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          WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K           |       \
  89          WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K           |       \
  90          WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K          |       \
  91          WIMLIB_EXTRACT_FLAG_COMPACT_LZX                        \
  92          )
  93
  94 /* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or
  95  * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA.  */
  96 int
  97 do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
  98 {
  99         ctx->count_until_file_progress = 500;  /* Arbitrary value to limit calls  */
 100         return extract_progress(ctx, msg);
 101 }
 102
 103 static int
 104 start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg)
 105 {
 106         ctx->progress.extract.current_file_count = 0;
 107         ctx->progress.extract.end_file_count = end_file_count;
 108         return do_file_extract_progress(ctx, msg);
 109 }
 110
 111 int
 112 start_file_structure_phase(struct apply_ctx *ctx, u64 end_file_count)
 113 {
 114         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 115 }
 116
 117 int
 118 start_file_metadata_phase(struct apply_ctx *ctx, u64 end_file_count)
 119 {
 120         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 121 }
 122
 123 static int
 124 end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
 125 {
 126         ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count;
 127         return do_file_extract_progress(ctx, msg);
 128 }
 129
 130 int
 131 end_file_structure_phase(struct apply_ctx *ctx)
 132 {
 133         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 134 }
 135
 136 int
 137 end_file_metadata_phase(struct apply_ctx *ctx)
 138 {
 139         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 140 }
 141
 142 #define PWM_FOUND_WIM_HDR (-1)
 143
 144 /* Read the header for a blob in a pipable WIM.  If @pwm_hdr_ret is not NULL,
 145  * also look for a pipable WIM header and return PWM_FOUND_WIM_HDR if found.  */
 146 static int
 147 read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE],
 148                      struct wim_reshdr *reshdr_ret,
 149                      struct wim_header_disk *pwm_hdr_ret)
 150 {
 151         int ret;
 152         struct pwm_blob_hdr blob_hdr;
 153         u64 magic;
 154
 155         ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr));
 156         if (unlikely(ret))
 157                 goto read_error;
 158
 159         magic = le64_to_cpu(blob_hdr.magic);
 160
 161         if (magic == PWM_MAGIC && pwm_hdr_ret != NULL) {
 162                 memcpy(pwm_hdr_ret, &blob_hdr, sizeof(blob_hdr));
 163                 ret = full_read(&pwm->in_fd,
 164                                 (u8 *)pwm_hdr_ret + sizeof(blob_hdr),
 165                                 sizeof(*pwm_hdr_ret) - sizeof(blob_hdr));
 166                 if (unlikely(ret))
 167                         goto read_error;
 168                 return PWM_FOUND_WIM_HDR;
 169         }
 170
 171         if (unlikely(magic != PWM_BLOB_MAGIC)) {
 172                 ERROR("Data read on pipe is invalid (expected blob header)");
 173                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 174         }
 175
 176         copy_hash(hash_ret, blob_hdr.hash);
 177
 178         reshdr_ret->size_in_wim = 0; /* Not available  */
 179         reshdr_ret->flags = le32_to_cpu(blob_hdr.flags);
 180         reshdr_ret->offset_in_wim = pwm->in_fd.offset;
 181         reshdr_ret->uncompressed_size = le64_to_cpu(blob_hdr.uncompressed_size);
 182
 183         if (unlikely(reshdr_ret->uncompressed_size == 0)) {
 184                 ERROR("Data read on pipe is invalid (resource is of 0 size)");
 185                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 186         }
 187
 188         return 0;
 189
 190 read_error:
 191         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE)
 192                 ERROR("The pipe ended before all needed data was sent!");
 193         else
 194                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
 195         return ret;
 196 }
 197
 198 static int
 199 read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 200 {
 201         int ret;
 202         u8 hash[SHA1_HASH_SIZE];
 203         struct wim_reshdr reshdr;
 204         struct wim_header_disk pwm_hdr;
 205         struct wim_resource_descriptor rdesc;
 206         struct blob_descriptor *blob;
 207
 208         copy_guid(ctx->progress.extract.guid, ctx->wim->hdr.guid);
 209         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
 210         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
 211         ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 212         if (ret)
 213                 return ret;
 214
 215         while (ctx->num_blobs_remaining) {
 216
 217                 ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr);
 218
 219                 if (ret == PWM_FOUND_WIM_HDR) {
 220                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
 221                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
 222
 223                         if (part_number == ctx->progress.extract.part_number &&
 224                             total_parts == ctx->progress.extract.total_parts &&
 225                             guids_equal(pwm_hdr.guid, ctx->progress.extract.guid))
 226                                 continue;
 227
 228                         copy_guid(ctx->progress.extract.guid, pwm_hdr.guid);
 229                         ctx->progress.extract.part_number = part_number;
 230                         ctx->progress.extract.total_parts = total_parts;
 231                         ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 232                         if (ret)
 233                                 return ret;
 234
 235                         continue;
 236                 }
 237
 238                 if (ret)
 239                         return ret;
 240
 241                 if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)
 242                     && (blob = lookup_blob(ctx->wim->blob_table, hash))
 243                     && (blob->out_refcnt))
 244                 {
 245                         wim_reshdr_to_desc_and_blob(&reshdr, ctx->wim, &rdesc, blob);
 246                         ret = read_blob_with_sha1(blob, cbs);
 247                         blob_unset_is_located_in_wim_resource(blob);
 248                         if (ret)
 249                                 return ret;
 250                         ctx->num_blobs_remaining--;
 251                 } else {
 252                         wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc);
 253                         ret = skip_wim_resource(&rdesc);
 254                         if (ret)
 255                                 return ret;
 256                 }
 257         }
 258
 259         return 0;
 260 }
 261
 262 /* Creates a temporary file opened for writing.  The open file descriptor is
 263  * returned in @fd_ret and its name is returned in @name_ret (dynamically
 264  * allocated).  */
 265 static int
 266 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
 267 {
 268         tchar *name;
 269         int open_flags;
 270         int raw_fd;
 271
 272 retry:
 273         name = ttempnam(NULL, T("wimlib"));
 274         if (!name) {
 275                 ERROR_WITH_ERRNO("Failed to create temporary filename");
 276                 return WIMLIB_ERR_NOMEM;
 277         }
 278
 279         open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
 280 #ifdef __WIN32__
 281         open_flags |= _O_SHORT_LIVED;
 282 #endif
 283         raw_fd = topen(name, open_flags, 0600);
 284
 285         if (raw_fd < 0) {
 286                 if (errno == EEXIST) {
 287                         FREE(name);
 288                         goto retry;
 289                 }
 290                 ERROR_WITH_ERRNO("Failed to create temporary file "
 291                                  "\"%"TS"\"", name);
 292                 FREE(name);
 293                 return WIMLIB_ERR_OPEN;
 294         }
 295
 296         filedes_init(fd_ret, raw_fd);
 297         *name_ret = name;
 298         return 0;
 299 }
 300
 301 static int
 302 begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
 303 {
 304         struct apply_ctx *ctx = _ctx;
 305
 306         ctx->cur_blob = blob;
 307         ctx->cur_blob_offset = 0;
 308
 309         if (unlikely(blob->out_refcnt > MAX_OPEN_FILES))
 310                 return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
 311
 312         return call_begin_blob(blob, ctx->saved_cbs);
 313 }
 314
 315 static int
 316 extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
 317 {
 318         struct apply_ctx *ctx = _ctx;
 319         union wimlib_progress_info *progress = &ctx->progress;
 320         int ret;
 321
 322         ctx->cur_blob_offset += size;
 323
 324         if (likely(ctx->supported_features.hard_links)) {
 325                 progress->extract.completed_bytes +=
 326                         (u64)size * ctx->cur_blob->out_refcnt;
 327                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 328                         progress->extract.completed_streams += ctx->cur_blob->out_refcnt;
 329         } else {
 330                 const struct blob_extraction_target *targets =
 331                         blob_extraction_targets(ctx->cur_blob);
 332                 for (u32 i = 0; i < ctx->cur_blob->out_refcnt; i++) {
 333                         const struct wim_inode *inode = targets[i].inode;
 334                         const struct wim_dentry *dentry;
 335
 336                         inode_for_each_extraction_alias(dentry, inode) {
 337                                 progress->extract.completed_bytes += size;
 338                                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 339                                         progress->extract.completed_streams++;
 340                         }
 341                 }
 342         }
 343         if (progress->extract.completed_bytes >= ctx->next_progress) {
 344
 345                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
 346                 if (ret)
 347                         return ret;
 348
 349                 set_next_progress(progress->extract.completed_bytes,
 350                                   progress->extract.total_bytes,
 351                                   &ctx->next_progress);
 352         }
 353
 354         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 355                 /* Just extracting to temporary file for now.  */
 356                 ret = full_write(&ctx->tmpfile_fd, chunk, size);
 357                 if (ret) {
 358                         ERROR_WITH_ERRNO("Error writing data to "
 359                                          "temporary file \"%"TS"\"",
 360                                          ctx->tmpfile_name);
 361                 }
 362                 return ret;
 363         }
 364
 365         return call_consume_chunk(chunk, size, ctx->saved_cbs);
 366 }
 367
 368 /* Copy the blob's data from the temporary file to each of its targets.
