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) 2012-2016 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 static int
 263 handle_pwm_metadata_resource(WIMStruct *pwm, int image, bool is_needed)
 264 {
 265         struct blob_descriptor *blob;
 266         struct wim_reshdr reshdr;
 267         struct wim_resource_descriptor *rdesc;
 268         int ret;
 269
 270         ret = WIMLIB_ERR_NOMEM;
 271         blob = new_blob_descriptor();
 272         if (!blob)
 273                 goto out;
 274
 275         ret = read_pwm_blob_header(pwm, blob->hash, &reshdr, NULL);
 276         if (ret)
 277                 goto out;
 278
 279         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
 280         if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
 281                 ERROR("Expected metadata resource, but found non-metadata "
 282                       "resource");
 283                 goto out;
 284         }
 285
 286         ret = WIMLIB_ERR_NOMEM;
 287         rdesc = MALLOC(sizeof(*rdesc));
 288         if (!rdesc)
 289                 goto out;
 290
 291         wim_reshdr_to_desc_and_blob(&reshdr, pwm, rdesc, blob);
 292         pwm->refcnt++;
 293
 294         ret = WIMLIB_ERR_NOMEM;
 295         pwm->image_metadata[image - 1] = new_unloaded_image_metadata(blob);
 296         if (!pwm->image_metadata[image - 1])
 297                 goto out;
 298         blob = NULL;
 299
 300         /* If the metadata resource is for the image being extracted, then parse
 301          * it and save the metadata in memory.  Otherwise, skip over it.  */
 302         if (is_needed)
 303                 ret = select_wim_image(pwm, image);
 304         else
 305                 ret = skip_wim_resource(rdesc);
 306 out:
 307         free_blob_descriptor(blob);
 308         return ret;
 309 }
 310
 311 /* Creates a temporary file opened for writing.  The open file descriptor is
 312  * returned in @fd_ret and its name is returned in @name_ret (dynamically
 313  * allocated).  */
 314 static int
 315 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
 316 {
 317         tchar *name;
 318         int open_flags;
 319         int raw_fd;
 320
 321 retry:
 322         name = ttempnam(NULL, T("wimlib"));
 323         if (!name) {
 324                 ERROR_WITH_ERRNO("Failed to create temporary filename");
 325                 return WIMLIB_ERR_NOMEM;
 326         }
 327
 328         open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
 329 #ifdef __WIN32__
 330         open_flags |= _O_SHORT_LIVED;
 331 #endif
 332         raw_fd = topen(name, open_flags, 0600);
 333
 334         if (raw_fd < 0) {
 335                 if (errno == EEXIST) {
 336                         FREE(name);
 337                         goto retry;
 338                 }
 339                 ERROR_WITH_ERRNO("Failed to create temporary file "
 340                                  "\"%"TS"\"", name);
 341                 FREE(name);
 342                 return WIMLIB_ERR_OPEN;
 343         }
 344
 345         filedes_init(fd_ret, raw_fd);
 346         *name_ret = name;
 347         return 0;
 348 }
 349
 350 static int
 351 begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
 352 {
 353         struct apply_ctx *ctx = _ctx;
 354
 355         ctx->cur_blob = blob;
 356         ctx->cur_blob_offset = 0;
 357
 358         if (unlikely(blob->out_refcnt > MAX_OPEN_FILES))
 359                 return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
 360
 361         return call_begin_blob(blob, ctx->saved_cbs);
 362 }
 363
 364 static int
 365 extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
 366 {
 367         struct apply_ctx *ctx = _ctx;
 368         union wimlib_progress_info *progress = &ctx->progress;
 369         int ret;
 370
 371         ctx->cur_blob_offset += size;
 372
 373         if (likely(ctx->supported_features.hard_links)) {
 374                 progress->extract.completed_bytes +=
 375                         (u64)size * ctx->cur_blob->out_refcnt;
 376                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 377                         progress->extract.completed_streams += ctx->cur_blob->out_refcnt;
 378         } else {
 379                 const struct blob_extraction_target *targets =
 380                         blob_extraction_targets(ctx->cur_blob);
 381                 for (u32 i = 0; i < ctx->cur_blob->out_refcnt; i++) {
 382                         const struct wim_inode *inode = targets[i].inode;
 383                         const struct wim_dentry *dentry;
 384
 385                         inode_for_each_extraction_alias(dentry, inode) {
 386                                 progress->extract.completed_bytes += size;
 387                                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 388                                         progress->extract.completed_streams++;
 389                         }
 390                 }
 391         }
 392         if (progress->extract.completed_bytes >= ctx->next_progress) {
 393
 394                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
 395                 if (ret)
 396                         return ret;
 397
 398                 set_next_progress(progress->extract.completed_bytes,
 399                                   progress->extract.total_bytes,
 400                                   &ctx->next_progress);
 401         }
 402
 403         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 404                 /* Just extracting to temporary file for now.  */
 405                 ret = full_write(&ctx->tmpfile_fd, chunk, size);
 406                 if (ret) {
 407                         ERROR_WITH_ERRNO("Error writing data to "
 408                                          "temporary file \"%"TS"\"",
 409                                          ctx->tmpfile_name);
 410                 }
 411                 return ret;
 412         }
 413
 414         return call_consume_chunk(chunk, size, ctx->saved_cbs);
 415 }
 416
 417 /* Copy the blob's data from the temporary file to each of its targets.
 418  *
 419  * This is executed only in the very uncommon case that a blob is being
 420  * extracted to more than MAX_OPEN_FILES targets!  */
 421 static int
 422 extract_from_tmpfile(const tchar *tmpfile_name,
 423                      const struct blob_descriptor *orig_blob,
 424                      const struct read_blob_callbacks *cbs)
 425 {
 426         struct blob_descriptor tmpfile_blob;
 427         const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob);
 428         int ret;
 429
 430         memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor));
 431         tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK;
 432         tmpfile_blob.file_on_disk = (tchar *)tmpfile_name;
 433         tmpfile_blob.out_refcnt = 1;
 434
 435         for (u32 i = 0; i < orig_blob->out_refcnt; i++) {
 436                 tmpfile_blob.inline_blob_extraction_targets[0] = targets[i];
 437                 ret = read_blob_with_cbs(&tmpfile_blob, cbs);
 438                 if (ret)
 439                         return ret;
 440         }
 441         return 0;
 442 }
 443
 444 static int
 445 end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
 446 {
 447         struct apply_ctx *ctx = _ctx;
 448
 449         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 450                 filedes_close(&ctx->tmpfile_fd);
 451                 if (!status)
 452                         status = extract_from_tmpfile(ctx->tmpfile_name, blob,
 453                                                       ctx->saved_cbs);
 454                 filedes_invalidate(&ctx->tmpfile_fd);
 455                 tunlink(ctx->tmpfile_name);
 456                 FREE(ctx->tmpfile_name);
 457                 return status;
 458         }
 459
 460         return call_end_blob(blob, status, ctx->saved_cbs);
 461 }
 462
 463 /*
 464  * Read the list of blobs to extract and feed their data into the specified
 465  * callback functions.
