_FLAG_STRICT_GLOB                |       \
  85          WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES              |       \
  86          WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE  |       \
  87          WIMLIB_EXTRACT_FLAG_WIMBOOT)
  88
  89 /* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or
  90  * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA.  */
  91 int
  92 do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
  93 {
  94         ctx->count_until_file_progress = 500;  /* Arbitrary value to limit calls  */
  95         return extract_progress(ctx, msg);
  96 }
  97
  98 static int
  99 start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg)
 100 {
 101         ctx->progress.extract.current_file_count = 0;
 102         ctx->progress.extract.end_file_count = end_file_count;
 103         return do_file_extract_progress(ctx, msg);
 104 }
 105
 106 int
 107 start_file_structure_phase(struct apply_ctx *ctx, u64 end_file_count)
 108 {
 109         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 110 }
 111
 112 int
 113 start_file_metadata_phase(struct apply_ctx *ctx, u64 end_file_count)
 114 {
 115         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 116 }
 117
 118 static int
 119 end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
 120 {
 121         ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count;
 122         return do_file_extract_progress(ctx, msg);
 123 }
 124
 125 int
 126 end_file_structure_phase(struct apply_ctx *ctx)
 127 {
 128         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 129 }
 130
 131 int
 132 end_file_metadata_phase(struct apply_ctx *ctx)
 133 {
 134         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 135 }
 136
 137 #define PWM_FOUND_WIM_HDR (-1)
 138
 139 /* Read the header for a blob in a pipable WIM.  If @pwm_hdr_ret is not NULL,
 140  * also look for a pipable WIM header and return PWM_FOUND_WIM_HDR if found.  */
 141 static int
 142 read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE],
 143                      struct wim_reshdr *reshdr_ret,
 144                      struct wim_header_disk *pwm_hdr_ret)
 145 {
 146         int ret;
 147         struct pwm_blob_hdr blob_hdr;
 148         u64 magic;
 149
 150         ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr));
 151         if (unlikely(ret))
 152                 goto read_error;
 153
 154         magic = le64_to_cpu(blob_hdr.magic);
 155
 156         if (magic == PWM_MAGIC && pwm_hdr_ret != NULL) {
 157                 memcpy(pwm_hdr_ret, &blob_hdr, sizeof(blob_hdr));
 158                 ret = full_read(&pwm->in_fd,
 159                                 (u8 *)pwm_hdr_ret + sizeof(blob_hdr),
 160                                 sizeof(*pwm_hdr_ret) - sizeof(blob_hdr));
 161                 if (unlikely(ret))
 162                         goto read_error;
 163                 return PWM_FOUND_WIM_HDR;
 164         }
 165
 166         if (unlikely(magic != PWM_BLOB_MAGIC)) {
 167                 ERROR("Data read on pipe is invalid (expected blob header)");
 168                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 169         }
 170
 171         copy_hash(hash_ret, blob_hdr.hash);
 172
 173         reshdr_ret->size_in_wim = 0; /* Not available  */
 174         reshdr_ret->flags = le32_to_cpu(blob_hdr.flags);
 175         reshdr_ret->offset_in_wim = pwm->in_fd.offset;
 176         reshdr_ret->uncompressed_size = le64_to_cpu(blob_hdr.uncompressed_size);
 177
 178         if (unlikely(reshdr_ret->uncompressed_size == 0)) {
 179                 ERROR("Data read on pipe is invalid (resource is of 0 size)");
 180                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 181         }
 182
 183         return 0;
 184
 185 read_error:
 186         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE)
 187                 ERROR("The pipe ended before all needed data was sent!");
 188         else
 189                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
 190         return ret;
 191 }
 192
 193 static int
 194 read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 195 {
 196         int ret;
 197         u8 hash[SHA1_HASH_SIZE];
 198         struct wim_reshdr reshdr;
 199         struct wim_header_disk pwm_hdr;
 200         struct wim_resource_descriptor rdesc;
 201         struct blob_descriptor *blob;
 202
 203         copy_guid(ctx->progress.extract.guid, ctx->wim->hdr.guid);
 204         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
 205         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
 206         ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 207         if (ret)
 208                 return ret;
 209
 210         while (ctx->num_blobs_remaining) {
 211
 212                 ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr);
 213
 214                 if (ret == PWM_FOUND_WIM_HDR) {
 215                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
 216                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
 217
 218                         if (part_number == ctx->progress.extract.part_number &&
 219                             total_parts == ctx->progress.extract.total_parts &&
 220                             guids_equal(pwm_hdr.guid, ctx->progress.extract.guid))
 221                                 continue;
 222
 223                         copy_guid(ctx->progress.extract.guid, pwm_hdr.guid);
 224                         ctx->progress.extract.part_number = part_number;
 225                         ctx->progress.extract.total_parts = total_parts;
 226                         ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 227                         if (ret)
 228                                 return ret;
 229
 230                         continue;
 231                 }
 232
 233                 if (ret)
 234                         return ret;
 235
 236                 if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)
 237                     && (blob = lookup_blob(ctx->wim->blob_table, hash))
 238                     && (blob->out_refcnt))
 239                 {
 240                         wim_reshdr_to_desc_and_blob(&reshdr, ctx->wim, &rdesc, blob);
 241                         ret = read_blob_with_sha1(blob, cbs);
 242                         blob_unset_is_located_in_wim_resource(blob);
 243                         if (ret)
 244                                 return ret;
 245                         ctx->num_blobs_remaining--;
 246                 } else {
 247                         wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc);
 248                         ret = skip_wim_resource(&rdesc);
 249                         if (ret)
 250                                 return ret;
 251                 }
 252         }
 253
 254         return 0;
 255 }
 256
 257 /* Creates a temporary file opened for writing.  The open file descriptor is
 258  * returned in @fd_ret and its name is returned in @name_ret (dynamically
 259  * allocated).  */
 260 static int
 261 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
 262 {
 263         tchar *name;
 264         int open_flags;
 265         int raw_fd;
 266
 267 retry:
 268         name = ttempnam(NULL, T("wimlib"));
 269         if (!name) {
 270                 ERROR_WITH_ERRNO("Failed to create temporary filename");
 271                 return WIMLIB_ERR_NOMEM;
 272         }
 273
 274         open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
 275 #ifdef __WIN32__
 276         open_flags |= _O_SHORT_LIVED;
 277 #endif
 278         raw_fd = topen(name, open_flags, 0600);
 279
 280         if (raw_fd < 0) {
 281                 if (errno == EEXIST) {
 282                         FREE(name);
 283                         goto retry;
 284                 }
 285                 ERROR_WITH_ERRNO("Failed to create temporary file "
 286                                  "\"%"TS"\"", name);
 287                 FREE(name);
 288                 return WIMLIB_ERR_OPEN;
 289         }
 290
 291         filedes_init(fd_ret, raw_fd);
 292         *name_ret = name;
 293         return 0;
 294 }
 295
 296 static int
 297 begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
 298 {
 299         struct apply_ctx *ctx = _ctx;
 300
 301         ctx->cur_blob = blob;
 302         ctx->cur_blob_offset = 0;
 303
 304         if (unlikely(blob->out_refcnt > MAX_OPEN_FILES))
 305                 return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
 306
 307         return call_begin_blob(blob, ctx->saved_cbs);
 308 }
 309
 310 static int
 311 extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
 312 {
 313         struct apply_ctx *ctx = _ctx;
 314         union wimlib_progress_info *progress = &ctx->progress;
 315         int ret;
 316
 317         ctx->cur_blob_offset += size;
 318
 319         if (likely(ctx->supported_features.hard_links)) {
 320                 progress->extract.completed_bytes +=
 321                         (u64)size * ctx->cur_blob->out_refcnt;
 322                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 323                         progress->extract.completed_streams += ctx->cur_blob->out_refcnt;
 324         } else {
 325                 const struct blob_extraction_target *targets =
 326                         blob_extraction_targets(ctx->cur_blob);
 327                 for (u32 i = 0; i < ctx->cur_blob->out_refcnt; i++) {
 328                         const struct wim_inode *inode = targets[i].inode;
 329                         const struct wim_dentry *dentry;
 330
 331                         inode_for_each_extraction_alias(dentry, inode) {
 332                                 progress->extract.completed_bytes += size;
 333                                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 334                                         progress->extract.completed_streams++;
 335                         }
 336                 }
 337         }
 338         if (progress->extract.completed_bytes >= ctx->next_progress) {
 339
 340                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
 341                 if (ret)
 342                         return ret;
 343
 344                 set_next_progress(progress->extract.completed_bytes,
 345                                   progress->extract.total_bytes,
 346                                   &ctx->next_progress);
 347         }
 348
 349         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 350                 /* Just extracting to temporary file for now.  */
 351                 ret = full_write(&ctx->tmpfile_fd, chunk, size);
 352                 if (ret) {
 353                         ERROR_WITH_ERRNO("Error writing data to "
 354                                          "temporary file \"%"TS"\"",
 355                                          ctx->tmpfile_name);
 356                 }
 357                 return ret;
 358         }
 359
 360         return call_consume_chunk(chunk, size, ctx->saved_cbs);
 361 }
 362
 363 /* Copy the blob's data from the temporary file to each of its targets.