 369  *
 370  * This is executed only in the very uncommon case that a blob is being
 371  * extracted to more than MAX_OPEN_FILES targets!  */
 372 static int
 373 extract_from_tmpfile(const tchar *tmpfile_name,
 374                      const struct blob_descriptor *orig_blob,
 375                      const struct read_blob_callbacks *cbs)
 376 {
 377         struct blob_descriptor tmpfile_blob;
 378         const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob);
 379         int ret;
 380
 381         memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor));
 382         tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK;
 383         tmpfile_blob.file_on_disk = (tchar *)tmpfile_name;
 384         tmpfile_blob.out_refcnt = 1;
 385
 386         for (u32 i = 0; i < orig_blob->out_refcnt; i++) {
 387                 tmpfile_blob.inline_blob_extraction_targets[0] = targets[i];
 388                 ret = read_blob_with_cbs(&tmpfile_blob, cbs);
 389                 if (ret)
 390                         return ret;
 391         }
 392         return 0;
 393 }
 394
 395 static int
 396 end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
 397 {
 398         struct apply_ctx *ctx = _ctx;
 399
 400         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 401                 filedes_close(&ctx->tmpfile_fd);
 402                 if (!status)
 403                         status = extract_from_tmpfile(ctx->tmpfile_name, blob,
 404                                                       ctx->saved_cbs);
 405                 filedes_invalidate(&ctx->tmpfile_fd);
 406                 tunlink(ctx->tmpfile_name);
 407                 FREE(ctx->tmpfile_name);
 408                 return status;
 409         }
 410
 411         return call_end_blob(blob, status, ctx->saved_cbs);
 412 }
 413
 414 /*
 415  * Read the list of blobs to extract and feed their data into the specified
 416  * callback functions.
 417  *
 418  * This handles checksumming each blob.
 419  *
 420  * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
 421  *
 422  * This also works if the WIM is being read from a pipe.
 423  *
 424  * This also will split up blobs that will need to be extracted to more than
 425  * MAX_OPEN_FILES locations, as measured by the 'out_refcnt' of each blob.
 426  * Therefore, the apply_operations implementation need not worry about running
 427  * out of file descriptors, unless it might open more than one file descriptor
 428  * per 'blob_extraction_target' (e.g. Win32 currently might because the
 429  * destination file system might not support hard links).
 430  */
 431 int
 432 extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 433 {
 434         struct read_blob_callbacks wrapper_cbs = {
 435                 .begin_blob     = begin_extract_blob_wrapper,
 436                 .consume_chunk  = extract_chunk_wrapper,
 437                 .end_blob       = end_extract_blob_wrapper,
 438                 .ctx            = ctx,
 439         };
 440         ctx->saved_cbs = cbs;
 441         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
 442                 return read_blobs_from_pipe(ctx, &wrapper_cbs);
 443         } else {
 444                 return read_blob_list(&ctx->blob_list,
 445                                       offsetof(struct blob_descriptor,
 446                                                extraction_list),
 447                                       &wrapper_cbs, VERIFY_BLOB_HASHES);
 448         }
 449 }
 450
 451 /* Extract a WIM dentry to standard output.
 452  *
 453  * This obviously doesn't make sense in all cases.  We return an error if the
 454  * dentry does not correspond to a regular file.  Otherwise we extract the
 455  * unnamed data stream only.  */
 456 static int
 457 extract_dentry_to_stdout(struct wim_dentry *dentry,
 458                          const struct blob_table *blob_table)
 459 {
 460         struct wim_inode *inode = dentry->d_inode;
 461         struct blob_descriptor *blob;
 462         struct filedes _stdout;
 463
 464         if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
 465                                    FILE_ATTRIBUTE_DIRECTORY |
 466                                    FILE_ATTRIBUTE_ENCRYPTED))
 467         {
 468                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 469                       "extracted to standard output", dentry_full_path(dentry));
 470                 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
 471         }
 472
 473         blob = inode_get_blob_for_unnamed_data_stream(inode, blob_table);
 474         if (!blob) {
 475                 const u8 *hash = inode_get_hash_of_unnamed_data_stream(inode);
 476                 if (!is_zero_hash(hash))
 477                         return blob_not_found_error(inode, hash);
 478                 return 0;
 479         }
 480
 481         filedes_init(&_stdout, STDOUT_FILENO);
 482         return extract_blob_to_fd(blob, &_stdout);
 483 }
 484
 485 static int
 486 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
 487                            const struct blob_table *blob_table)
 488 {
 489         for (size_t i = 0; i < num_dentries; i++) {
 490                 int ret = extract_dentry_to_stdout(dentries[i], blob_table);
 491                 if (ret)
 492                         return ret;
 493         }
 494         return 0;
 495 }
 496
 497 /**********************************************************************/
 498
 499 /*
 500  * Removes duplicate dentries from the array.
 501  *
 502  * Returns the new number of dentries, packed at the front of the array.
 503  */
 504 static size_t
 505 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
 506 {
 507         size_t i, j = 0;
 508         for (i = 0; i < num_trees; i++) {
 509                 if (!trees[i]->d_tmp_flag) {
 510                         /* Found distinct dentry.  */
 511                         trees[i]->d_tmp_flag = 1;
 512                         trees[j++] = trees[i];
 513                 }
 514         }
 515         for (i = 0; i < j; i++)
 516                 trees[i]->d_tmp_flag = 0;
 517         return j;
 518 }
 519
 520 /*
 521  * Remove dentries that are descendants of other dentries in the array.
 522  *
 523  * Returns the new number of dentries, packed at the front of the array.
 524  */
 525 static size_t
 526 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
 527 {
 528         size_t i, j = 0;
 529         for (i = 0; i < num_trees; i++)
 530                 trees[i]->d_tmp_flag = 1;
 531         for (i = 0; i < num_trees; i++) {
 532                 struct wim_dentry *d = trees[i];
 533                 while (!dentry_is_root(d)) {
 534                         d = d->d_parent;
 535                         if (d->d_tmp_flag)
 536                                 goto tree_contained;
 537                 }
 538                 trees[j++] = trees[i];
 539                 continue;
 540
 541         tree_contained:
 542                 trees[i]->d_tmp_flag = 0;
 543         }
 544
 545         for (i = 0; i < j; i++)
 546                 trees[i]->d_tmp_flag = 0;
 547         return j;
 548 }
 549
 550 static int
 551 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
 552 {
 553         struct list_head *dentry_list = _dentry_list;
 554         list_add_tail(&dentry->d_extraction_list_node, dentry_list);
 555         return 0;
 556 }
 557
 558 static void
 559 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
 560 {
 561         dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
 562 }
 563
 564 static int
 565 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
 566 {
 567         if (will_extract_dentry(dentry)) {
 568                 list_del(&dentry->d_extraction_list_node);
 569                 dentry_reset_extraction_list_node(dentry);
 570         }
 571         return 0;
 572 }
 573
 574 /*
 575  * Build the preliminary list of dentries to be extracted.