 466  *
 467  * This handles checksumming each blob.
 468  *
 469  * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
 470  *
 471  * This also works if the WIM is being read from a pipe.
 472  *
 473  * This also will split up blobs that will need to be extracted to more than
 474  * MAX_OPEN_FILES locations, as measured by the 'out_refcnt' of each blob.
 475  * Therefore, the apply_operations implementation need not worry about running
 476  * out of file descriptors, unless it might open more than one file descriptor
 477  * per 'blob_extraction_target' (e.g. Win32 currently might because the
 478  * destination file system might not support hard links).
 479  */
 480 int
 481 extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 482 {
 483         struct read_blob_callbacks wrapper_cbs = {
 484                 .begin_blob     = begin_extract_blob_wrapper,
 485                 .consume_chunk  = extract_chunk_wrapper,
 486                 .end_blob       = end_extract_blob_wrapper,
 487                 .ctx            = ctx,
 488         };
 489         ctx->saved_cbs = cbs;
 490         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
 491                 return read_blobs_from_pipe(ctx, &wrapper_cbs);
 492         } else {
 493                 return read_blob_list(&ctx->blob_list,
 494                                       offsetof(struct blob_descriptor,
 495                                                extraction_list),
 496                                       &wrapper_cbs, VERIFY_BLOB_HASHES);
 497         }
 498 }
 499
 500 /* Extract a WIM dentry to standard output.
 501  *
 502  * This obviously doesn't make sense in all cases.  We return an error if the
 503  * dentry does not correspond to a regular file.  Otherwise we extract the
 504  * unnamed data stream only.  */
 505 static int
 506 extract_dentry_to_stdout(struct wim_dentry *dentry,
 507                          const struct blob_table *blob_table)
 508 {
 509         struct wim_inode *inode = dentry->d_inode;
 510         struct blob_descriptor *blob;
 511         struct filedes _stdout;
 512
 513         if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
 514                                    FILE_ATTRIBUTE_DIRECTORY |
 515                                    FILE_ATTRIBUTE_ENCRYPTED))
 516         {
 517                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 518                       "extracted to standard output", dentry_full_path(dentry));
 519                 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
 520         }
 521
 522         blob = inode_get_blob_for_unnamed_data_stream(inode, blob_table);
 523         if (!blob) {
 524                 const u8 *hash = inode_get_hash_of_unnamed_data_stream(inode);
 525                 if (!is_zero_hash(hash))
 526                         return blob_not_found_error(inode, hash);
 527                 return 0;
 528         }
 529
 530         filedes_init(&_stdout, STDOUT_FILENO);
 531         return extract_blob_to_fd(blob, &_stdout);
 532 }
 533
 534 static int
 535 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
 536                            const struct blob_table *blob_table)
 537 {
 538         for (size_t i = 0; i < num_dentries; i++) {
 539                 int ret = extract_dentry_to_stdout(dentries[i], blob_table);
 540                 if (ret)
 541                         return ret;
 542         }
 543         return 0;
 544 }
 545
 546 /**********************************************************************/
 547
 548 /*
 549  * Removes duplicate dentries from the array.
 550  *
 551  * Returns the new number of dentries, packed at the front of the array.
 552  */
 553 static size_t
 554 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
 555 {
 556         size_t i, j = 0;
 557         for (i = 0; i < num_trees; i++) {
 558                 if (!trees[i]->d_tmp_flag) {
 559                         /* Found distinct dentry.  */
 560                         trees[i]->d_tmp_flag = 1;
 561                         trees[j++] = trees[i];
 562                 }
 563         }
 564         for (i = 0; i < j; i++)
 565                 trees[i]->d_tmp_flag = 0;
 566         return j;
 567 }
 568
 569 /*
 570  * Remove dentries that are descendants of other dentries in the array.
 571  *
 572  * Returns the new number of dentries, packed at the front of the array.
 573  */
 574 static size_t
 575 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
 576 {
 577         size_t i, j = 0;
 578         for (i = 0; i < num_trees; i++)
 579                 trees[i]->d_tmp_flag = 1;
 580         for (i = 0; i < num_trees; i++) {
 581                 struct wim_dentry *d = trees[i];
 582                 while (!dentry_is_root(d)) {
 583                         d = d->d_parent;
 584                         if (d->d_tmp_flag)
 585                                 goto tree_contained;
 586                 }
 587                 trees[j++] = trees[i];
 588                 continue;
 589
 590         tree_contained:
 591                 trees[i]->d_tmp_flag = 0;
 592         }
 593
 594         for (i = 0; i < j; i++)
 595                 trees[i]->d_tmp_flag = 0;
 596         return j;
 597 }
 598
 599 static int
 600 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
 601 {
 602         struct list_head *dentry_list = _dentry_list;
 603         list_add_tail(&dentry->d_extraction_list_node, dentry_list);
 604         return 0;
 605 }
 606
 607 static void
 608 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
 609 {
 610         dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
 611 }
 612
 613 static int
 614 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
 615 {
 616         if (will_extract_dentry(dentry)) {
 617                 list_del(&dentry->d_extraction_list_node);
 618                 dentry_reset_extraction_list_node(dentry);
 619         }
 620         return 0;
 621 }
 622
 623 /*
 624  * Build the preliminary list of dentries to be extracted.
 625  *
 626  * The list maintains the invariant that if d1 and d2 are in the list and d1 is
 627  * an ancestor of d2, then d1 appears before d2 in the list.