 364  *
 365  * This is executed only in the very uncommon case that a blob is being
 366  * extracted to more than MAX_OPEN_FILES targets!  */
 367 static int
 368 extract_from_tmpfile(const tchar *tmpfile_name,
 369                      const struct blob_descriptor *orig_blob,
 370                      const struct read_blob_callbacks *cbs)
 371 {
 372         struct blob_descriptor tmpfile_blob;
 373         const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob);
 374         int ret;
 375
 376         memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor));
 377         tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK;
 378         tmpfile_blob.file_on_disk = (tchar *)tmpfile_name;
 379         tmpfile_blob.out_refcnt = 1;
 380
 381         for (u32 i = 0; i < orig_blob->out_refcnt; i++) {
 382                 tmpfile_blob.inline_blob_extraction_targets[0] = targets[i];
 383                 ret = read_blob_with_cbs(&tmpfile_blob, cbs);
 384                 if (ret)
 385                         return ret;
 386         }
 387         return 0;
 388 }
 389
 390 static int
 391 end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
 392 {
 393         struct apply_ctx *ctx = _ctx;
 394
 395         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 396                 filedes_close(&ctx->tmpfile_fd);
 397                 if (!status)
 398                         status = extract_from_tmpfile(ctx->tmpfile_name, blob,
 399                                                       ctx->saved_cbs);
 400                 filedes_invalidate(&ctx->tmpfile_fd);
 401                 tunlink(ctx->tmpfile_name);
 402                 FREE(ctx->tmpfile_name);
 403                 return status;
 404         }
 405
 406         return call_end_blob(blob, status, ctx->saved_cbs);
 407 }
 408
 409 /*
 410  * Read the list of blobs to extract and feed their data into the specified
 411  * callback functions.
 412  *
 413  * This handles checksumming each blob.
 414  *
 415  * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
 416  *
 417  * This also works if the WIM is being read from a pipe.
 418  *
 419  * This also will split up blobs that will need to be extracted to more than
 420  * MAX_OPEN_FILES locations, as measured by the 'out_refcnt' of each blob.
 421  * Therefore, the apply_operations implementation need not worry about running
 422  * out of file descriptors, unless it might open more than one file descriptor
 423  * per 'blob_extraction_target' (e.g. Win32 currently might because the
 424  * destination file system might not support hard links).
 425  */
 426 int
 427 extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 428 {
 429         struct read_blob_callbacks wrapper_cbs = {
 430                 .begin_blob     = begin_extract_blob_wrapper,
 431                 .consume_chunk  = extract_chunk_wrapper,
 432                 .end_blob       = end_extract_blob_wrapper,
 433                 .ctx            = ctx,
 434         };
 435         ctx->saved_cbs = cbs;
 436         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
 437                 return read_blobs_from_pipe(ctx, &wrapper_cbs);
 438         } else {
 439                 return read_blob_list(&ctx->blob_list,
 440                                       offsetof(struct blob_descriptor,
 441                                                extraction_list),
 442                                       &wrapper_cbs, VERIFY_BLOB_HASHES);
 443         }
 444 }
 445
 446 /* Extract a WIM dentry to standard output.
 447  *
 448  * This obviously doesn't make sense in all cases.  We return an error if the
 449  * dentry does not correspond to a regular file.  Otherwise we extract the
 450  * unnamed data stream only.  */
 451 static int
 452 extract_dentry_to_stdout(struct wim_dentry *dentry,
 453                          const struct blob_table *blob_table)
 454 {
 455         struct wim_inode *inode = dentry->d_inode;
 456         struct blob_descriptor *blob;
 457         struct filedes _stdout;
 458
 459         if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
 460                                    FILE_ATTRIBUTE_DIRECTORY |
 461                                    FILE_ATTRIBUTE_ENCRYPTED))
 462         {
 463                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 464                       "extracted to standard output", dentry_full_path(dentry));
 465                 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
 466         }
 467
 468         blob = inode_get_blob_for_unnamed_data_stream(inode, blob_table);
 469         if (!blob) {
 470                 const u8 *hash = inode_get_hash_of_unnamed_data_stream(inode);
 471                 if (!is_zero_hash(hash))
 472                         return blob_not_found_error(inode, hash);
 473                 return 0;
 474         }
 475
 476         filedes_init(&_stdout, STDOUT_FILENO);
 477         return extract_blob_to_fd(blob, &_stdout);
 478 }
 479
 480 static int
 481 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
 482                            const struct blob_table *blob_table)
 483 {
 484         for (size_t i = 0; i < num_dentries; i++) {
 485                 int ret = extract_dentry_to_stdout(dentries[i], blob_table);
 486                 if (ret)
 487                         return ret;
 488         }
 489         return 0;
 490 }
 491
 492 /**********************************************************************/
 493
 494 /*
 495  * Removes duplicate dentries from the array.
 496  *
 497  * Returns the new number of dentries, packed at the front of the array.
 498  */
 499 static size_t
 500 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
 501 {
 502         size_t i, j = 0;
 503         for (i = 0; i < num_trees; i++) {
 504                 if (!trees[i]->d_tmp_flag) {
 505                         /* Found distinct dentry.  */
 506                         trees[i]->d_tmp_flag = 1;
 507                         trees[j++] = trees[i];
 508                 }
 509         }
 510         for (i = 0; i < j; i++)
 511                 trees[i]->d_tmp_flag = 0;
 512         return j;
 513 }
 514
 515 /*
 516  * Remove dentries that are descendants of other dentries in the array.
 517  *
 518  * Returns the new number of dentries, packed at the front of the array.