 576  *
 577  * The list maintains the invariant that if d1 and d2 are in the list and d1 is
 578  * an ancestor of d2, then d1 appears before d2 in the list.
 579  */
 580 static void
 581 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
 582                   size_t num_trees, bool add_ancestors)
 583 {
 584         INIT_LIST_HEAD(dentry_list);
 585
 586         /* Add the trees recursively.  */
 587         for (size_t i = 0; i < num_trees; i++)
 588                 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
 589
 590         /* If requested, add ancestors of the trees.  */
 591         if (add_ancestors) {
 592                 for (size_t i = 0; i < num_trees; i++) {
 593                         struct wim_dentry *dentry = trees[i];
 594                         struct wim_dentry *ancestor;
 595                         struct list_head *place_after;
 596
 597                         if (dentry_is_root(dentry))
 598                                 continue;
 599
 600                         place_after = dentry_list;
 601                         ancestor = dentry;
 602                         do {
 603                                 ancestor = ancestor->d_parent;
 604                                 if (will_extract_dentry(ancestor)) {
 605                                         place_after = &ancestor->d_extraction_list_node;
 606                                         break;
 607                                 }
 608                         } while (!dentry_is_root(ancestor));
 609
 610                         ancestor = dentry;
 611                         do {
 612                                 ancestor = ancestor->d_parent;
 613                                 if (will_extract_dentry(ancestor))
 614                                         break;
 615                                 list_add(&ancestor->d_extraction_list_node, place_after);
 616                         } while (!dentry_is_root(ancestor));
 617                 }
 618         }
 619 }
 620
 621 static void
 622 destroy_dentry_list(struct list_head *dentry_list)
 623 {
 624         struct wim_dentry *dentry, *tmp;
 625         struct wim_inode *inode;
 626
 627         list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
 628                 inode = dentry->d_inode;
 629                 dentry_reset_extraction_list_node(dentry);
 630                 inode->i_visited = 0;
 631                 inode->i_can_externally_back = 0;
 632                 if ((void *)dentry->d_extraction_name != (void *)dentry->d_name)
 633                         FREE(dentry->d_extraction_name);
 634                 dentry->d_extraction_name = NULL;
 635                 dentry->d_extraction_name_nchars = 0;
 636         }
 637 }
 638
 639 static void
 640 destroy_blob_list(struct list_head *blob_list)
 641 {
 642         struct blob_descriptor *blob;
 643
 644         list_for_each_entry(blob, blob_list, extraction_list)
 645                 if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets))
 646                         FREE(blob->blob_extraction_targets);
 647 }
 648
 649 #ifdef __WIN32__
 650 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
 651 #else
 652 static const utf16lechar replacement_char = cpu_to_le16('?');
 653 #endif
 654
 655 static bool
 656 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
 657 {
 658         size_t i;
 659
 660         if (num_chars == 0)
 661                 return true;
 662         for (i = 0; i < num_chars; i++) {
 663                 switch (le16_to_cpu(name[i])) {
 664         #ifdef __WIN32__
 665                 case '\x01'...'\x1F':
 666                 case '\\':
 667                 case ':':
 668                 case '*':
 669                 case '?':
 670                 case '"':
 671                 case '<':
 672                 case '>':
 673                 case '|':
 674         #endif
 675                 case '/':
 676                 case '\0':
 677                         if (fix)
 678                                 name[i] = replacement_char;
 679                         else
 680                                 return false;
 681                 }
 682         }
 683
 684         return true;
 685 }
 686
 687 static int
 688 dentry_calculate_extraction_name(struct wim_dentry *dentry,
 689                                  struct apply_ctx *ctx)
 690 {
 691         int ret;
 692
 693         if (dentry_is_root(dentry))
 694                 return 0;
 695
 696 #ifdef WITH_NTFS_3G
 697         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 698                 dentry->d_extraction_name = dentry->d_name;
 699                 dentry->d_extraction_name_nchars = dentry->d_name_nbytes /
 700                                                    sizeof(utf16lechar);
 701                 return 0;
 702         }
 703 #endif
 704
 705         if (!ctx->supported_features.case_sensitive_filenames) {
 706                 struct wim_dentry *other;
 707                 list_for_each_entry(other, &dentry->d_ci_conflict_list,
 708                                     d_ci_conflict_list)
 709                 {
 710                         if (will_extract_dentry(other)) {
 711                                 if (ctx->extract_flags &
 712                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
 713                                         WARNING("\"%"TS"\" has the same "
 714                                                 "case-insensitive name as "
 715                                                 "\"%"TS"\"; extracting "
 716                                                 "dummy name instead",
 717                                                 dentry_full_path(dentry),
 718                                                 dentry_full_path(other));
 719                                         goto out_replace;
 720                                 } else {
 721                                         WARNING("Not extracting \"%"TS"\": "
 722                                                 "has same case-insensitive "
 723                                                 "name as \"%"TS"\"",
 724                                                 dentry_full_path(dentry),
 725                                                 dentry_full_path(other));
 726                                         goto skip_dentry;
 727                                 }
 728                         }
 729                 }
 730         }
 731
 732         if (file_name_valid(dentry->d_name, dentry->d_name_nbytes / 2, false)) {
 733                 size_t nbytes = 0;
 734                 ret = utf16le_get_tstr(dentry->d_name,
 735                                        dentry->d_name_nbytes,
 736                                        (const tchar **)&dentry->d_extraction_name,
 737                                        &nbytes);
 738                 dentry->d_extraction_name_nchars = nbytes / sizeof(tchar);
 739                 return ret;
 740         } else {
 741                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
 742                 {
 743                         WARNING("\"%"TS"\" has an invalid filename "
 744                                 "that is not supported on this platform; "
 745                                 "extracting dummy name instead",
 746                                 dentry_full_path(dentry));
 747                         goto out_replace;
 748                 } else {
 749                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
 750                                 "that is not supported on this platform",
 751                                 dentry_full_path(dentry));
 752                         goto skip_dentry;
 753                 }
 754         }
 755
 756 out_replace:
 757         {
 758                 utf16lechar utf16_name_copy[dentry->d_name_nbytes / 2];
 759
 760                 memcpy(utf16_name_copy, dentry->d_name, dentry->d_name_nbytes);
 761                 file_name_valid(utf16_name_copy, dentry->d_name_nbytes / 2, true);
 762
 763                 const tchar *tchar_name;
 764                 size_t tchar_nchars;
 765
 766                 ret = utf16le_get_tstr(utf16_name_copy,
 767                                        dentry->d_name_nbytes,
 768                                        &tchar_name, &tchar_nchars);
 769                 if (ret)
 770                         return ret;
 771
 772                 tchar_nchars /= sizeof(tchar);
 773
 774                 size_t fixed_name_num_chars = tchar_nchars;
 775                 tchar fixed_name[tchar_nchars + 50];
 776
 777                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
 778                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
 779                                                  T(" (invalid filename #%lu)"),
 780                                                  ++ctx->invalid_sequence);
 781
 782                 utf16le_put_tstr(tchar_name);
 783
 784                 dentry->d_extraction_name = TSTRDUP(fixed_name);
 785                 if (!dentry->d_extraction_name)
 786                         return WIMLIB_ERR_NOMEM;
 787                 dentry->d_extraction_name_nchars = fixed_name_num_chars;
 788         }
 789         return 0;
 790
 791 skip_dentry:
 792         for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
 793         return 0;
 794 }
 795
 796 /*
 797  * Calculate the actual filename component at which each WIM dentry will be
 798  * extracted, with special handling for dentries that are unsupported by the
 799  * extraction backend or have invalid names.