 628  */
 629 static void
 630 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
 631                   size_t num_trees, bool add_ancestors)
 632 {
 633         INIT_LIST_HEAD(dentry_list);
 634
 635         /* Add the trees recursively.  */
 636         for (size_t i = 0; i < num_trees; i++)
 637                 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
 638
 639         /* If requested, add ancestors of the trees.  */
 640         if (add_ancestors) {
 641                 for (size_t i = 0; i < num_trees; i++) {
 642                         struct wim_dentry *dentry = trees[i];
 643                         struct wim_dentry *ancestor;
 644                         struct list_head *place_after;
 645
 646                         if (dentry_is_root(dentry))
 647                                 continue;
 648
 649                         place_after = dentry_list;
 650                         ancestor = dentry;
 651                         do {
 652                                 ancestor = ancestor->d_parent;
 653                                 if (will_extract_dentry(ancestor)) {
 654                                         place_after = &ancestor->d_extraction_list_node;
 655                                         break;
 656                                 }
 657                         } while (!dentry_is_root(ancestor));
 658
 659                         ancestor = dentry;
 660                         do {
 661                                 ancestor = ancestor->d_parent;
 662                                 if (will_extract_dentry(ancestor))
 663                                         break;
 664                                 list_add(&ancestor->d_extraction_list_node, place_after);
 665                         } while (!dentry_is_root(ancestor));
 666                 }
 667         }
 668 }
 669
 670 static void
 671 destroy_dentry_list(struct list_head *dentry_list)
 672 {
 673         struct wim_dentry *dentry, *tmp;
 674         struct wim_inode *inode;
 675
 676         list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
 677                 inode = dentry->d_inode;
 678                 dentry_reset_extraction_list_node(dentry);
 679                 inode->i_visited = 0;
 680                 inode->i_can_externally_back = 0;
 681                 if ((void *)dentry->d_extraction_name != (void *)dentry->d_name)
 682                         FREE(dentry->d_extraction_name);
 683                 dentry->d_extraction_name = NULL;
 684                 dentry->d_extraction_name_nchars = 0;
 685         }
 686 }
 687
 688 static void
 689 destroy_blob_list(struct list_head *blob_list)
 690 {
 691         struct blob_descriptor *blob;
 692
 693         list_for_each_entry(blob, blob_list, extraction_list)
 694                 if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets))
 695                         FREE(blob->blob_extraction_targets);
 696 }
 697
 698 #ifdef __WIN32__
 699 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
 700 #else
 701 static const utf16lechar replacement_char = cpu_to_le16('?');
 702 #endif
 703
 704 static bool
 705 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
 706 {
 707         size_t i;
 708
 709         if (num_chars == 0)
 710                 return true;
 711         for (i = 0; i < num_chars; i++) {
 712                 switch (le16_to_cpu(name[i])) {
 713         #ifdef __WIN32__
 714                 case '\x01'...'\x1F':
 715                 case '\\':
 716                 case ':':
 717                 case '*':
 718                 case '?':
 719                 case '"':
 720                 case '<':
 721                 case '>':
 722                 case '|':
 723         #endif
 724                 case '/':
 725                 case '\0':
 726                         if (fix)
 727                                 name[i] = replacement_char;
 728                         else
 729                                 return false;
 730                 }
 731         }
 732
 733         return true;
 734 }
 735
 736 static int
 737 dentry_calculate_extraction_name(struct wim_dentry *dentry,
 738                                  struct apply_ctx *ctx)
 739 {
 740         int ret;
 741
 742         if (dentry_is_root(dentry))
 743                 return 0;
 744
 745 #ifdef WITH_NTFS_3G
 746         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 747                 dentry->d_extraction_name = dentry->d_name;
 748                 dentry->d_extraction_name_nchars = dentry->d_name_nbytes /
 749                                                    sizeof(utf16lechar);
 750                 return 0;
 751         }
 752 #endif
 753
 754         if (!ctx->supported_features.case_sensitive_filenames) {
 755                 struct wim_dentry *other;
 756                 list_for_each_entry(other, &dentry->d_ci_conflict_list,
 757                                     d_ci_conflict_list)
 758                 {
 759                         if (will_extract_dentry(other)) {
 760                                 if (ctx->extract_flags &
 761                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
 762                                         WARNING("\"%"TS"\" has the same "
 763                                                 "case-insensitive name as "
 764                                                 "\"%"TS"\"; extracting "
 765                                                 "dummy name instead",
 766                                                 dentry_full_path(dentry),
 767                                                 dentry_full_path(other));
 768                                         goto out_replace;
 769                                 } else {
 770                                         WARNING("Not extracting \"%"TS"\": "
 771                                                 "has same case-insensitive "
 772                                                 "name as \"%"TS"\"",
 773                                                 dentry_full_path(dentry),
 774                                                 dentry_full_path(other));
 775                                         goto skip_dentry;
 776                                 }
 777                         }
 778                 }
 779         }
 780
 781         if (file_name_valid(dentry->d_name, dentry->d_name_nbytes / 2, false)) {
 782                 size_t nbytes = 0;
 783                 ret = utf16le_get_tstr(dentry->d_name,
 784                                        dentry->d_name_nbytes,
 785                                        (const tchar **)&dentry->d_extraction_name,
 786                                        &nbytes);
 787                 dentry->d_extraction_name_nchars = nbytes / sizeof(tchar);
 788                 return ret;
 789         } else {
 790                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
 791                 {
 792                         WARNING("\"%"TS"\" has an invalid filename "
 793                                 "that is not supported on this platform; "
 794                                 "extracting dummy name instead",
 795                                 dentry_full_path(dentry));
 796                         goto out_replace;
 797                 } else {
 798                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
 799                                 "that is not supported on this platform",
 800                                 dentry_full_path(dentry));
 801                         goto skip_dentry;
 802                 }
 803         }
 804
 805 out_replace:
 806         {
 807                 utf16lechar utf16_name_copy[dentry->d_name_nbytes / 2];
 808
 809                 memcpy(utf16_name_copy, dentry->d_name, dentry->d_name_nbytes);
 810                 file_name_valid(utf16_name_copy, dentry->d_name_nbytes / 2, true);
 811
 812                 const tchar *tchar_name;
 813                 size_t tchar_nchars;
 814
 815                 ret = utf16le_get_tstr(utf16_name_copy,
 816                                        dentry->d_name_nbytes,
 817                                        &tchar_name, &tchar_nchars);
 818                 if (ret)
 819                         return ret;
 820
 821                 tchar_nchars /= sizeof(tchar);
 822
 823                 size_t fixed_name_num_chars = tchar_nchars;
 824                 tchar fixed_name[tchar_nchars + 50];
 825
 826                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
 827                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
 828                                                  T(" (invalid filename #%lu)"),
 829                                                  ++ctx->invalid_sequence);
 830
 831                 utf16le_put_tstr(tchar_name);
 832
 833                 dentry->d_extraction_name = TSTRDUP(fixed_name);
 834                 if (!dentry->d_extraction_name)
 835                         return WIMLIB_ERR_NOMEM;
 836                 dentry->d_extraction_name_nchars = fixed_name_num_chars;
 837         }
 838         return 0;
 839
 840 skip_dentry:
 841         for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
 842         return 0;
 843 }
 844
 845 /*
 846  * Calculate the actual filename component at which each WIM dentry will be
 847  * extracted, with special handling for dentries that are unsupported by the
 848  * extraction backend or have invalid names.
 849  *
 850  * ctx->supported_features must be filled in.