 519  */
 520 static size_t
 521 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
 522 {
 523         size_t i, j = 0;
 524         for (i = 0; i < num_trees; i++)
 525                 trees[i]->d_tmp_flag = 1;
 526         for (i = 0; i < num_trees; i++) {
 527                 struct wim_dentry *d = trees[i];
 528                 while (!dentry_is_root(d)) {
 529                         d = d->d_parent;
 530                         if (d->d_tmp_flag)
 531                                 goto tree_contained;
 532                 }
 533                 trees[j++] = trees[i];
 534                 continue;
 535
 536         tree_contained:
 537                 trees[i]->d_tmp_flag = 0;
 538         }
 539
 540         for (i = 0; i < j; i++)
 541                 trees[i]->d_tmp_flag = 0;
 542         return j;
 543 }
 544
 545 static int
 546 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
 547 {
 548         struct list_head *dentry_list = _dentry_list;
 549         list_add_tail(&dentry->d_extraction_list_node, dentry_list);
 550         return 0;
 551 }
 552
 553 static void
 554 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
 555 {
 556         dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
 557 }
 558
 559 static int
 560 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
 561 {
 562         if (will_extract_dentry(dentry)) {
 563                 list_del(&dentry->d_extraction_list_node);
 564                 dentry_reset_extraction_list_node(dentry);
 565         }
 566         return 0;
 567 }
 568
 569 /*
 570  * Build the preliminary list of dentries to be extracted.
 571  *
 572  * The list maintains the invariant that if d1 and d2 are in the list and d1 is
 573  * an ancestor of d2, then d1 appears before d2 in the list.
 574  */
 575 static void
 576 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
 577                   size_t num_trees, bool add_ancestors)
 578 {
 579         INIT_LIST_HEAD(dentry_list);
 580
 581         /* Add the trees recursively.  */
 582         for (size_t i = 0; i < num_trees; i++)
 583                 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
 584
 585         /* If requested, add ancestors of the trees.  */
 586         if (add_ancestors) {
 587                 for (size_t i = 0; i < num_trees; i++) {
 588                         struct wim_dentry *dentry = trees[i];
 589                         struct wim_dentry *ancestor;
 590                         struct list_head *place_after;
 591
 592                         if (dentry_is_root(dentry))
 593                                 continue;
 594
 595                         place_after = dentry_list;
 596                         ancestor = dentry;
 597                         do {
 598                                 ancestor = ancestor->d_parent;
 599                                 if (will_extract_dentry(ancestor)) {
 600                                         place_after = &ancestor->d_extraction_list_node;
 601                                         break;
 602                                 }
 603                         } while (!dentry_is_root(ancestor));
 604
 605                         ancestor = dentry;
 606                         do {
 607                                 ancestor = ancestor->d_parent;
 608                                 if (will_extract_dentry(ancestor))
 609                                         break;
 610                                 list_add(&ancestor->d_extraction_list_node, place_after);
 611                         } while (!dentry_is_root(ancestor));
 612                 }
 613         }
 614 }
 615
 616 static void
 617 destroy_dentry_list(struct list_head *dentry_list)
 618 {
 619         struct wim_dentry *dentry, *tmp;
 620         struct wim_inode *inode;
 621
 622         list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
 623                 inode = dentry->d_inode;
 624                 dentry_reset_extraction_list_node(dentry);
 625                 inode->i_visited = 0;
 626                 inode->i_can_externally_back = 0;
 627                 if ((void *)dentry->d_extraction_name != (void *)dentry->d_name)
 628                         FREE(dentry->d_extraction_name);
 629                 dentry->d_extraction_name = NULL;
 630                 dentry->d_extraction_name_nchars = 0;
 631         }
 632 }
 633
 634 static void
 635 destroy_blob_list(struct list_head *blob_list)
 636 {
 637         struct blob_descriptor *blob;
 638
 639         list_for_each_entry(blob, blob_list, extraction_list)
 640                 if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets))
 641                         FREE(blob->blob_extraction_targets);
 642 }
 643
 644 #ifdef __WIN32__
 645 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
 646 #else
 647 static const utf16lechar replacement_char = cpu_to_le16('?');
 648 #endif
 649
 650 static bool
 651 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
 652 {
 653         size_t i;
 654
 655         if (num_chars == 0)
 656                 return true;
 657         for (i = 0; i < num_chars; i++) {
 658                 switch (le16_to_cpu(name[i])) {
 659         #ifdef __WIN32__
 660                 case '\x01'...'\x1F':
 661                 case '\\':
 662                 case ':':
 663                 case '*':
 664                 case '?':
 665                 case '"':
 666                 case '<':
 667                 case '>':
 668                 case '|':
 669         #endif
 670                 case '/':
 671                 case '\0':
 672                         if (fix)
 673                                 name[i] = replacement_char;
 674                         else
 675                                 return false;
 676                 }
 677         }
 678
 679         return true;
 680 }
 681
 682 static int
 683 dentry_calculate_extraction_name(struct wim_dentry *dentry,
 684                                  struct apply_ctx *ctx)
 685 {
 686         int ret;
 687
 688         if (dentry_is_root(dentry))
 689                 return 0;
 690
 691 #ifdef WITH_NTFS_3G
 692         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 693                 dentry->d_extraction_name = dentry->d_name;
 694                 dentry->d_extraction_name_nchars = dentry->d_name_nbytes /
 695                                                    sizeof(utf16lechar);
 696                 return 0;
 697         }
 698 #endif
 699
 700         if (!ctx->supported_features.case_sensitive_filenames) {
 701                 struct wim_dentry *other;
 702                 list_for_each_entry(other, &dentry->d_ci_conflict_list,
 703                                     d_ci_conflict_list)
 704                 {
 705                         if (will_extract_dentry(other)) {
 706                                 if (ctx->extract_flags &
 707                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
 708                                         WARNING("\"%"TS"\" has the same "
 709                                                 "case-insensitive name as "
 710                                                 "\"%"TS"\"; extracting "
 711                                                 "dummy name instead",
 712                                                 dentry_full_path(dentry),
 713                                                 dentry_full_path(other));
 714                                         goto out_replace;
 715                                 } else {
 716                                         WARNING("Not extracting \"%"TS"\": "
 717                                                 "has same case-insensitive "
 718                                                 "name as \"%"TS"\"",
 719                                                 dentry_full_path(dentry),
 720                                                 dentry_full_path(other));
 721                                         goto skip_dentry;
 722                                 }
 723                         }
 724                 }
 725         }
 726
 727         if (file_name_valid(dentry->d_name, dentry->d_name_nbytes / 2, false)) {
 728                 size_t nbytes = 0;
 729                 ret = utf16le_get_tstr(dentry->d_name,
 730                                        dentry->d_name_nbytes,
 731                                        (const tchar **)&dentry->d_extraction_name,
 732                                        &nbytes);
 733                 dentry->d_extraction_name_nchars = nbytes / sizeof(tchar);
 734                 return ret;
 735         } else {
 736                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
 737                 {
 738                         WARNING("\"%"TS"\" has an invalid filename "
 739                                 "that is not supported on this platform; "
 740                                 "extracting dummy name instead",
 741                                 dentry_full_path(dentry));
 742                         goto out_replace;
 743                 } else {
 744                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
 745                                 "that is not supported on this platform",
 746                                 dentry_full_path(dentry));
 747                         goto skip_dentry;
 748                 }
 749         }
 750
 751 out_replace:
 752         {
 753                 utf16lechar utf16_name_copy[dentry->d_name_nbytes / 2];
 754
 755                 memcpy(utf16_name_copy, dentry->d_name, dentry->d_name_nbytes);
 756                 file_name_valid(utf16_name_copy, dentry->d_name_nbytes / 2, true);
 757
 758                 const tchar *tchar_name;
 759                 size_t tchar_nchars;
 760
 761                 ret = utf16le_get_tstr(utf16_name_copy,
 762                                        dentry->d_name_nbytes,
 763                                        &tchar_name, &tchar_nchars);
 764                 if (ret)
 765                         return ret;
 766
 767                 tchar_nchars /= sizeof(tchar);
 768
 769                 size_t fixed_name_num_chars = tchar_nchars;
 770                 tchar fixed_name[tchar_nchars + 50];
 771
 772                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
 773                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
 774                                                  T(" (invalid filename #%lu)"),
 775                                                  ++ctx->invalid_sequence);
 776
 777                 utf16le_put_tstr(tchar_name);
 778
 779                 dentry->d_extraction_name = TSTRDUP(fixed_name);
 780                 if (!dentry->d_extraction_name)
 781                         return WIMLIB_ERR_NOMEM;
 782                 dentry->d_extraction_name_nchars = fixed_name_num_chars;
 783         }
 784         return 0;
 785
 786 skip_dentry:
 787         for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
 788         return 0;
 789 }
 790
 791 /*
 792  * Calculate the actual filename component at which each WIM dentry will be
 793  * extracted, with special handling for dentries that are unsupported by the
 794  * extraction backend or have invalid names.