 800  *
 801  * ctx->supported_features must be filled in.
 802  *
 803  * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
 804  */
 805 static int
 806 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
 807                                        struct apply_ctx *ctx)
 808 {
 809         struct list_head *prev, *cur;
 810
 811         /* Can't use list_for_each_entry() because a call to
 812          * dentry_calculate_extraction_name() may delete the current dentry and
 813          * its children from the list.  */
 814
 815         prev = dentry_list;
 816         for (;;) {
 817                 struct wim_dentry *dentry;
 818                 int ret;
 819
 820                 cur = prev->next;
 821                 if (cur == dentry_list)
 822                         break;
 823
 824                 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
 825
 826                 ret = dentry_calculate_extraction_name(dentry, ctx);
 827                 if (ret)
 828                         return ret;
 829
 830                 if (prev->next == cur)
 831                         prev = cur;
 832                 else
 833                         ; /* Current dentry and its children (which follow in
 834                              the list) were deleted.  prev stays the same.  */
 835         }
 836         return 0;
 837 }
 838
 839 static int
 840 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
 841                        struct blob_table *blob_table)
 842 {
 843         struct wim_inode *inode = dentry->d_inode;
 844         struct blob_descriptor *blob;
 845         int ret;
 846         bool force = false;
 847
 848         /* Special case:  when extracting from a pipe, the WIM blob table is
 849          * initially empty, so "resolving" an inode's streams is initially not
 850          * possible.  However, we still need to keep track of which blobs,
 851          * identified by SHA-1 message digests, need to be extracted, so we
 852          * "resolve" the inode's streams anyway by allocating a 'struct
 853          * blob_descriptor' for each one.  */
 854         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
 855                 force = true;
 856         ret = inode_resolve_streams(inode, blob_table, force);
 857         if (ret)
 858                 return ret;
 859         for (unsigned i = 0; i < inode->i_num_streams; i++) {
 860                 blob = stream_blob_resolved(&inode->i_streams[i]);
 861                 if (blob)
 862                         blob->out_refcnt = 0;
 863         }
 864         return 0;
 865 }
 866
 867 /*
 868  * For each dentry to be extracted, resolve all streams in the corresponding
 869  * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
 870  *
 871  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
 872  */
 873 static int
 874 dentry_list_resolve_streams(struct list_head *dentry_list,
 875                             struct apply_ctx *ctx)
 876 {
 877         struct wim_dentry *dentry;
 878         int ret;
 879
 880         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
 881                 ret = dentry_resolve_streams(dentry,
 882                                              ctx->extract_flags,
 883                                              ctx->wim->blob_table);
 884                 if (ret)
 885                         return ret;
 886         }
 887         return 0;
 888 }
 889
 890 static int
 891 ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
 892            struct apply_ctx *ctx)
 893 {
 894         struct wim_inode *inode = dentry->d_inode;
 895         struct blob_descriptor *blob = stream_blob_resolved(strm);
 896         struct blob_extraction_target *targets;
 897
 898         if (!blob)
 899                 return 0;
 900
 901         /* Tally the size only for each actual extraction of the stream (not
 902          * additional hard links to the inode).  */
 903         if (inode->i_visited && ctx->supported_features.hard_links)
 904                 return 0;
 905
 906         ctx->progress.extract.total_bytes += blob->size;
 907         ctx->progress.extract.total_streams++;
 908
 909         if (inode->i_visited)
 910                 return 0;
 911
 912         /* Add each blob to 'ctx->blob_list' only one time, regardless of how
 913          * many extraction targets it will have.  */
 914         if (blob->out_refcnt == 0) {
 915                 list_add_tail(&blob->extraction_list, &ctx->blob_list);
 916                 ctx->num_blobs_remaining++;
 917         }
 918
 919         /* Set this stream as an extraction target of 'blob'.  */
 920
 921         if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 922                 targets = blob->inline_blob_extraction_targets;
 923         } else {
 924                 struct blob_extraction_target *prev_targets;
 925                 size_t alloc_blob_extraction_targets;
 926
 927                 if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 928                         prev_targets = NULL;
 929                         alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets);
 930                 } else {
 931                         prev_targets = blob->blob_extraction_targets;
 932                         alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
 933                 }
 934
 935                 if (blob->out_refcnt == alloc_blob_extraction_targets) {
 936                         alloc_blob_extraction_targets *= 2;
 937                         targets = REALLOC(prev_targets,
 938                                           alloc_blob_extraction_targets *
 939                                           sizeof(targets[0]));
 940                         if (!targets)
 941                                 return WIMLIB_ERR_NOMEM;
 942                         if (!prev_targets) {
 943                                 memcpy(targets,
 944                                        blob->inline_blob_extraction_targets,
 945                                        sizeof(blob->inline_blob_extraction_targets));
 946                         }
 947                         blob->blob_extraction_targets = targets;
 948                         blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
 949                 }
 950                 targets = blob->blob_extraction_targets;
 951         }
 952         targets[blob->out_refcnt].inode = inode;
 953         targets[blob->out_refcnt].stream = strm;
 954         blob->out_refcnt++;
 955         return 0;
 956 }
 957
 958 static int
 959 ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode,
 960                      struct wim_inode_stream *strm, struct apply_ctx *ctx)
 961 {
 962         bool need_stream = false;
 963         switch (strm->stream_type) {
 964         case STREAM_TYPE_DATA:
 965                 if (stream_is_named(strm)) {
 966                         /* Named data stream  */
 967                         if (ctx->supported_features.named_data_streams)
 968                                 need_stream = true;
 969                 } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
 970                                                     FILE_ATTRIBUTE_ENCRYPTED))
 971                            && !(inode_is_symlink(inode)
 972                                 && !ctx->supported_features.reparse_points
 973                                 && ctx->supported_features.symlink_reparse_points))
 974                 {
 975                         /*
 976                          * Unnamed data stream.  Skip if any of the following is true:
 977                          *
 978                          * - file is a directory
 979                          * - file is encrypted
 980                          * - backend needs to create the file as UNIX symlink
 981                          * - backend will extract the stream as externally
 982                          *   backed from the WIM archive itself
 983                          */
 984                         if (ctx->apply_ops->will_back_from_wim) {
 985                                 int ret = (*ctx->apply_ops->will_back_from_wim)(dentry, ctx);
 986                                 if (ret > 0) /* Error?  */
 987                                         return ret;
 988                                 if (ret < 0) /* Won't externally back?  */
 989                                         need_stream = true;
 990                         } else {
 991                                 need_stream = true;
 992                         }
 993                 }
 994                 break;
 995         case STREAM_TYPE_REPARSE_POINT:
 996                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT);
 997                 if (ctx->supported_features.reparse_points ||
 998                     (inode_is_symlink(inode) &&
 999                      ctx->supported_features.symlink_reparse_points))
1000                         need_stream = true;
1001                 break;
1002         case STREAM_TYPE_EFSRPC_RAW_DATA:
1003                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED);
1004                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
1005                         if (ctx->supported_features.encrypted_directories)
1006                                 need_stream = true;
1007                 } else {
1008                         if (ctx->supported_features.encrypted_files)
1009                                 need_stream = true;
1010                 }
1011                 break;
1012         }
1013         if (need_stream)
1014                 return ref_stream(strm, dentry, ctx);
1015         return 0;
1016 }
1017
1018 static int
1019 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
1020 {
1021         struct wim_inode *inode = dentry->d_inode;
1022         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1023                 int ret = ref_stream_if_needed(dentry, inode,
1024                                                &inode->i_streams[i], ctx);
1025                 if (ret)
1026                         return ret;
1027         }
1028         inode->i_visited = 1;
1029         return 0;
1030 }
1031
1032 /*
1033  * Given a list of dentries to be extracted, build the list of blobs that need
1034  * to be extracted, and for each blob determine the streams to which that blob
1035  * will be extracted.