 851  *
 852  * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
 853  */
 854 static int
 855 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
 856                                        struct apply_ctx *ctx)
 857 {
 858         struct list_head *prev, *cur;
 859
 860         /* Can't use list_for_each_entry() because a call to
 861          * dentry_calculate_extraction_name() may delete the current dentry and
 862          * its children from the list.  */
 863
 864         prev = dentry_list;
 865         for (;;) {
 866                 struct wim_dentry *dentry;
 867                 int ret;
 868
 869                 cur = prev->next;
 870                 if (cur == dentry_list)
 871                         break;
 872
 873                 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
 874
 875                 ret = dentry_calculate_extraction_name(dentry, ctx);
 876                 if (ret)
 877                         return ret;
 878
 879                 if (prev->next == cur)
 880                         prev = cur;
 881                 else
 882                         ; /* Current dentry and its children (which follow in
 883                              the list) were deleted.  prev stays the same.  */
 884         }
 885         return 0;
 886 }
 887
 888 static int
 889 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
 890                        struct blob_table *blob_table)
 891 {
 892         struct wim_inode *inode = dentry->d_inode;
 893         struct blob_descriptor *blob;
 894         int ret;
 895         bool force = false;
 896
 897         /* Special case:  when extracting from a pipe, the WIM blob table is
 898          * initially empty, so "resolving" an inode's streams is initially not
 899          * possible.  However, we still need to keep track of which blobs,
 900          * identified by SHA-1 message digests, need to be extracted, so we
 901          * "resolve" the inode's streams anyway by allocating a 'struct
 902          * blob_descriptor' for each one.  */
 903         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
 904                 force = true;
 905         ret = inode_resolve_streams(inode, blob_table, force);
 906         if (ret)
 907                 return ret;
 908         for (unsigned i = 0; i < inode->i_num_streams; i++) {
 909                 blob = stream_blob_resolved(&inode->i_streams[i]);
 910                 if (blob)
 911                         blob->out_refcnt = 0;
 912         }
 913         return 0;
 914 }
 915
 916 /*
 917  * For each dentry to be extracted, resolve all streams in the corresponding
 918  * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
 919  *
 920  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
 921  */
 922 static int
 923 dentry_list_resolve_streams(struct list_head *dentry_list,
 924                             struct apply_ctx *ctx)
 925 {
 926         struct wim_dentry *dentry;
 927         int ret;
 928
 929         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
 930                 ret = dentry_resolve_streams(dentry,
 931                                              ctx->extract_flags,
 932                                              ctx->wim->blob_table);
 933                 if (ret)
 934                         return ret;
 935         }
 936         return 0;
 937 }
 938
 939 static int
 940 ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
 941            struct apply_ctx *ctx)
 942 {
 943         struct wim_inode *inode = dentry->d_inode;
 944         struct blob_descriptor *blob = stream_blob_resolved(strm);
 945         struct blob_extraction_target *targets;
 946
 947         if (!blob)
 948                 return 0;
 949
 950         /* Tally the size only for each actual extraction of the stream (not
 951          * additional hard links to the inode).  */
 952         if (inode->i_visited && ctx->supported_features.hard_links)
 953                 return 0;
 954
 955         ctx->progress.extract.total_bytes += blob->size;
 956         ctx->progress.extract.total_streams++;
 957
 958         if (inode->i_visited)
 959                 return 0;
 960
 961         /* Add each blob to 'ctx->blob_list' only one time, regardless of how
 962          * many extraction targets it will have.  */
 963         if (blob->out_refcnt == 0) {
 964                 list_add_tail(&blob->extraction_list, &ctx->blob_list);
 965                 ctx->num_blobs_remaining++;
 966         }
 967
 968         /* Set this stream as an extraction target of 'blob'.  */
 969
 970         if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 971                 targets = blob->inline_blob_extraction_targets;
 972         } else {
 973                 struct blob_extraction_target *prev_targets;
 974                 size_t alloc_blob_extraction_targets;
 975
 976                 if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 977                         prev_targets = NULL;
 978                         alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets);
 979                 } else {
 980                         prev_targets = blob->blob_extraction_targets;
 981                         alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
 982                 }
 983
 984                 if (blob->out_refcnt == alloc_blob_extraction_targets) {
 985                         alloc_blob_extraction_targets *= 2;
 986                         targets = REALLOC(prev_targets,
 987                                           alloc_blob_extraction_targets *
 988                                           sizeof(targets[0]));
 989                         if (!targets)
 990                                 return WIMLIB_ERR_NOMEM;
 991                         if (!prev_targets) {
 992                                 memcpy(targets,
 993                                        blob->inline_blob_extraction_targets,
 994                                        sizeof(blob->inline_blob_extraction_targets));
 995                         }
 996                         blob->blob_extraction_targets = targets;
 997                         blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
 998                 }
 999                 targets = blob->blob_extraction_targets;
1000         }
1001         targets[blob->out_refcnt].inode = inode;
1002         targets[blob->out_refcnt].stream = strm;
1003         blob->out_refcnt++;
1004         return 0;
1005 }
1006
1007 static int
1008 ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode,
1009                      struct wim_inode_stream *strm, struct apply_ctx *ctx)
1010 {
1011         bool need_stream = false;
1012         switch (strm->stream_type) {
1013         case STREAM_TYPE_DATA:
1014                 if (stream_is_named(strm)) {
1015                         /* Named data stream  */
1016                         if (ctx->supported_features.named_data_streams)
1017                                 need_stream = true;
1018                 } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
1019                                                     FILE_ATTRIBUTE_ENCRYPTED))
1020                            && !(inode_is_symlink(inode)
1021                                 && !ctx->supported_features.reparse_points
1022                                 && ctx->supported_features.symlink_reparse_points))
1023                 {
1024                         /*
1025                          * Unnamed data stream.  Skip if any of the following is true:
1026                          *
1027                          * - file is a directory
1028                          * - file is encrypted
1029                          * - backend needs to create the file as UNIX symlink
1030                          * - backend will extract the stream as externally
1031                          *   backed from the WIM archive itself
1032                          */
1033                         if (ctx->apply_ops->will_back_from_wim) {
1034                                 int ret = (*ctx->apply_ops->will_back_from_wim)(dentry, ctx);
1035                                 if (ret > 0) /* Error?  */
1036                                         return ret;
1037                                 if (ret < 0) /* Won't externally back?  */
1038                                         need_stream = true;
1039                         } else {
1040                                 need_stream = true;
1041                         }
1042                 }
1043                 break;
1044         case STREAM_TYPE_REPARSE_POINT:
1045                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT);
1046                 if (ctx->supported_features.reparse_points ||
1047                     (inode_is_symlink(inode) &&
1048                      ctx->supported_features.symlink_reparse_points))
1049                         need_stream = true;
1050                 break;
1051         case STREAM_TYPE_EFSRPC_RAW_DATA:
1052                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED);
1053                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
1054                         if (ctx->supported_features.encrypted_directories)
1055                                 need_stream = true;
1056                 } else {
1057                         if (ctx->supported_features.encrypted_files)
1058                                 need_stream = true;
1059                 }
1060                 break;
1061         }
1062         if (need_stream)
1063                 return ref_stream(strm, dentry, ctx);
1064         return 0;
1065 }
1066
1067 static int
1068 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
1069 {
1070         struct wim_inode *inode = dentry->d_inode;
1071         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1072                 int ret = ref_stream_if_needed(dentry, inode,
1073                                                &inode->i_streams[i], ctx);
1074                 if (ret)
1075                         return ret;
1076         }
1077         inode->i_visited = 1;
1078         return 0;
1079 }
1080
1081 /*
1082  * Given a list of dentries to be extracted, build the list of blobs that need
1083  * to be extracted, and for each blob determine the streams to which that blob
1084  * will be extracted.