 795  *
 796  * ctx->supported_features must be filled in.
 797  *
 798  * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
 799  */
 800 static int
 801 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
 802                                        struct apply_ctx *ctx)
 803 {
 804         struct list_head *prev, *cur;
 805
 806         /* Can't use list_for_each_entry() because a call to
 807          * dentry_calculate_extraction_name() may delete the current dentry and
 808          * its children from the list.  */
 809
 810         prev = dentry_list;
 811         for (;;) {
 812                 struct wim_dentry *dentry;
 813                 int ret;
 814
 815                 cur = prev->next;
 816                 if (cur == dentry_list)
 817                         break;
 818
 819                 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
 820
 821                 ret = dentry_calculate_extraction_name(dentry, ctx);
 822                 if (ret)
 823                         return ret;
 824
 825                 if (prev->next == cur)
 826                         prev = cur;
 827                 else
 828                         ; /* Current dentry and its children (which follow in
 829                              the list) were deleted.  prev stays the same.  */
 830         }
 831         return 0;
 832 }
 833
 834 static int
 835 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
 836                        struct blob_table *blob_table)
 837 {
 838         struct wim_inode *inode = dentry->d_inode;
 839         struct blob_descriptor *blob;
 840         int ret;
 841         bool force = false;
 842
 843         /* Special case:  when extracting from a pipe, the WIM blob table is
 844          * initially empty, so "resolving" an inode's streams is initially not
 845          * possible.  However, we still need to keep track of which blobs,
 846          * identified by SHA-1 message digests, need to be extracted, so we
 847          * "resolve" the inode's streams anyway by allocating a 'struct
 848          * blob_descriptor' for each one.  */
 849         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
 850                 force = true;
 851         ret = inode_resolve_streams(inode, blob_table, force);
 852         if (ret)
 853                 return ret;
 854         for (unsigned i = 0; i < inode->i_num_streams; i++) {
 855                 blob = stream_blob_resolved(&inode->i_streams[i]);
 856                 if (blob)
 857                         blob->out_refcnt = 0;
 858         }
 859         return 0;
 860 }
 861
 862 /*
 863  * For each dentry to be extracted, resolve all streams in the corresponding
 864  * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
 865  *
 866  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
 867  */
 868 static int
 869 dentry_list_resolve_streams(struct list_head *dentry_list,
 870                             struct apply_ctx *ctx)
 871 {
 872         struct wim_dentry *dentry;
 873         int ret;
 874
 875         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
 876                 ret = dentry_resolve_streams(dentry,
 877                                              ctx->extract_flags,
 878                                              ctx->wim->blob_table);
 879                 if (ret)
 880                         return ret;
 881         }
 882         return 0;
 883 }
 884
 885 static int
 886 ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
 887            struct apply_ctx *ctx)
 888 {
 889         struct wim_inode *inode = dentry->d_inode;
 890         struct blob_descriptor *blob = stream_blob_resolved(strm);
 891         struct blob_extraction_target *targets;
 892
 893         if (!blob)
 894                 return 0;
 895
 896         /* Tally the size only for each actual extraction of the stream (not
 897          * additional hard links to the inode).  */
 898         if (inode->i_visited && ctx->supported_features.hard_links)
 899                 return 0;
 900
 901         ctx->progress.extract.total_bytes += blob->size;
 902         ctx->progress.extract.total_streams++;
 903
 904         if (inode->i_visited)
 905                 return 0;
 906
 907         /* Add each blob to 'ctx->blob_list' only one time, regardless of how
 908          * many extraction targets it will have.  */
 909         if (blob->out_refcnt == 0) {
 910                 list_add_tail(&blob->extraction_list, &ctx->blob_list);
 911                 ctx->num_blobs_remaining++;
 912         }
 913
 914         /* Set this stream as an extraction target of 'blob'.  */
 915
 916         if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 917                 targets = blob->inline_blob_extraction_targets;
 918         } else {
 919                 struct blob_extraction_target *prev_targets;
 920                 size_t alloc_blob_extraction_targets;
 921
 922                 if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 923                         prev_targets = NULL;
 924                         alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets);
 925                 } else {
 926                         prev_targets = blob->blob_extraction_targets;
 927                         alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
 928                 }
 929
 930                 if (blob->out_refcnt == alloc_blob_extraction_targets) {
 931                         alloc_blob_extraction_targets *= 2;
 932                         targets = REALLOC(prev_targets,
 933                                           alloc_blob_extraction_targets *
 934                                           sizeof(targets[0]));
 935                         if (!targets)
 936                                 return WIMLIB_ERR_NOMEM;
 937                         if (!prev_targets) {
 938                                 memcpy(targets,
 939                                        blob->inline_blob_extraction_targets,
 940                                        sizeof(blob->inline_blob_extraction_targets));
 941                         }
 942                         blob->blob_extraction_targets = targets;
 943                         blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
 944                 }
 945                 targets = blob->blob_extraction_targets;
 946         }
 947         targets[blob->out_refcnt].inode = inode;
 948         targets[blob->out_refcnt].stream = strm;
 949         blob->out_refcnt++;
 950         return 0;
 951 }
 952
 953 static int
 954 ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode,
 955                      struct wim_inode_stream *strm, struct apply_ctx *ctx)
 956 {
 957         bool need_stream = false;
 958         switch (strm->stream_type) {
 959         case STREAM_TYPE_DATA:
 960                 if (stream_is_named(strm)) {
 961                         /* Named data stream  */
 962                         if (ctx->supported_features.named_data_streams)
 963                                 need_stream = true;
 964                 } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
 965                                                     FILE_ATTRIBUTE_ENCRYPTED))
 966                            && !(inode_is_symlink(inode)
 967                                 && !ctx->supported_features.reparse_points
 968                                 && ctx->supported_features.symlink_reparse_points))
 969                 {
 970                         /*
 971                          * Unnamed data stream.  Skip if any of the following is true:
 972                          *
 973                          * - file is a directory
 974                          * - file is encrypted
 975                          * - backend needs to create the file as UNIX symlink
 976                          * - backend will extract the stream as externally backed
 977                          */
 978                         if (ctx->apply_ops->will_externally_back) {
 979                                 int ret = (*ctx->apply_ops->will_externally_back)(dentry, ctx);
 980                                 if (ret > 0) /* Error?  */
 981                                         return ret;
 982                                 if (ret < 0) /* Won't externally back?  */
 983                                         need_stream = true;
 984                         } else {
 985                                 need_stream = true;
 986                         }
 987                 }
 988                 break;
 989         case STREAM_TYPE_REPARSE_POINT:
 990                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT);
 991                 if (ctx->supported_features.reparse_points ||
 992                     (inode_is_symlink(inode) &&
 993                      ctx->supported_features.symlink_reparse_points))
 994                         need_stream = true;
 995                 break;
 996         case STREAM_TYPE_EFSRPC_RAW_DATA:
 997                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED);
 998                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
 999                         if (ctx->supported_features.encrypted_directories)
1000                                 need_stream = true;
1001                 } else {
1002                         if (ctx->supported_features.encrypted_files)
1003                                 need_stream = true;
1004                 }
1005                 break;
1006         }
1007         if (need_stream)
1008                 return ref_stream(strm, dentry, ctx);
1009         return 0;
1010 }
1011
1012 static int
1013 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
1014 {
1015         struct wim_inode *inode = dentry->d_inode;
1016         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1017                 int ret = ref_stream_if_needed(dentry, inode,
1018                                                &inode->i_streams[i], ctx);
1019                 if (ret)
1020                         return ret;
1021         }
1022         inode->i_visited = 1;
1023         return 0;
1024 }
1025
1026 /*
1027  * Given a list of dentries to be extracted, build the list of blobs that need
1028  * to be extracted, and for each blob determine the streams to which that blob
1029  * will be extracted.