1036  *
1037  * This also initializes the extract progress info with byte and blob
1038  * information.
1039  *
1040  * ctx->supported_features must be filled in.
1041  */
1042 static int
1043 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
1044 {
1045         struct wim_dentry *dentry;
1046         int ret;
1047
1048         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1049                 ret = dentry_ref_streams(dentry, ctx);
1050                 if (ret)
1051                         return ret;
1052         }
1053         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1054                 dentry->d_inode->i_visited = 0;
1055         return 0;
1056 }
1057
1058 static void
1059 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
1060 {
1061         struct wim_dentry *dentry;
1062
1063         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1064                 dentry->d_inode->i_first_extraction_alias = NULL;
1065
1066         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1067                 dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias;
1068                 dentry->d_inode->i_first_extraction_alias = dentry;
1069         }
1070 }
1071
1072 static void
1073 inode_tally_features(const struct wim_inode *inode,
1074                      struct wim_features *features)
1075 {
1076         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1077                 features->archive_files++;
1078         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1079                 features->hidden_files++;
1080         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1081                 features->system_files++;
1082         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1083                 features->compressed_files++;
1084         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1085                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1086                         features->encrypted_directories++;
1087                 else
1088                         features->encrypted_files++;
1089         }
1090         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1091                 features->not_context_indexed_files++;
1092         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1093                 features->sparse_files++;
1094         if (inode_has_named_data_stream(inode))
1095                 features->named_data_streams++;
1096         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1097                 features->reparse_points++;
1098                 if (inode_is_symlink(inode))
1099                         features->symlink_reparse_points++;
1100                 else
1101                         features->other_reparse_points++;
1102         }
1103         if (inode_has_security_descriptor(inode))
1104                 features->security_descriptors++;
1105         if (inode_has_unix_data(inode))
1106                 features->unix_data++;
1107 }
1108
1109 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1110 static void
1111 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
1112 {
1113         struct wim_inode *inode = dentry->d_inode;
1114
1115         if (dentry_has_short_name(dentry))
1116                 features->short_names++;
1117
1118         if (inode->i_visited) {
1119                 features->hard_links++;
1120         } else {
1121                 inode_tally_features(inode, features);
1122                 inode->i_visited = 1;
1123         }
1124 }
1125
1126 /* Tally the features necessary to extract the specified dentries.  */
1127 static void
1128 dentry_list_get_features(struct list_head *dentry_list,
1129                          struct wim_features *features)
1130 {
1131         struct wim_dentry *dentry;
1132
1133         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1134                 dentry_tally_features(dentry, features);
1135
1136         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1137                 dentry->d_inode->i_visited = 0;
1138 }
1139
1140 static int
1141 do_feature_check(const struct wim_features *required_features,
1142                  const struct wim_features *supported_features,
1143                  int extract_flags)
1144 {
1145         /* Encrypted files.  */
1146         if (required_features->encrypted_files &&
1147             !supported_features->encrypted_files)
1148                 WARNING("Ignoring EFS-encrypted data of %lu files",
1149                         required_features->encrypted_files);
1150
1151         /* Named data streams.  */
1152         if (required_features->named_data_streams &&
1153             !supported_features->named_data_streams)
1154                 WARNING("Ignoring named data streams of %lu files",
1155                         required_features->named_data_streams);
1156
1157         /* File attributes.  */
1158         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1159                 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1160                  * We're an archive program, so theoretically we can do what we
1161                  * want with it.  */
1162
1163                 if (required_features->hidden_files &&
1164                     !supported_features->hidden_files)
1165                         WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1166                                 required_features->hidden_files);
1167
1168                 if (required_features->system_files &&
1169                     !supported_features->system_files)
1170                         WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1171                                 required_features->system_files);
1172
1173                 if (required_features->compressed_files &&
1174                     !supported_features->compressed_files)
1175                         WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1176                                 required_features->compressed_files);
1177
1178                 if (required_features->not_context_indexed_files &&
1179                     !supported_features->not_context_indexed_files)
1180                         WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1181                                 required_features->not_context_indexed_files);
1182
1183                 if (required_features->sparse_files &&
1184                     !supported_features->sparse_files)
1185                         WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1186                                 required_features->sparse_files);
1187
1188                 if (required_features->encrypted_directories &&
1189                     !supported_features->encrypted_directories)
1190                         WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1191                                 required_features->encrypted_directories);
1192         }
1193
1194         /* Hard links.  */
1195         if (required_features->hard_links && !supported_features->hard_links)
1196                 WARNING("Extracting %lu hard links as independent files",
1197                         required_features->hard_links);
1198
1199         /* Symbolic links and reparse points.  */
1200         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1201             required_features->symlink_reparse_points &&
1202             !supported_features->symlink_reparse_points &&
1203             !supported_features->reparse_points)
1204         {
1205                 ERROR("Extraction backend does not support symbolic links!");
1206                 return WIMLIB_ERR_UNSUPPORTED;
1207         }
1208         if (required_features->reparse_points &&
1209             !supported_features->reparse_points)
1210         {
1211                 if (supported_features->symlink_reparse_points) {
1212                         if (required_features->other_reparse_points) {
1213                                 WARNING("Ignoring reparse data of %lu non-symlink/junction files",
1214                                         required_features->other_reparse_points);
1215                         }
1216                 } else {
1217                         WARNING("Ignoring reparse data of %lu files",
1218                                 required_features->reparse_points);
1219                 }
1220         }
1221
1222         /* Security descriptors.  */
1223         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1224                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1225              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1226             required_features->security_descriptors &&
1227             !supported_features->security_descriptors)
1228         {
1229                 ERROR("Extraction backend does not support security descriptors!");
1230                 return WIMLIB_ERR_UNSUPPORTED;
1231         }
1232         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1233             required_features->security_descriptors &&
1234             !supported_features->security_descriptors)
1235                 WARNING("Ignoring Windows NT security descriptors of %lu files",
1236                         required_features->security_descriptors);
1237
1238         /* UNIX data.  */
1239         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1240             required_features->unix_data && !supported_features->unix_data)
1241         {
1242                 ERROR("Extraction backend does not support UNIX data!");
1243                 return WIMLIB_ERR_UNSUPPORTED;
1244         }
1245
1246         if (required_features->unix_data &&
1247             !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1248         {
1249                 WARNING("Ignoring UNIX metadata of %lu files",
1250                         required_features->unix_data);
1251         }
1252
1253         /* DOS Names.  */
1254         if (required_features->short_names &&