1085  *
1086  * This also initializes the extract progress info with byte and blob
1087  * information.
1088  *
1089  * ctx->supported_features must be filled in.
1090  */
1091 static int
1092 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
1093 {
1094         struct wim_dentry *dentry;
1095         int ret;
1096
1097         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1098                 ret = dentry_ref_streams(dentry, ctx);
1099                 if (ret)
1100                         return ret;
1101         }
1102         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1103                 dentry->d_inode->i_visited = 0;
1104         return 0;
1105 }
1106
1107 static void
1108 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
1109 {
1110         struct wim_dentry *dentry;
1111
1112         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1113                 dentry->d_inode->i_first_extraction_alias = NULL;
1114
1115         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1116                 dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias;
1117                 dentry->d_inode->i_first_extraction_alias = dentry;
1118         }
1119 }
1120
1121 static void
1122 inode_tally_features(const struct wim_inode *inode,
1123                      struct wim_features *features)
1124 {
1125         if (inode->i_attributes & FILE_ATTRIBUTE_READONLY)
1126                 features->readonly_files++;
1127         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1128                 features->hidden_files++;
1129         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1130                 features->system_files++;
1131         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1132                 features->archive_files++;
1133         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1134                 features->compressed_files++;
1135         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1136                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1137                         features->encrypted_directories++;
1138                 else
1139                         features->encrypted_files++;
1140         }
1141         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1142                 features->not_context_indexed_files++;
1143         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1144                 features->sparse_files++;
1145         if (inode_has_named_data_stream(inode))
1146                 features->named_data_streams++;
1147         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1148                 features->reparse_points++;
1149                 if (inode_is_symlink(inode))
1150                         features->symlink_reparse_points++;
1151                 else
1152                         features->other_reparse_points++;
1153         }
1154         if (inode_has_security_descriptor(inode))
1155                 features->security_descriptors++;
1156         if (inode_has_unix_data(inode))
1157                 features->unix_data++;
1158 }
1159
1160 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1161 static void
1162 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
1163 {
1164         struct wim_inode *inode = dentry->d_inode;
1165
1166         if (dentry_has_short_name(dentry))
1167                 features->short_names++;
1168
1169         if (inode->i_visited) {
1170                 features->hard_links++;
1171         } else {
1172                 inode_tally_features(inode, features);
1173                 inode->i_visited = 1;
1174         }
1175 }
1176
1177 /* Tally the features necessary to extract the specified dentries.  */
1178 static void
1179 dentry_list_get_features(struct list_head *dentry_list,
1180                          struct wim_features *features)
1181 {
1182         struct wim_dentry *dentry;
1183
1184         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1185                 dentry_tally_features(dentry, features);
1186
1187         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1188                 dentry->d_inode->i_visited = 0;
1189 }
1190
1191 static int
1192 do_feature_check(const struct wim_features *required_features,
1193                  const struct wim_features *supported_features,
1194                  int extract_flags)
1195 {
1196         /* Encrypted files.  */
1197         if (required_features->encrypted_files &&
1198             !supported_features->encrypted_files)
1199                 WARNING("Ignoring EFS-encrypted data of %lu files",
1200                         required_features->encrypted_files);
1201
1202         /* Named data streams.  */
1203         if (required_features->named_data_streams &&
1204             !supported_features->named_data_streams)
1205                 WARNING("Ignoring named data streams of %lu files",
1206                         required_features->named_data_streams);
1207
1208         /* File attributes.  */
1209         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1210
1211                 if (required_features->readonly_files &&
1212                     !supported_features->readonly_files)
1213                         WARNING("Ignoring FILE_ATTRIBUTE_READONLY of %lu files",
1214                                 required_features->readonly_files);
1215
1216                 if (required_features->hidden_files &&
1217                     !supported_features->hidden_files)
1218                         WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1219                                 required_features->hidden_files);
1220
1221                 if (required_features->system_files &&
1222                     !supported_features->system_files)
1223                         WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1224                                 required_features->system_files);
1225
1226                 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1227                  * We're an archive program, so theoretically we can do what we
1228                  * want with it.  */
1229
1230                 if (required_features->compressed_files &&
1231                     !supported_features->compressed_files)
1232                         WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1233                                 required_features->compressed_files);
1234
1235                 if (required_features->not_context_indexed_files &&
1236                     !supported_features->not_context_indexed_files)
1237                         WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1238                                 required_features->not_context_indexed_files);
1239
1240                 if (required_features->sparse_files &&
1241                     !supported_features->sparse_files)
1242                         WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1243                                 required_features->sparse_files);
1244
1245                 if (required_features->encrypted_directories &&
1246                     !supported_features->encrypted_directories)
1247                         WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1248                                 required_features->encrypted_directories);
1249         }
1250
1251         /* Hard links.  */
1252         if (required_features->hard_links && !supported_features->hard_links)
1253                 WARNING("Extracting %lu hard links as independent files",
1254                         required_features->hard_links);
1255
1256         /* Symbolic links and reparse points.  */
1257         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1258             required_features->symlink_reparse_points &&
1259             !supported_features->symlink_reparse_points &&
1260             !supported_features->reparse_points)
1261         {
1262                 ERROR("Extraction backend does not support symbolic links!");
1263                 return WIMLIB_ERR_UNSUPPORTED;
1264         }
1265         if (required_features->reparse_points &&
1266             !supported_features->reparse_points)
1267         {
1268                 if (supported_features->symlink_reparse_points) {
1269                         if (required_features->other_reparse_points) {
1270                                 WARNING("Ignoring reparse data of %lu non-symlink/junction files",
1271                                         required_features->other_reparse_points);
1272                         }
1273                 } else {
1274                         WARNING("Ignoring reparse data of %lu files",
1275                                 required_features->reparse_points);
1276                 }
1277         }
1278
1279         /* Security descriptors.  */
1280         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1281                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1282              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1283             required_features->security_descriptors &&
1284             !supported_features->security_descriptors)
1285         {
1286                 ERROR("Extraction backend does not support security descriptors!");
1287                 return WIMLIB_ERR_UNSUPPORTED;
1288         }
1289         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1290             required_features->security_descriptors &&
1291             !supported_features->security_descriptors)
1292                 WARNING("Ignoring Windows NT security descriptors of %lu files",
1293                         required_features->security_descriptors);
1294
1295         /* UNIX data.  */
1296         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1297             required_features->unix_data && !supported_features->unix_data)
1298         {
1299                 ERROR("Extraction backend does not support UNIX data!");
1300                 return WIMLIB_ERR_UNSUPPORTED;
1301         }
1302
1303         if (required_features->unix_data &&
1304             !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1305         {