1030  *
1031  * This also initializes the extract progress info with byte and blob
1032  * information.
1033  *
1034  * ctx->supported_features must be filled in.
1035  */
1036 static int
1037 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
1038 {
1039         struct wim_dentry *dentry;
1040         int ret;
1041
1042         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1043                 ret = dentry_ref_streams(dentry, ctx);
1044                 if (ret)
1045                         return ret;
1046         }
1047         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1048                 dentry->d_inode->i_visited = 0;
1049         return 0;
1050 }
1051
1052 static void
1053 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
1054 {
1055         struct wim_dentry *dentry;
1056
1057         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1058                 dentry->d_inode->i_first_extraction_alias = NULL;
1059
1060         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1061                 dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias;
1062                 dentry->d_inode->i_first_extraction_alias = dentry;
1063         }
1064 }
1065
1066 static void
1067 inode_tally_features(const struct wim_inode *inode,
1068                      struct wim_features *features)
1069 {
1070         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1071                 features->archive_files++;
1072         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1073                 features->hidden_files++;
1074         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1075                 features->system_files++;
1076         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1077                 features->compressed_files++;
1078         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1079                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1080                         features->encrypted_directories++;
1081                 else
1082                         features->encrypted_files++;
1083         }
1084         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1085                 features->not_context_indexed_files++;
1086         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1087                 features->sparse_files++;
1088         if (inode_has_named_data_stream(inode))
1089                 features->named_data_streams++;
1090         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1091                 features->reparse_points++;
1092                 if (inode_is_symlink(inode))
1093                         features->symlink_reparse_points++;
1094                 else
1095                         features->other_reparse_points++;
1096         }
1097         if (inode_has_security_descriptor(inode))
1098                 features->security_descriptors++;
1099         if (inode_has_unix_data(inode))
1100                 features->unix_data++;
1101 }
1102
1103 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1104 static void
1105 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
1106 {
1107         struct wim_inode *inode = dentry->d_inode;
1108
1109         if (dentry_has_short_name(dentry))
1110                 features->short_names++;
1111
1112         if (inode->i_visited) {
1113                 features->hard_links++;
1114         } else {
1115                 inode_tally_features(inode, features);
1116                 inode->i_visited = 1;
1117         }
1118 }
1119
1120 /* Tally the features necessary to extract the specified dentries.  */
1121 static void
1122 dentry_list_get_features(struct list_head *dentry_list,
1123                          struct wim_features *features)
1124 {
1125         struct wim_dentry *dentry;
1126
1127         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1128                 dentry_tally_features(dentry, features);
1129
1130         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1131                 dentry->d_inode->i_visited = 0;
1132 }
1133
1134 static int
1135 do_feature_check(const struct wim_features *required_features,
1136                  const struct wim_features *supported_features,
1137                  int extract_flags)
1138 {
1139         /* Encrypted files.  */
1140         if (required_features->encrypted_files &&
1141             !supported_features->encrypted_files)
1142                 WARNING("Ignoring EFS-encrypted data of %lu files",
1143                         required_features->encrypted_files);
1144
1145         /* Named data streams.  */
1146         if (required_features->named_data_streams &&
1147             !supported_features->named_data_streams)
1148                 WARNING("Ignoring named data streams of %lu files",
1149                         required_features->named_data_streams);
1150
1151         /* File attributes.  */
1152         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1153                 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1154                  * We're an archive program, so theoretically we can do what we
1155                  * want with it.  */
1156
1157                 if (required_features->hidden_files &&
1158                     !supported_features->hidden_files)
1159                         WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1160                                 required_features->hidden_files);
1161
1162                 if (required_features->system_files &&
1163                     !supported_features->system_files)
1164                         WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1165                                 required_features->system_files);
1166
1167                 if (required_features->compressed_files &&
1168                     !supported_features->compressed_files)
1169                         WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1170                                 required_features->compressed_files);
1171
1172                 if (required_features->not_context_indexed_files &&
1173                     !supported_features->not_context_indexed_files)
1174                         WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1175                                 required_features->not_context_indexed_files);
1176
1177                 if (required_features->sparse_files &&
1178                     !supported_features->sparse_files)
1179                         WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1180                                 required_features->sparse_files);
1181
1182                 if (required_features->encrypted_directories &&
1183                     !supported_features->encrypted_directories)
1184                         WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1185                                 required_features->encrypted_directories);
1186         }
1187
1188         /* Hard links.  */
1189         if (required_features->hard_links && !supported_features->hard_links)
1190                 WARNING("Extracting %lu hard links as independent files",
1191                         required_features->hard_links);
1192
1193         /* Symbolic links and reparse points.  */
1194         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1195             required_features->symlink_reparse_points &&
1196             !supported_features->symlink_reparse_points &&
1197             !supported_features->reparse_points)
1198         {
1199                 ERROR("Extraction backend does not support symbolic links!");
1200                 return WIMLIB_ERR_UNSUPPORTED;
1201         }
1202         if (required_features->reparse_points &&
1203             !supported_features->reparse_points)
1204         {
1205                 if (supported_features->symlink_reparse_points) {
1206                         if (required_features->other_reparse_points) {
1207                                 WARNING("Ignoring reparse data of %lu non-symlink/junction files",
1208                                         required_features->other_reparse_points);
1209                         }
1210                 } else {
1211                         WARNING("Ignoring reparse data of %lu files",
1212                                 required_features->reparse_points);
1213                 }
1214         }
1215
1216         /* Security descriptors.  */
1217         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1218                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1219              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1220             required_features->security_descriptors &&
1221             !supported_features->security_descriptors)
1222         {
1223                 ERROR("Extraction backend does not support security descriptors!");
1224                 return WIMLIB_ERR_UNSUPPORTED;
1225         }
1226         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1227             required_features->security_descriptors &&
1228             !supported_features->security_descriptors)
1229                 WARNING("Ignoring Windows NT security descriptors of %lu files",
1230                         required_features->security_descriptors);
1231
1232         /* UNIX data.  */
1233         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1234             required_features->unix_data && !supported_features->unix_data)
1235         {