1255             !supported_features->short_names)
1256         {
1257                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1258                         ERROR("Extraction backend does not support DOS names!");
1259                         return WIMLIB_ERR_UNSUPPORTED;
1260                 }
1261                 WARNING("Ignoring DOS names of %lu files",
1262                         required_features->short_names);
1263         }
1264
1265         /* Timestamps.  */
1266         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1267             !supported_features->timestamps)
1268         {
1269                 ERROR("Extraction backend does not support timestamps!");
1270                 return WIMLIB_ERR_UNSUPPORTED;
1271         }
1272
1273         return 0;
1274 }
1275
1276 static const struct apply_operations *
1277 select_apply_operations(int extract_flags)
1278 {
1279 #ifdef WITH_NTFS_3G
1280         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1281                 return &ntfs_3g_apply_ops;
1282 #endif
1283 #ifdef __WIN32__
1284         return &win32_apply_ops;
1285 #else
1286         return &unix_apply_ops;
1287 #endif
1288 }
1289
1290 static int
1291 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1292               const tchar *target, int extract_flags)
1293 {
1294         const struct apply_operations *ops;
1295         struct apply_ctx *ctx;
1296         int ret;
1297         LIST_HEAD(dentry_list);
1298
1299         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1300                 ret = extract_dentries_to_stdout(trees, num_trees,
1301                                                  wim->blob_table);
1302                 goto out;
1303         }
1304
1305         num_trees = remove_duplicate_trees(trees, num_trees);
1306         num_trees = remove_contained_trees(trees, num_trees);
1307
1308         ops = select_apply_operations(extract_flags);
1309
1310         if (num_trees > 1 && ops->single_tree_only) {
1311                 ERROR("Extracting multiple directory trees "
1312                       "at once is not supported in %s extraction mode!",
1313                       ops->name);
1314                 ret = WIMLIB_ERR_UNSUPPORTED;
1315                 goto out;
1316         }
1317
1318         ctx = CALLOC(1, ops->context_size);
1319         if (!ctx) {
1320                 ret = WIMLIB_ERR_NOMEM;
1321                 goto out;
1322         }
1323
1324         ctx->wim = wim;
1325         ctx->target = target;
1326         ctx->target_nchars = tstrlen(target);
1327         ctx->extract_flags = extract_flags;
1328         if (ctx->wim->progfunc) {
1329                 ctx->progfunc = ctx->wim->progfunc;
1330                 ctx->progctx = ctx->wim->progctx;
1331                 ctx->progress.extract.image = wim->current_image;
1332                 ctx->progress.extract.extract_flags = (extract_flags &
1333                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
1334                 ctx->progress.extract.wimfile_name = wim->filename;
1335                 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1336                                                                          wim->current_image);
1337                 ctx->progress.extract.target = target;
1338         }
1339         INIT_LIST_HEAD(&ctx->blob_list);
1340         filedes_invalidate(&ctx->tmpfile_fd);
1341         ctx->apply_ops = ops;
1342
1343         ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1344         if (ret)
1345                 goto out_cleanup;
1346
1347         build_dentry_list(&dentry_list, trees, num_trees,
1348                           !(extract_flags &
1349                             WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1350
1351         dentry_list_get_features(&dentry_list, &ctx->required_features);
1352
1353         ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1354                                ctx->extract_flags);
1355         if (ret)
1356                 goto out_cleanup;
1357
1358         ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1359         if (ret)
1360                 goto out_cleanup;
1361
1362         if (unlikely(list_empty(&dentry_list))) {
1363                 WARNING("There is nothing to extract!");
1364                 goto out_cleanup;
1365         }
1366
1367         ret = dentry_list_resolve_streams(&dentry_list, ctx);
1368         if (ret)
1369                 goto out_cleanup;
1370
1371         dentry_list_build_inode_alias_lists(&dentry_list);
1372
1373         ret = dentry_list_ref_streams(&dentry_list, ctx);
1374         if (ret)
1375                 goto out_cleanup;
1376
1377         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1378                 /* When extracting from a pipe, the number of bytes of data to
1379                  * extract can't be determined in the normal way (examining the
1380                  * blob table), since at this point all we have is a set of
1381                  * SHA-1 message digests of blobs that need to be extracted.
1382                  * However, we can get a reasonably accurate estimate by taking
1383                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1384                  * data.  This does assume that a full image is being extracted,
1385                  * but currently there is no API for doing otherwise.  (Also,
1386                  * subtract <HARDLINKBYTES> from this if hard links are
1387                  * supported by the extraction mode.)  */
1388                 ctx->progress.extract.total_bytes =
1389                         xml_get_image_total_bytes(wim->xml_info,
1390                                                   wim->current_image);
1391                 if (ctx->supported_features.hard_links) {
1392                         ctx->progress.extract.total_bytes -=
1393                                 xml_get_image_hard_link_bytes(wim->xml_info,
1394                                                               wim->current_image);
1395                 }
1396         }
1397
1398         ret = extract_progress(ctx,
1399                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1400                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
1401                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
1402         if (ret)
1403                 goto out_cleanup;
1404
1405         ret = (*ops->extract)(&dentry_list, ctx);
1406         if (ret)
1407                 goto out_cleanup;
1408
1409         if (ctx->progress.extract.completed_bytes <
1410             ctx->progress.extract.total_bytes)
1411         {
1412                 ctx->progress.extract.completed_bytes =
1413                         ctx->progress.extract.total_bytes;
1414                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
1415                 if (ret)
1416                         goto out_cleanup;
1417         }
1418
1419         ret = extract_progress(ctx,
1420                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1421                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
1422                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
1423 out_cleanup:
1424         destroy_blob_list(&ctx->blob_list);
1425         destroy_dentry_list(&dentry_list);
1426         FREE(ctx);
1427 out:
1428         return ret;
1429 }
1430
1431 static int
1432 mkdir_if_needed(const tchar *target)
1433 {
1434         if (!tmkdir(target, 0755))
1435                 return 0;
1436
1437         if (errno == EEXIST)
1438                 return 0;
1439
1440 #ifdef __WIN32__
1441         /* _wmkdir() fails with EACCES if called on a drive root directory.  */
1442         if (errno == EACCES)
1443                 return 0;
1444 #endif
1445
1446         ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
1447         return WIMLIB_ERR_MKDIR;
1448 }
1449
1450 /* Make sure the extraction flags make sense, and update them if needed.  */
1451 static int
1452 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1453 {
1454         int extract_flags = *extract_flags_p;
1455
1456         /* Check for invalid flag combinations  */
1457
1458         if ((extract_flags &
1459              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1460               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1461                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1462                 return WIMLIB_ERR_INVALID_PARAM;
1463
1464         if ((extract_flags &
1465              (WIMLIB_EXTRACT_FLAG_RPFIX |
1466               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1467                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
1468                 return WIMLIB_ERR_INVALID_PARAM;
1469
1470 #ifndef WITH_NTFS_3G
1471         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1472                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1473                       "        it cannot apply a WIM image directly to an NTFS volume.");
1474                 return WIMLIB_ERR_UNSUPPORTED;
1475         }
1476 #endif
1477
1478         if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1479 #ifdef __WIN32__
1480                 if (!wim->filename)
1481                         return WIMLIB_ERR_NO_FILENAME;
1482 #else
1483                 ERROR("WIMBoot extraction is only supported on Windows!");
1484                 return WIMLIB_ERR_UNSUPPORTED;
1485 #endif
1486         }
1487
1488         if (extract_flags & (WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K |
1489                              WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K |
1490                              WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K |
1491                              WIMLIB_EXTRACT_FLAG_COMPACT_LZX))
1492         {
1493         #ifdef __WIN32__
1494                 int count = 0;
1495                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K) != 0);