1306                 WARNING("Ignoring UNIX metadata of %lu files",
1307                         required_features->unix_data);
1308         }
1309
1310         /* DOS Names.  */
1311         if (required_features->short_names &&
1312             !supported_features->short_names)
1313         {
1314                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1315                         ERROR("Extraction backend does not support DOS names!");
1316                         return WIMLIB_ERR_UNSUPPORTED;
1317                 }
1318                 WARNING("Ignoring DOS names of %lu files",
1319                         required_features->short_names);
1320         }
1321
1322         /* Timestamps.  */
1323         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1324             !supported_features->timestamps)
1325         {
1326                 ERROR("Extraction backend does not support timestamps!");
1327                 return WIMLIB_ERR_UNSUPPORTED;
1328         }
1329
1330         return 0;
1331 }
1332
1333 static const struct apply_operations *
1334 select_apply_operations(int extract_flags)
1335 {
1336 #ifdef WITH_NTFS_3G
1337         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1338                 return &ntfs_3g_apply_ops;
1339 #endif
1340 #ifdef __WIN32__
1341         return &win32_apply_ops;
1342 #else
1343         return &unix_apply_ops;
1344 #endif
1345 }
1346
1347 static int
1348 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1349               const tchar *target, int extract_flags)
1350 {
1351         const struct apply_operations *ops;
1352         struct apply_ctx *ctx;
1353         int ret;
1354         LIST_HEAD(dentry_list);
1355
1356         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1357                 ret = extract_dentries_to_stdout(trees, num_trees,
1358                                                  wim->blob_table);
1359                 goto out;
1360         }
1361
1362         num_trees = remove_duplicate_trees(trees, num_trees);
1363         num_trees = remove_contained_trees(trees, num_trees);
1364
1365         ops = select_apply_operations(extract_flags);
1366
1367         if (num_trees > 1 && ops->single_tree_only) {
1368                 ERROR("Extracting multiple directory trees "
1369                       "at once is not supported in %s extraction mode!",
1370                       ops->name);
1371                 ret = WIMLIB_ERR_UNSUPPORTED;
1372                 goto out;
1373         }
1374
1375         ctx = CALLOC(1, ops->context_size);
1376         if (!ctx) {
1377                 ret = WIMLIB_ERR_NOMEM;
1378                 goto out;
1379         }
1380
1381         ctx->wim = wim;
1382         ctx->target = target;
1383         ctx->target_nchars = tstrlen(target);
1384         ctx->extract_flags = extract_flags;
1385         if (ctx->wim->progfunc) {
1386                 ctx->progfunc = ctx->wim->progfunc;
1387                 ctx->progctx = ctx->wim->progctx;
1388                 ctx->progress.extract.image = wim->current_image;
1389                 ctx->progress.extract.extract_flags = (extract_flags &
1390                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
1391                 ctx->progress.extract.wimfile_name = wim->filename;
1392                 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1393                                                                          wim->current_image);
1394                 ctx->progress.extract.target = target;
1395         }
1396         INIT_LIST_HEAD(&ctx->blob_list);
1397         filedes_invalidate(&ctx->tmpfile_fd);
1398         ctx->apply_ops = ops;
1399
1400         ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1401         if (ret)
1402                 goto out_cleanup;
1403
1404         build_dentry_list(&dentry_list, trees, num_trees,
1405                           !(extract_flags &
1406                             WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1407
1408         dentry_list_get_features(&dentry_list, &ctx->required_features);
1409
1410         ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1411                                ctx->extract_flags);
1412         if (ret)
1413                 goto out_cleanup;
1414
1415         ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1416         if (ret)
1417                 goto out_cleanup;
1418
1419         if (unlikely(list_empty(&dentry_list))) {
1420                 WARNING("There is nothing to extract!");
1421                 goto out_cleanup;
1422         }
1423
1424         ret = dentry_list_resolve_streams(&dentry_list, ctx);
1425         if (ret)
1426                 goto out_cleanup;
1427
1428         dentry_list_build_inode_alias_lists(&dentry_list);
1429
1430         ret = dentry_list_ref_streams(&dentry_list, ctx);
1431         if (ret)
1432                 goto out_cleanup;
1433
1434         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1435                 /* When extracting from a pipe, the number of bytes of data to
1436                  * extract can't be determined in the normal way (examining the
1437                  * blob table), since at this point all we have is a set of
1438                  * SHA-1 message digests of blobs that need to be extracted.
1439                  * However, we can get a reasonably accurate estimate by taking
1440                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1441                  * data.  This does assume that a full image is being extracted,
1442                  * but currently there is no API for doing otherwise.  (Also,
1443                  * subtract <HARDLINKBYTES> from this if hard links are
1444                  * supported by the extraction mode.)  */
1445                 ctx->progress.extract.total_bytes =
1446                         xml_get_image_total_bytes(wim->xml_info,
1447                                                   wim->current_image);
1448                 if (ctx->supported_features.hard_links) {
1449                         ctx->progress.extract.total_bytes -=
1450                                 xml_get_image_hard_link_bytes(wim->xml_info,
1451                                                               wim->current_image);
1452                 }
1453         }
1454
1455         ret = extract_progress(ctx,
1456                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1457                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
1458                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
1459         if (ret)
1460                 goto out_cleanup;
1461
1462         ret = (*ops->extract)(&dentry_list, ctx);
1463         if (ret)
1464                 goto out_cleanup;
1465
1466         if (ctx->progress.extract.completed_bytes <
1467             ctx->progress.extract.total_bytes)
1468         {
1469                 ctx->progress.extract.completed_bytes =
1470                         ctx->progress.extract.total_bytes;
1471                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
1472                 if (ret)
1473                         goto out_cleanup;
1474         }
1475
1476         ret = extract_progress(ctx,
1477                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1478                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
1479                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
1480 out_cleanup:
1481         destroy_blob_list(&ctx->blob_list);
1482         destroy_dentry_list(&dentry_list);
1483         FREE(ctx);
1484 out:
1485         return ret;
1486 }
1487
1488 static int
1489 mkdir_if_needed(const tchar *target)
1490 {
1491         if (!tmkdir(target, 0755))
1492                 return 0;
1493
1494         if (errno == EEXIST)
1495                 return 0;
1496
1497 #ifdef __WIN32__
1498         /* _wmkdir() fails with EACCES if called on a drive root directory.  */
1499         if (errno == EACCES)
1500                 return 0;
1501 #endif
1502
1503         ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
1504         return WIMLIB_ERR_MKDIR;
1505 }
1506
1507 /* Make sure the extraction flags make sense, and update them if needed.  */
1508 static int
1509 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1510 {
1511         int extract_flags = *extract_flags_p;
1512
1513         /* Check for invalid flag combinations  */
1514
1515         if ((extract_flags &
1516              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1517               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1518                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1519                 return WIMLIB_ERR_INVALID_PARAM;
1520
1521         if ((extract_flags &
1522              (WIMLIB_EXTRACT_FLAG_RPFIX |
1523               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1524                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
1525                 return WIMLIB_ERR_INVALID_PARAM;
1526
1527 #ifndef WITH_NTFS_3G
1528         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1529                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1530                       "        it cannot apply a WIM image directly to an NTFS volume.");
1531                 return WIMLIB_ERR_UNSUPPORTED;
1532         }
1533 #endif
1534
1535         if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1536 #ifdef __WIN32__
1537                 if (!wim->filename)
1538                         return WIMLIB_ERR_NO_FILENAME;
1539 #else
1540                 ERROR("WIMBoot extraction is only supported on Windows!");
1541                 return WIMLIB_ERR_UNSUPPORTED;
1542 #endif
1543         }
1544
1545         if (extract_flags & (WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K |
1546                              WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K |
1547                              WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K |
1548                              WIMLIB_EXTRACT_FLAG_COMPACT_LZX))