1236                 ERROR("Extraction backend does not support UNIX data!");
1237                 return WIMLIB_ERR_UNSUPPORTED;
1238         }
1239
1240         if (required_features->unix_data &&
1241             !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1242         {
1243                 WARNING("Ignoring UNIX metadata of %lu files",
1244                         required_features->unix_data);
1245         }
1246
1247         /* DOS Names.  */
1248         if (required_features->short_names &&
1249             !supported_features->short_names)
1250         {
1251                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1252                         ERROR("Extraction backend does not support DOS names!");
1253                         return WIMLIB_ERR_UNSUPPORTED;
1254                 }
1255                 WARNING("Ignoring DOS names of %lu files",
1256                         required_features->short_names);
1257         }
1258
1259         /* Timestamps.  */
1260         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1261             !supported_features->timestamps)
1262         {
1263                 ERROR("Extraction backend does not support timestamps!");
1264                 return WIMLIB_ERR_UNSUPPORTED;
1265         }
1266
1267         return 0;
1268 }
1269
1270 static const struct apply_operations *
1271 select_apply_operations(int extract_flags)
1272 {
1273 #ifdef WITH_NTFS_3G
1274         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1275                 return &ntfs_3g_apply_ops;
1276 #endif
1277 #ifdef __WIN32__
1278         return &win32_apply_ops;
1279 #else
1280         return &unix_apply_ops;
1281 #endif
1282 }
1283
1284 static int
1285 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1286               const tchar *target, int extract_flags)
1287 {
1288         const struct apply_operations *ops;
1289         struct apply_ctx *ctx;
1290         int ret;
1291         LIST_HEAD(dentry_list);
1292
1293         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1294                 ret = extract_dentries_to_stdout(trees, num_trees,
1295                                                  wim->blob_table);
1296                 goto out;
1297         }
1298
1299         num_trees = remove_duplicate_trees(trees, num_trees);
1300         num_trees = remove_contained_trees(trees, num_trees);
1301
1302         ops = select_apply_operations(extract_flags);
1303
1304         if (num_trees > 1 && ops->single_tree_only) {
1305                 ERROR("Extracting multiple directory trees "
1306                       "at once is not supported in %s extraction mode!",
1307                       ops->name);
1308                 ret = WIMLIB_ERR_UNSUPPORTED;
1309                 goto out;
1310         }
1311
1312         ctx = CALLOC(1, ops->context_size);
1313         if (!ctx) {
1314                 ret = WIMLIB_ERR_NOMEM;
1315                 goto out;
1316         }
1317
1318         ctx->wim = wim;
1319         ctx->target = target;
1320         ctx->target_nchars = tstrlen(target);
1321         ctx->extract_flags = extract_flags;
1322         if (ctx->wim->progfunc) {
1323                 ctx->progfunc = ctx->wim->progfunc;
1324                 ctx->progctx = ctx->wim->progctx;
1325                 ctx->progress.extract.image = wim->current_image;
1326                 ctx->progress.extract.extract_flags = (extract_flags &
1327                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
1328                 ctx->progress.extract.wimfile_name = wim->filename;
1329                 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1330                                                                          wim->current_image);
1331                 ctx->progress.extract.target = target;
1332         }
1333         INIT_LIST_HEAD(&ctx->blob_list);
1334         filedes_invalidate(&ctx->tmpfile_fd);
1335         ctx->apply_ops = ops;
1336
1337         ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1338         if (ret)
1339                 goto out_cleanup;
1340
1341         build_dentry_list(&dentry_list, trees, num_trees,
1342                           !(extract_flags &
1343                             WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1344
1345         dentry_list_get_features(&dentry_list, &ctx->required_features);
1346
1347         ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1348                                ctx->extract_flags);
1349         if (ret)
1350                 goto out_cleanup;
1351
1352         ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1353         if (ret)
1354                 goto out_cleanup;
1355
1356         if (unlikely(list_empty(&dentry_list))) {
1357                 WARNING("There is nothing to extract!");
1358                 goto out_cleanup;
1359         }
1360
1361         ret = dentry_list_resolve_streams(&dentry_list, ctx);
1362         if (ret)
1363                 goto out_cleanup;
1364
1365         dentry_list_build_inode_alias_lists(&dentry_list);
1366
1367         ret = dentry_list_ref_streams(&dentry_list, ctx);
1368         if (ret)
1369                 goto out_cleanup;
1370
1371         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1372                 /* When extracting from a pipe, the number of bytes of data to
1373                  * extract can't be determined in the normal way (examining the
1374                  * blob table), since at this point all we have is a set of
1375                  * SHA-1 message digests of blobs that need to be extracted.
1376                  * However, we can get a reasonably accurate estimate by taking
1377                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1378                  * data.  This does assume that a full image is being extracted,
1379                  * but currently there is no API for doing otherwise.  (Also,
1380                  * subtract <HARDLINKBYTES> from this if hard links are
1381                  * supported by the extraction mode.)  */
1382                 ctx->progress.extract.total_bytes =
1383                         wim_info_get_image_total_bytes(wim->wim_info,
1384                                                        wim->current_image);
1385                 if (ctx->supported_features.hard_links) {
1386                         ctx->progress.extract.total_bytes -=
1387                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
1388                                                                    wim->current_image);
1389                 }
1390         }
1391
1392         ret = extract_progress(ctx,
1393                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1394                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
1395                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
1396         if (ret)
1397                 goto out_cleanup;
1398
1399         ret = (*ops->extract)(&dentry_list, ctx);
1400         if (ret)
1401                 goto out_cleanup;
1402
1403         if (ctx->progress.extract.completed_bytes <
1404             ctx->progress.extract.total_bytes)
1405         {
1406                 ctx->progress.extract.completed_bytes =
1407                         ctx->progress.extract.total_bytes;
1408                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
1409                 if (ret)
1410                         goto out_cleanup;
1411         }
1412
1413         ret = extract_progress(ctx,
1414                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1415                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
1416                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
1417 out_cleanup:
1418         destroy_blob_list(&ctx->blob_list);
1419         destroy_dentry_list(&dentry_list);
1420         FREE(ctx);
1421 out:
1422         return ret;
1423 }
1424
1425 static int
1426 mkdir_if_needed(const tchar *target)
1427 {
1428         if (!tmkdir(target, 0755))
1429                 return 0;
1430
1431         if (errno == EEXIST)
1432                 return 0;
1433
1434 #ifdef __WIN32__
1435         /* _wmkdir() fails with EACCES if called on a drive root directory.  */
1436         if (errno == EACCES)
1437                 return 0;
1438 #endif
1439
1440         ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
1441         return WIMLIB_ERR_MKDIR;
1442 }
1443
1444 /* Make sure the extraction flags make sense, and update them if needed.  */
1445 static int
1446 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1447 {
1448         int extract_flags = *extract_flags_p;
1449
1450         /* Check for invalid flag combinations  */
1451
1452         if ((extract_flags &
1453              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1454               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1455                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1456                 return WIMLIB_ERR_INVALID_PARAM;
1457
1458         if ((extract_flags &
1459              (WIMLIB_EXTRACT_FLAG_RPFIX |
1460               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1461                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
1462                 return WIMLIB_ERR_INVALID_PARAM;
1463
1464 #ifndef WITH_NTFS_3G
1465         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1466                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1467                       "        it cannot apply a WIM image directly to an NTFS volume.");