1496                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K) != 0);
1497                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K) != 0);
1498                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_LZX) != 0);
1499                 if (count != 1) {
1500                         ERROR("Only one compression format can be specified "
1501                               "for compact-mode extraction!");
1502                         return WIMLIB_ERR_INVALID_PARAM;
1503                 }
1504                 if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1505                         ERROR("Compact-mode extraction and WIMBoot-mode "
1506                               "extraction are mutually exclusive!");
1507                         return WIMLIB_ERR_INVALID_PARAM;
1508                 }
1509         #else
1510                 ERROR("Compact-mode extraction (System Compression) "
1511                       "is only supported on Windows!");
1512                 return WIMLIB_ERR_UNSUPPORTED;
1513         #endif
1514         }
1515
1516
1517         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1518                               WIMLIB_EXTRACT_FLAG_NORPFIX |
1519                               WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1520                                         WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1521         {
1522                 /* For full-image extraction, do reparse point fixups by default
1523                  * if the WIM header says they are enabled.  */
1524                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1525                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1526         }
1527
1528         *extract_flags_p = extract_flags;
1529         return 0;
1530 }
1531
1532 struct append_dentry_ctx {
1533         struct wim_dentry **dentries;
1534         size_t num_dentries;
1535         size_t num_alloc_dentries;
1536 };
1537
1538 static int
1539 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1540 {
1541         struct append_dentry_ctx *ctx = _ctx;
1542
1543         if (ctx->num_dentries == ctx->num_alloc_dentries) {
1544                 struct wim_dentry **new_dentries;
1545                 size_t new_length;
1546
1547                 new_length = max(ctx->num_alloc_dentries + 8,
1548                                  ctx->num_alloc_dentries * 3 / 2);
1549                 new_dentries = REALLOC(ctx->dentries,
1550                                        new_length * sizeof(ctx->dentries[0]));
1551                 if (new_dentries == NULL)
1552                         return WIMLIB_ERR_NOMEM;
1553                 ctx->dentries = new_dentries;
1554                 ctx->num_alloc_dentries = new_length;
1555         }
1556         ctx->dentries[ctx->num_dentries++] = dentry;
1557         return 0;
1558 }
1559
1560 /* Append dentries matched by a path which can contain wildcard characters.  */
1561 static int
1562 append_matched_dentries(WIMStruct *wim, const tchar *orig_pattern,
1563                         int extract_flags, struct append_dentry_ctx *ctx)
1564 {
1565         const size_t count_before = ctx->num_dentries;
1566         tchar *pattern;
1567         int ret;
1568
1569         pattern = canonicalize_wim_path(orig_pattern);
1570         if (!pattern)
1571                 return WIMLIB_ERR_NOMEM;
1572         ret = expand_path_pattern(wim_get_current_root_dentry(wim), pattern,
1573                                   append_dentry_cb, ctx);
1574         FREE(pattern);
1575         if (ret || ctx->num_dentries > count_before)
1576                 return ret;
1577         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB) {
1578                 ERROR("No matches for path pattern \"%"TS"\"", orig_pattern);
1579                 return WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1580         }
1581         WARNING("No matches for path pattern \"%"TS"\"", orig_pattern);
1582         return 0;
1583 }
1584
1585 static int
1586 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1587                         const tchar * const *paths, size_t num_paths,
1588                         int extract_flags)
1589 {
1590         int ret;
1591         struct wim_dentry **trees;
1592         size_t num_trees;
1593
1594         if (wim == NULL || target == NULL || target[0] == T('\0') ||
1595             (num_paths != 0 && paths == NULL))
1596                 return WIMLIB_ERR_INVALID_PARAM;
1597
1598         ret = check_extract_flags(wim, &extract_flags);
1599         if (ret)
1600                 return ret;
1601
1602         ret = select_wim_image(wim, image);
1603         if (ret)
1604                 return ret;
1605
1606         ret = wim_checksum_unhashed_blobs(wim);
1607         if (ret)
1608                 return ret;
1609
1610         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1611                               WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1612             (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1613         {
1614                 ret = mkdir_if_needed(target);
1615                 if (ret)
1616                         return ret;
1617         }
1618
1619         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1620
1621                 struct append_dentry_ctx append_dentry_ctx = {
1622                         .dentries = NULL,
1623                         .num_dentries = 0,
1624                         .num_alloc_dentries = 0,
1625                 };
1626
1627                 for (size_t i = 0; i < num_paths; i++) {
1628                         ret = append_matched_dentries(wim, paths[i],
1629                                                       extract_flags,
1630                                                       &append_dentry_ctx);
1631                         if (ret) {
1632                                 trees = append_dentry_ctx.dentries;
1633                                 goto out_free_trees;
1634                         }
1635                 }
1636                 trees = append_dentry_ctx.dentries;
1637                 num_trees = append_dentry_ctx.num_dentries;
1638         } else {
1639                 trees = MALLOC(num_paths * sizeof(trees[0]));
1640                 if (trees == NULL)
1641                         return WIMLIB_ERR_NOMEM;
1642
1643                 for (size_t i = 0; i < num_paths; i++) {
1644
1645                         tchar *path = canonicalize_wim_path(paths[i]);
1646                         if (path == NULL) {
1647                                 ret = WIMLIB_ERR_NOMEM;
1648                                 goto out_free_trees;
1649                         }
1650
1651                         trees[i] = get_dentry(wim, path,
1652                                               WIMLIB_CASE_PLATFORM_DEFAULT);
1653                         FREE(path);
1654                         if (trees[i] == NULL) {
1655                                   ERROR("Path \"%"TS"\" does not exist "
1656                                         "in WIM image %d",
1657                                         paths[i], wim->current_image);
1658                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1659                                   goto out_free_trees;
1660                         }
1661                 }
1662                 num_trees = num_paths;
1663         }
1664
1665         if (num_trees == 0) {
1666                 ret = 0;
1667                 goto out_free_trees;
1668         }
1669
1670         ret = extract_trees(wim, trees, num_trees, target, extract_flags);
1671 out_free_trees:
1672         FREE(trees);
1673         return ret;
1674 }
1675
1676 static int
1677 extract_single_image(WIMStruct *wim, int image,
1678                      const tchar *target, int extract_flags)
1679 {
1680         const tchar *path = WIMLIB_WIM_ROOT_PATH;
1681         extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1682         return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
1683 }
1684
1685 static const tchar * const filename_forbidden_chars =
1686 T(
1687 #ifdef __WIN32__
1688 "<>:\"/\\|?*"
1689 #else
1690 "/"
1691 #endif
1692 );
1693
1694 /* This function checks if it is okay to use a WIM image's name as a directory
1695  * name.  */
1696 static bool
1697 image_name_ok_as_dir(const tchar *image_name)
1698 {
1699         return image_name && *image_name &&
1700                 !tstrpbrk(image_name, filename_forbidden_chars) &&
1701                 tstrcmp(image_name, T(".")) &&
1702                 tstrcmp(image_name, T("..")) &&
1703                 tstrlen(image_name) <= 128;
1704 }
1705
1706 /* Extracts all images from the WIM to the directory @target, with the images
1707  * placed in subdirectories named by their image names. */
1708 static int
1709 extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
1710 {
1711         size_t output_path_len = tstrlen(target);
1712         tchar buf[output_path_len + 1 + 128 + 1];
1713         int ret;
1714         int image;
1715         const tchar *image_name;
1716
1717         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1718                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1719                 return WIMLIB_ERR_INVALID_PARAM;
1720         }
1721
1722         ret = mkdir_if_needed(target);
1723         if (ret)
1724                 return ret;
1725         tmemcpy(buf, target, output_path_len);
1726         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1727         for (image = 1; image <= wim->hdr.image_count; image++) {
1728                 image_name = wimlib_get_image_name(wim, image);
1729                 if (image_name_ok_as_dir(image_name)) {
1730                         tstrcpy(buf + output_path_len + 1, image_name);
1731                 } else {
1732                         /* Image name is empty or contains forbidden characters.