1549         {
1550         #ifdef __WIN32__
1551                 int count = 0;
1552                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K) != 0);
1553                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K) != 0);
1554                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K) != 0);
1555                 count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_LZX) != 0);
1556                 if (count != 1) {
1557                         ERROR("Only one compression format can be specified "
1558                               "for compact-mode extraction!");
1559                         return WIMLIB_ERR_INVALID_PARAM;
1560                 }
1561                 if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1562                         ERROR("Compact-mode extraction and WIMBoot-mode "
1563                               "extraction are mutually exclusive!");
1564                         return WIMLIB_ERR_INVALID_PARAM;
1565                 }
1566         #else
1567                 ERROR("Compact-mode extraction (System Compression) "
1568                       "is only supported on Windows!");
1569                 return WIMLIB_ERR_UNSUPPORTED;
1570         #endif
1571         }
1572
1573
1574         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1575                               WIMLIB_EXTRACT_FLAG_NORPFIX |
1576                               WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1577                                         WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1578         {
1579                 /* For full-image extraction, do reparse point fixups by default
1580                  * if the WIM header says they are enabled.  */
1581                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1582                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1583         }
1584
1585         *extract_flags_p = extract_flags;
1586         return 0;
1587 }
1588
1589 struct append_dentry_ctx {
1590         struct wim_dentry **dentries;
1591         size_t num_dentries;
1592         size_t num_alloc_dentries;
1593 };
1594
1595 static int
1596 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1597 {
1598         struct append_dentry_ctx *ctx = _ctx;
1599
1600         if (ctx->num_dentries == ctx->num_alloc_dentries) {
1601                 struct wim_dentry **new_dentries;
1602                 size_t new_length;
1603
1604                 new_length = max(ctx->num_alloc_dentries + 8,
1605                                  ctx->num_alloc_dentries * 3 / 2);
1606                 new_dentries = REALLOC(ctx->dentries,
1607                                        new_length * sizeof(ctx->dentries[0]));
1608                 if (new_dentries == NULL)
1609                         return WIMLIB_ERR_NOMEM;
1610                 ctx->dentries = new_dentries;
1611                 ctx->num_alloc_dentries = new_length;
1612         }
1613         ctx->dentries[ctx->num_dentries++] = dentry;
1614         return 0;
1615 }
1616
1617 /* Append dentries matched by a path which can contain wildcard characters.  */
1618 static int
1619 append_matched_dentries(WIMStruct *wim, const tchar *orig_pattern,
1620                         int extract_flags, struct append_dentry_ctx *ctx)
1621 {
1622         const size_t count_before = ctx->num_dentries;
1623         tchar *pattern;
1624         int ret;
1625
1626         pattern = canonicalize_wim_path(orig_pattern);
1627         if (!pattern)
1628                 return WIMLIB_ERR_NOMEM;
1629         ret = expand_path_pattern(wim_get_current_root_dentry(wim), pattern,
1630                                   append_dentry_cb, ctx);
1631         FREE(pattern);
1632         if (ret || ctx->num_dentries > count_before)
1633                 return ret;
1634         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB) {
1635                 ERROR("No matches for path pattern \"%"TS"\"", orig_pattern);
1636                 return WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1637         }
1638         WARNING("No matches for path pattern \"%"TS"\"", orig_pattern);
1639         return 0;
1640 }
1641
1642 static int
1643 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1644                         const tchar * const *paths, size_t num_paths,
1645                         int extract_flags)
1646 {
1647         int ret;
1648         struct wim_dentry **trees;
1649         size_t num_trees;
1650
1651         if (wim == NULL || target == NULL || target[0] == T('\0') ||
1652             (num_paths != 0 && paths == NULL))
1653                 return WIMLIB_ERR_INVALID_PARAM;
1654
1655         ret = check_extract_flags(wim, &extract_flags);
1656         if (ret)
1657                 return ret;
1658
1659         ret = select_wim_image(wim, image);
1660         if (ret)
1661                 return ret;
1662
1663         ret = wim_checksum_unhashed_blobs(wim);
1664         if (ret)
1665                 return ret;
1666
1667         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1668                               WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1669             (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1670         {
1671                 ret = mkdir_if_needed(target);
1672                 if (ret)
1673                         return ret;
1674         }
1675
1676         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1677
1678                 struct append_dentry_ctx append_dentry_ctx = {
1679                         .dentries = NULL,
1680                         .num_dentries = 0,
1681                         .num_alloc_dentries = 0,
1682                 };
1683
1684                 for (size_t i = 0; i < num_paths; i++) {
1685                         ret = append_matched_dentries(wim, paths[i],
1686                                                       extract_flags,
1687                                                       &append_dentry_ctx);
1688                         if (ret) {
1689                                 trees = append_dentry_ctx.dentries;
1690                                 goto out_free_trees;
1691                         }
1692                 }
1693                 trees = append_dentry_ctx.dentries;
1694                 num_trees = append_dentry_ctx.num_dentries;
1695         } else {
1696                 trees = MALLOC(num_paths * sizeof(trees[0]));
1697                 if (trees == NULL)
1698                         return WIMLIB_ERR_NOMEM;
1699
1700                 for (size_t i = 0; i < num_paths; i++) {
1701
1702                         tchar *path = canonicalize_wim_path(paths[i]);
1703                         if (path == NULL) {
1704                                 ret = WIMLIB_ERR_NOMEM;
1705                                 goto out_free_trees;
1706                         }
1707
1708                         trees[i] = get_dentry(wim, path,
1709                                               WIMLIB_CASE_PLATFORM_DEFAULT);
1710                         FREE(path);
1711                         if (trees[i] == NULL) {
1712                                   ERROR("Path \"%"TS"\" does not exist "
1713                                         "in WIM image %d",
1714                                         paths[i], wim->current_image);
1715                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1716                                   goto out_free_trees;
1717                         }
1718                 }
1719                 num_trees = num_paths;
1720         }
1721
1722         if (num_trees == 0) {
1723                 ret = 0;
1724                 goto out_free_trees;
1725         }
1726
1727         ret = extract_trees(wim, trees, num_trees, target, extract_flags);
1728 out_free_trees:
1729         FREE(trees);
1730         return ret;
1731 }
1732
1733 static int
1734 extract_single_image(WIMStruct *wim, int image,
1735                      const tchar *target, int extract_flags)
1736 {
1737         const tchar *path = WIMLIB_WIM_ROOT_PATH;
1738         extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1739         return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
1740 }
1741
1742 static const tchar * const filename_forbidden_chars =
1743 T(
1744 #ifdef __WIN32__
1745 "<>:\"/\\|?*"
1746 #else
1747 "/"
1748 #endif
1749 );
1750
1751 /* This function checks if it is okay to use a WIM image's name as a directory
1752  * name.  */
1753 static bool
1754 image_name_ok_as_dir(const tchar *image_name)
1755 {
1756         return image_name && *image_name &&
1757                 !tstrpbrk(image_name, filename_forbidden_chars) &&
1758                 tstrcmp(image_name, T(".")) &&
1759                 tstrcmp(image_name, T("..")) &&
1760                 tstrlen(image_name) <= 128;
1761 }
1762
1763 /* Extracts all images from the WIM to the directory @target, with the images
1764  * placed in subdirectories named by their image names. */
1765 static int
1766 extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
1767 {
1768         size_t output_path_len = tstrlen(target);
1769         tchar buf[output_path_len + 1 + 128 + 1];
1770         int ret;
1771         int image;
1772         const tchar *image_name;
1773
1774         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1775                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1776                 return WIMLIB_ERR_INVALID_PARAM;
1777         }
1778
1779         ret = mkdir_if_needed(target);
1780         if (ret)
1781                 return ret;
1782         tmemcpy(buf, target, output_path_len);
1783         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1784         for (image = 1; image <= wim->hdr.image_count; image++) {
1785                 image_name = wimlib_get_image_name(wim, image);
1786                 if (image_name_ok_as_dir(image_name)) {
1787                         tstrcpy(buf + output_path_len + 1, image_name);
1788                 } else {
1789                         /* Image name is empty or contains forbidden characters.