1468                 return WIMLIB_ERR_UNSUPPORTED;
1469         }
1470 #endif
1471
1472         if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1473 #ifdef __WIN32__
1474                 if (!wim->filename)
1475                         return WIMLIB_ERR_NO_FILENAME;
1476 #else
1477                 ERROR("WIMBoot extraction is only supported on Windows!");
1478                 return WIMLIB_ERR_UNSUPPORTED;
1479 #endif
1480         }
1481
1482
1483         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1484                               WIMLIB_EXTRACT_FLAG_NORPFIX |
1485                               WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1486                                         WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1487         {
1488                 /* For full-image extraction, do reparse point fixups by default
1489                  * if the WIM header says they are enabled.  */
1490                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1491                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1492         }
1493
1494         *extract_flags_p = extract_flags;
1495         return 0;
1496 }
1497
1498 struct append_dentry_ctx {
1499         struct wim_dentry **dentries;
1500         size_t num_dentries;
1501         size_t num_alloc_dentries;
1502 };
1503
1504 static int
1505 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1506 {
1507         struct append_dentry_ctx *ctx = _ctx;
1508
1509         if (ctx->num_dentries == ctx->num_alloc_dentries) {
1510                 struct wim_dentry **new_dentries;
1511                 size_t new_length;
1512
1513                 new_length = max(ctx->num_alloc_dentries + 8,
1514                                  ctx->num_alloc_dentries * 3 / 2);
1515                 new_dentries = REALLOC(ctx->dentries,
1516                                        new_length * sizeof(ctx->dentries[0]));
1517                 if (new_dentries == NULL)
1518                         return WIMLIB_ERR_NOMEM;
1519                 ctx->dentries = new_dentries;
1520                 ctx->num_alloc_dentries = new_length;
1521         }
1522         ctx->dentries[ctx->num_dentries++] = dentry;
1523         return 0;
1524 }
1525
1526 /* Append dentries matched by a path which can contain wildcard characters.  */
1527 static int
1528 append_matched_dentries(WIMStruct *wim, const tchar *orig_pattern,
1529                         int extract_flags, struct append_dentry_ctx *ctx)
1530 {
1531         const size_t count_before = ctx->num_dentries;
1532         tchar *pattern;
1533         int ret;
1534
1535         pattern = canonicalize_wim_path(orig_pattern);
1536         if (!pattern)
1537                 return WIMLIB_ERR_NOMEM;
1538         ret = expand_path_pattern(wim_get_current_root_dentry(wim), pattern,
1539                                   append_dentry_cb, ctx);
1540         FREE(pattern);
1541         if (ret || ctx->num_dentries > count_before)
1542                 return ret;
1543         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB) {
1544                 ERROR("No matches for path pattern \"%"TS"\"", orig_pattern);
1545                 return WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1546         }
1547         WARNING("No matches for path pattern \"%"TS"\"", orig_pattern);
1548         return 0;
1549 }
1550
1551 static int
1552 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1553                         const tchar * const *paths, size_t num_paths,
1554                         int extract_flags)
1555 {
1556         int ret;
1557         struct wim_dentry **trees;
1558         size_t num_trees;
1559
1560         if (wim == NULL || target == NULL || target[0] == T('\0') ||
1561             (num_paths != 0 && paths == NULL))
1562                 return WIMLIB_ERR_INVALID_PARAM;
1563
1564         ret = check_extract_flags(wim, &extract_flags);
1565         if (ret)
1566                 return ret;
1567
1568         ret = select_wim_image(wim, image);
1569         if (ret)
1570                 return ret;
1571
1572         ret = wim_checksum_unhashed_blobs(wim);
1573         if (ret)
1574                 return ret;
1575
1576         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1577                               WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1578             (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1579         {
1580                 ret = mkdir_if_needed(target);
1581                 if (ret)
1582                         return ret;
1583         }
1584
1585         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1586
1587                 struct append_dentry_ctx append_dentry_ctx = {
1588                         .dentries = NULL,
1589                         .num_dentries = 0,
1590                         .num_alloc_dentries = 0,
1591                 };
1592
1593                 for (size_t i = 0; i < num_paths; i++) {
1594                         ret = append_matched_dentries(wim, paths[i],
1595                                                       extract_flags,
1596                                                       &append_dentry_ctx);
1597                         if (ret) {
1598                                 trees = append_dentry_ctx.dentries;
1599                                 goto out_free_trees;
1600                         }
1601                 }
1602                 trees = append_dentry_ctx.dentries;
1603                 num_trees = append_dentry_ctx.num_dentries;
1604         } else {
1605                 trees = MALLOC(num_paths * sizeof(trees[0]));
1606                 if (trees == NULL)
1607                         return WIMLIB_ERR_NOMEM;
1608
1609                 for (size_t i = 0; i < num_paths; i++) {
1610
1611                         tchar *path = canonicalize_wim_path(paths[i]);
1612                         if (path == NULL) {
1613                                 ret = WIMLIB_ERR_NOMEM;
1614                                 goto out_free_trees;
1615                         }
1616
1617                         trees[i] = get_dentry(wim, path,
1618                                               WIMLIB_CASE_PLATFORM_DEFAULT);
1619                         FREE(path);
1620                         if (trees[i] == NULL) {
1621                                   ERROR("Path \"%"TS"\" does not exist "
1622                                         "in WIM image %d",
1623                                         paths[i], wim->current_image);
1624                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1625                                   goto out_free_trees;
1626                         }
1627                 }
1628                 num_trees = num_paths;
1629         }
1630
1631         if (num_trees == 0) {
1632                 ret = 0;
1633                 goto out_free_trees;
1634         }
1635
1636         ret = extract_trees(wim, trees, num_trees, target, extract_flags);
1637 out_free_trees:
1638         FREE(trees);
1639         return ret;
1640 }
1641
1642 static int
1643 extract_single_image(WIMStruct *wim, int image,
1644                      const tchar *target, int extract_flags)
1645 {
1646         const tchar *path = WIMLIB_WIM_ROOT_PATH;
1647         extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1648         return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
1649 }
1650
1651 static const tchar * const filename_forbidden_chars =
1652 T(
1653 #ifdef __WIN32__
1654 "<>:\"/\\|?*"
1655 #else
1656 "/"
1657 #endif
1658 );
1659
1660 /* This function checks if it is okay to use a WIM image's name as a directory
1661  * name.  */
1662 static bool
1663 image_name_ok_as_dir(const tchar *image_name)
1664 {
1665         return image_name && *image_name &&
1666                 !tstrpbrk(image_name, filename_forbidden_chars) &&
1667                 tstrcmp(image_name, T(".")) &&
1668                 tstrcmp(image_name, T(".."));
1669 }
1670
1671 /* Extracts all images from the WIM to the directory @target, with the images
1672  * placed in subdirectories named by their image names. */
1673 static int
1674 extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
1675 {
1676         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
1677         size_t output_path_len = tstrlen(target);
1678         tchar buf[output_path_len + 1 + image_name_max_len + 1];
1679         int ret;
1680         int image;
1681         const tchar *image_name;
1682
1683         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1684                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1685                 return WIMLIB_ERR_INVALID_PARAM;
1686         }
1687
1688         ret = mkdir_if_needed(target);
1689         if (ret)
1690                 return ret;
1691         tmemcpy(buf, target, output_path_len);
1692         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1693         for (image = 1; image <= wim->hdr.image_count; image++) {
1694                 image_name = wimlib_get_image_name(wim, image);
1695                 if (image_name_ok_as_dir(image_name)) {
1696                         tstrcpy(buf + output_path_len + 1, image_name);
1697                 } else {
1698                         /* Image name is empty or contains forbidden characters.