1733                          * Use image number instead. */
1734                         tsprintf(buf + output_path_len + 1, T("%d"), image);
1735                 }
1736                 ret = extract_single_image(wim, image, buf, extract_flags);
1737                 if (ret)
1738                         return ret;
1739         }
1740         return 0;
1741 }
1742
1743 static int
1744 do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1745                         int extract_flags)
1746 {
1747         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1748                              WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1749                              WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1750                 return WIMLIB_ERR_INVALID_PARAM;
1751
1752         if (image == WIMLIB_ALL_IMAGES)
1753                 return extract_all_images(wim, target, extract_flags);
1754         else
1755                 return extract_single_image(wim, image, target, extract_flags);
1756 }
1757
1758
1759 /****************************************************************************
1760  *                          Extraction API                                  *
1761  ****************************************************************************/
1762
1763 WIMLIBAPI int
1764 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1765                      const tchar * const *paths, size_t num_paths,
1766                      int extract_flags)
1767 {
1768         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1769                 return WIMLIB_ERR_INVALID_PARAM;
1770
1771         return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1772                                        extract_flags);
1773 }
1774
1775 WIMLIBAPI int
1776 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1777                         const tchar *path_list_file, int extract_flags)
1778 {
1779         int ret;
1780         tchar **paths;
1781         size_t num_paths;
1782         void *mem;
1783
1784         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1785         if (ret) {
1786                 ERROR("Failed to read path list file \"%"TS"\"",
1787                       path_list_file);
1788                 return ret;
1789         }
1790
1791         ret = wimlib_extract_paths(wim, image, target,
1792                                    (const tchar * const *)paths, num_paths,
1793                                    extract_flags);
1794         FREE(paths);
1795         FREE(mem);
1796         return ret;
1797 }
1798
1799 WIMLIBAPI int
1800 wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
1801                                              const tchar *image_num_or_name,
1802                                              const tchar *target,
1803                                              int extract_flags,
1804                                              wimlib_progress_func_t progfunc,
1805                                              void *progctx)
1806 {
1807         int ret;
1808         WIMStruct *pwm;
1809         struct filedes *in_fd;
1810         int image;
1811         unsigned i;
1812
1813         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1814                 return WIMLIB_ERR_INVALID_PARAM;
1815
1816         /* Read the WIM header from the pipe and get a WIMStruct to represent
1817          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
1818          * wimlib_open_wim(), getting a WIMStruct in this way will result in an
1819          * empty blob table, no XML data read, and no filename set.  */
1820         ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
1821                                     progfunc, progctx);
1822         if (ret)
1823                 return ret;
1824
1825         /* Sanity check to make sure this is a pipable WIM.  */
1826         if (pwm->hdr.magic != PWM_MAGIC) {
1827                 ERROR("The WIM being read from file descriptor %d "
1828                       "is not pipable!", pipe_fd);
1829                 ret = WIMLIB_ERR_NOT_PIPABLE;
1830                 goto out_wimlib_free;
1831         }
1832
1833         /* Sanity check to make sure the first part of a pipable split WIM is
1834          * sent over the pipe first.  */
1835         if (pwm->hdr.part_number != 1) {
1836                 ERROR("The first part of the split WIM must be "
1837                       "sent over the pipe first.");
1838                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1839                 goto out_wimlib_free;
1840         }
1841
1842         in_fd = &pwm->in_fd;
1843         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1844
1845         /* As mentioned, the WIMStruct we created from the pipe does not have
1846          * XML data yet.  Fix this by reading the extra copy of the XML data
1847          * that directly follows the header in pipable WIMs.  (Note: see
1848          * write_pipable_wim() for more details about the format of pipable
1849          * WIMs.)  */
1850         {
1851                 u8 hash[SHA1_HASH_SIZE];
1852
1853                 ret = read_pwm_blob_header(pwm, hash,
1854                                            &pwm->hdr.xml_data_reshdr, NULL);
1855                 if (ret)
1856                         goto out_wimlib_free;
1857
1858                 if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1859                         ERROR("Expected XML data, but found non-metadata resource.");
1860                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1861                         goto out_wimlib_free;
1862                 }
1863
1864                 ret = read_wim_xml_data(pwm);
1865                 if (ret)
1866                         goto out_wimlib_free;
1867
1868                 if (xml_get_image_count(pwm->xml_info) != pwm->hdr.image_count) {
1869                         ERROR("Image count in XML data is not the same as in WIM header.");
1870                         ret = WIMLIB_ERR_IMAGE_COUNT;
1871                         goto out_wimlib_free;
1872                 }
1873         }
1874
1875         /* Get image index (this may use the XML data that was just read to
1876          * resolve an image name).  */
1877         if (image_num_or_name) {
1878                 image = wimlib_resolve_image(pwm, image_num_or_name);
1879                 if (image == WIMLIB_NO_IMAGE) {
1880                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1881                               image_num_or_name);
1882                         ret = WIMLIB_ERR_INVALID_IMAGE;
1883                         goto out_wimlib_free;
1884                 } else if (image == WIMLIB_ALL_IMAGES) {
1885                         ERROR("Applying all images from a pipe is not supported!");
1886                         ret = WIMLIB_ERR_INVALID_IMAGE;
1887                         goto out_wimlib_free;
1888                 }
1889         } else {
1890                 if (pwm->hdr.image_count != 1) {
1891                         ERROR("No image was specified, but the pipable WIM "
1892                               "did not contain exactly 1 image");
1893                         ret = WIMLIB_ERR_INVALID_IMAGE;
1894                         goto out_wimlib_free;
1895                 }
1896                 image = 1;
1897         }
1898
1899         /* Load the needed metadata resource.  */
1900         for (i = 1; i <= pwm->hdr.image_count; i++) {
1901                 struct wim_image_metadata *imd;
1902                 struct wim_reshdr reshdr;
1903                 struct wim_resource_descriptor *metadata_rdesc;
1904
1905                 imd = pwm->image_metadata[i - 1];
1906
1907                 ret = WIMLIB_ERR_NOMEM;
1908                 imd->metadata_blob = new_blob_descriptor();
1909                 if (!imd->metadata_blob)
1910                         goto out_wimlib_free;
1911
1912                 imd->metadata_blob->is_metadata = 1;
1913
1914                 ret = read_pwm_blob_header(pwm, imd->metadata_blob->hash,
1915                                            &reshdr, NULL);
1916                 if (ret)
1917                         goto out_wimlib_free;
1918
1919                 if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1920                         ERROR("Expected metadata resource, but found "
1921                               "non-metadata resource");
1922                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1923                         goto out_wimlib_free;
1924                 }
1925
1926                 ret = WIMLIB_ERR_NOMEM;
1927                 metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
1928                 if (!metadata_rdesc)
1929                         goto out_wimlib_free;
1930                 wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc,
1931                                             imd->metadata_blob);
1932                 pwm->refcnt++;
1933
1934                 if (i == image) {
1935                         /* Metadata resource is for the image being extracted.
1936                          * Parse it and save the metadata in memory.  */
1937                         ret = read_metadata_resource(imd);
1938                         if (ret)
1939                                 goto out_wimlib_free;
1940                 } else {
1941                         /* Metadata resource is not for the image being
1942                          * extracted.  Skip over it.  */
1943                         ret = skip_wim_resource(metadata_rdesc);
1944                         if (ret)
1945                                 goto out_wimlib_free;
1946                 }
1947         }
1948         /* Extract the image.  */
1949         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
1950         ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
1951         /* Clean up and return.  */
1952 out_wimlib_free:
1953         wimlib_free(pwm);
1954         return ret;
1955 }
1956
1957
1958 WIMLIBAPI int
1959 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
1960                                const tchar *target, int extract_flags)
1961 {
1962         return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
1963                                                             image_num_or_name,
1964                                                             target,
1965                                                             extract_flags,
1966                                                             NULL,
1967                                                             NULL);
1968 }
1969
1970 WIMLIBAPI int
1971 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1972                      int extract_flags)
1973 {
1974         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1975                 return WIMLIB_ERR_INVALID_PARAM;
1976         return do_wimlib_extract_image(wim, image, target, extract_flags);
1977 }