1699                          * Use image number instead. */
1700                         tsprintf(buf + output_path_len + 1, T("%d"), image);
1701                 }
1702                 ret = extract_single_image(wim, image, buf, extract_flags);
1703                 if (ret)
1704                         return ret;
1705         }
1706         return 0;
1707 }
1708
1709 static int
1710 do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1711                         int extract_flags)
1712 {
1713         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1714                              WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1715                              WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1716                 return WIMLIB_ERR_INVALID_PARAM;
1717
1718         if (image == WIMLIB_ALL_IMAGES)
1719                 return extract_all_images(wim, target, extract_flags);
1720         else
1721                 return extract_single_image(wim, image, target, extract_flags);
1722 }
1723
1724
1725 /****************************************************************************
1726  *                          Extraction API                                  *
1727  ****************************************************************************/
1728
1729 WIMLIBAPI int
1730 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1731                      const tchar * const *paths, size_t num_paths,
1732                      int extract_flags)
1733 {
1734         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1735                 return WIMLIB_ERR_INVALID_PARAM;
1736
1737         return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1738                                        extract_flags);
1739 }
1740
1741 WIMLIBAPI int
1742 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1743                         const tchar *path_list_file, int extract_flags)
1744 {
1745         int ret;
1746         tchar **paths;
1747         size_t num_paths;
1748         void *mem;
1749
1750         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1751         if (ret) {
1752                 ERROR("Failed to read path list file \"%"TS"\"",
1753                       path_list_file);
1754                 return ret;
1755         }
1756
1757         ret = wimlib_extract_paths(wim, image, target,
1758                                    (const tchar * const *)paths, num_paths,
1759                                    extract_flags);
1760         FREE(paths);
1761         FREE(mem);
1762         return ret;
1763 }
1764
1765 WIMLIBAPI int
1766 wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
1767                                              const tchar *image_num_or_name,
1768                                              const tchar *target,
1769                                              int extract_flags,
1770                                              wimlib_progress_func_t progfunc,
1771                                              void *progctx)
1772 {
1773         int ret;
1774         WIMStruct *pwm;
1775         struct filedes *in_fd;
1776         int image;
1777         unsigned i;
1778
1779         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1780                 return WIMLIB_ERR_INVALID_PARAM;
1781
1782         /* Read the WIM header from the pipe and get a WIMStruct to represent
1783          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
1784          * wimlib_open_wim(), getting a WIMStruct in this way will result in an
1785          * empty blob table, no XML data read, and no filename set.  */
1786         ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
1787                                     progfunc, progctx);
1788         if (ret)
1789                 return ret;
1790
1791         /* Sanity check to make sure this is a pipable WIM.  */
1792         if (pwm->hdr.magic != PWM_MAGIC) {
1793                 ERROR("The WIM being read from file descriptor %d "
1794                       "is not pipable!", pipe_fd);
1795                 ret = WIMLIB_ERR_NOT_PIPABLE;
1796                 goto out_wimlib_free;
1797         }
1798
1799         /* Sanity check to make sure the first part of a pipable split WIM is
1800          * sent over the pipe first.  */
1801         if (pwm->hdr.part_number != 1) {
1802                 ERROR("The first part of the split WIM must be "
1803                       "sent over the pipe first.");
1804                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1805                 goto out_wimlib_free;
1806         }
1807
1808         in_fd = &pwm->in_fd;
1809         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1810
1811         /* As mentioned, the WIMStruct we created from the pipe does not have
1812          * XML data yet.  Fix this by reading the extra copy of the XML data
1813          * that directly follows the header in pipable WIMs.  (Note: see
1814          * write_pipable_wim() for more details about the format of pipable
1815          * WIMs.)  */
1816         {
1817                 u8 hash[SHA1_HASH_SIZE];
1818
1819                 ret = read_pwm_blob_header(pwm, hash,
1820                                            &pwm->hdr.xml_data_reshdr, NULL);
1821                 if (ret)
1822                         goto out_wimlib_free;
1823
1824                 if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1825                         ERROR("Expected XML data, but found non-metadata resource.");
1826                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1827                         goto out_wimlib_free;
1828                 }
1829
1830                 ret = read_wim_xml_data(pwm);
1831                 if (ret)
1832                         goto out_wimlib_free;
1833
1834                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
1835                         ERROR("Image count in XML data is not the same as in WIM header.");
1836                         ret = WIMLIB_ERR_IMAGE_COUNT;
1837                         goto out_wimlib_free;
1838                 }
1839         }
1840
1841         /* Get image index (this may use the XML data that was just read to
1842          * resolve an image name).  */
1843         if (image_num_or_name) {
1844                 image = wimlib_resolve_image(pwm, image_num_or_name);
1845                 if (image == WIMLIB_NO_IMAGE) {
1846                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1847                               image_num_or_name);
1848                         ret = WIMLIB_ERR_INVALID_IMAGE;
1849                         goto out_wimlib_free;
1850                 } else if (image == WIMLIB_ALL_IMAGES) {
1851                         ERROR("Applying all images from a pipe is not supported!");
1852                         ret = WIMLIB_ERR_INVALID_IMAGE;
1853                         goto out_wimlib_free;
1854                 }
1855         } else {
1856                 if (pwm->hdr.image_count != 1) {
1857                         ERROR("No image was specified, but the pipable WIM "
1858                               "did not contain exactly 1 image");
1859                         ret = WIMLIB_ERR_INVALID_IMAGE;
1860                         goto out_wimlib_free;
1861                 }
1862                 image = 1;
1863         }
1864
1865         /* Load the needed metadata resource.  */
1866         for (i = 1; i <= pwm->hdr.image_count; i++) {
1867                 struct wim_image_metadata *imd;
1868                 struct wim_reshdr reshdr;
1869                 struct wim_resource_descriptor *metadata_rdesc;
1870
1871                 imd = pwm->image_metadata[i - 1];
1872
1873                 ret = WIMLIB_ERR_NOMEM;
1874                 imd->metadata_blob = new_blob_descriptor();
1875                 if (!imd->metadata_blob)
1876                         goto out_wimlib_free;
1877
1878                 imd->metadata_blob->is_metadata = 1;
1879
1880                 ret = read_pwm_blob_header(pwm, imd->metadata_blob->hash,
1881                                            &reshdr, NULL);
1882                 if (ret)
1883                         goto out_wimlib_free;
1884
1885                 if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1886                         ERROR("Expected metadata resource, but found "
1887                               "non-metadata resource");
1888                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1889                         goto out_wimlib_free;
1890                 }
1891
1892                 ret = WIMLIB_ERR_NOMEM;
1893                 metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
1894                 if (!metadata_rdesc)
1895                         goto out_wimlib_free;
1896                 wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc,
1897                                             imd->metadata_blob);
1898
1899                 if (i == image) {
1900                         /* Metadata resource is for the image being extracted.
1901                          * Parse it and save the metadata in memory.  */
1902                         ret = read_metadata_resource(imd);
1903                         if (ret)
1904                                 goto out_wimlib_free;
1905                         imd->modified = 1;
1906                 } else {
1907                         /* Metadata resource is not for the image being
1908                          * extracted.  Skip over it.  */
1909                         ret = skip_wim_resource(metadata_rdesc);
1910                         if (ret)
1911                                 goto out_wimlib_free;
1912                 }
1913         }
1914         /* Extract the image.  */
1915         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
1916         ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
1917         /* Clean up and return.  */
1918 out_wimlib_free:
1919         wimlib_free(pwm);
1920         return ret;
1921 }
1922
1923
1924 WIMLIBAPI int
1925 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
1926                                const tchar *target, int extract_flags)
1927 {
1928         return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
1929                                                             image_num_or_name,
1930                                                             target,
1931                                                             extract_flags,
1932                                                             NULL,
1933                                                             NULL);
1934 }
1935
1936 WIMLIBAPI int
1937 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1938                      int extract_flags)
1939 {
1940         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1941                 return WIMLIB_ERR_INVALID_PARAM;
1942         return do_wimlib_extract_image(wim, image, target, extract_flags);
1943 }