1790                          * Use image number instead. */
1791                         tsprintf(buf + output_path_len + 1, T("%d"), image);
1792                 }
1793                 ret = extract_single_image(wim, image, buf, extract_flags);
1794                 if (ret)
1795                         return ret;
1796         }
1797         return 0;
1798 }
1799
1800 static int
1801 do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1802                         int extract_flags)
1803 {
1804         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1805                              WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1806                              WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1807                 return WIMLIB_ERR_INVALID_PARAM;
1808
1809         if (image == WIMLIB_ALL_IMAGES)
1810                 return extract_all_images(wim, target, extract_flags);
1811         else
1812                 return extract_single_image(wim, image, target, extract_flags);
1813 }
1814
1815
1816 /****************************************************************************
1817  *                          Extraction API                                  *
1818  ****************************************************************************/
1819
1820 WIMLIBAPI int
1821 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1822                      const tchar * const *paths, size_t num_paths,
1823                      int extract_flags)
1824 {
1825         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1826                 return WIMLIB_ERR_INVALID_PARAM;
1827
1828         return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1829                                        extract_flags);
1830 }
1831
1832 WIMLIBAPI int
1833 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1834                         const tchar *path_list_file, int extract_flags)
1835 {
1836         int ret;
1837         tchar **paths;
1838         size_t num_paths;
1839         void *mem;
1840
1841         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1842         if (ret) {
1843                 ERROR("Failed to read path list file \"%"TS"\"",
1844                       path_list_file);
1845                 return ret;
1846         }
1847
1848         ret = wimlib_extract_paths(wim, image, target,
1849                                    (const tchar * const *)paths, num_paths,
1850                                    extract_flags);
1851         FREE(paths);
1852         FREE(mem);
1853         return ret;
1854 }
1855
1856 WIMLIBAPI int
1857 wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
1858                                              const tchar *image_num_or_name,
1859                                              const tchar *target,
1860                                              int extract_flags,
1861                                              wimlib_progress_func_t progfunc,
1862                                              void *progctx)
1863 {
1864         int ret;
1865         WIMStruct *pwm;
1866         struct filedes *in_fd;
1867         int image;
1868         unsigned i;
1869
1870         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1871                 return WIMLIB_ERR_INVALID_PARAM;
1872
1873         /* Read the WIM header from the pipe and get a WIMStruct to represent
1874          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
1875          * wimlib_open_wim(), getting a WIMStruct in this way will result in an
1876          * empty blob table, no XML data read, and no filename set.  */
1877         ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
1878                                     progfunc, progctx);
1879         if (ret)
1880                 return ret;
1881
1882         /* Sanity check to make sure this is a pipable WIM.  */
1883         if (pwm->hdr.magic != PWM_MAGIC) {
1884                 ERROR("The WIM being read from file descriptor %d "
1885                       "is not pipable!", pipe_fd);
1886                 ret = WIMLIB_ERR_NOT_PIPABLE;
1887                 goto out_wimlib_free;
1888         }
1889
1890         /* Sanity check to make sure the first part of a pipable split WIM is
1891          * sent over the pipe first.  */
1892         if (pwm->hdr.part_number != 1) {
1893                 ERROR("The first part of the split WIM must be "
1894                       "sent over the pipe first.");
1895                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1896                 goto out_wimlib_free;
1897         }
1898
1899         in_fd = &pwm->in_fd;
1900         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1901
1902         /* As mentioned, the WIMStruct we created from the pipe does not have
1903          * XML data yet.  Fix this by reading the extra copy of the XML data
1904          * that directly follows the header in pipable WIMs.  (Note: see
1905          * write_pipable_wim() for more details about the format of pipable
1906          * WIMs.)  */
1907         {
1908                 u8 hash[SHA1_HASH_SIZE];
1909
1910                 ret = read_pwm_blob_header(pwm, hash,
1911                                            &pwm->hdr.xml_data_reshdr, NULL);
1912                 if (ret)
1913                         goto out_wimlib_free;
1914
1915                 if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1916                         ERROR("Expected XML data, but found non-metadata resource.");
1917                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1918                         goto out_wimlib_free;
1919                 }
1920
1921                 ret = read_wim_xml_data(pwm);
1922                 if (ret)
1923                         goto out_wimlib_free;
1924
1925                 if (xml_get_image_count(pwm->xml_info) != pwm->hdr.image_count) {
1926                         ERROR("Image count in XML data is not the same as in WIM header.");
1927                         ret = WIMLIB_ERR_IMAGE_COUNT;
1928                         goto out_wimlib_free;
1929                 }
1930         }
1931
1932         /* Get image index (this may use the XML data that was just read to
1933          * resolve an image name).  */
1934         if (image_num_or_name) {
1935                 image = wimlib_resolve_image(pwm, image_num_or_name);
1936                 if (image == WIMLIB_NO_IMAGE) {
1937                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1938                               image_num_or_name);
1939                         ret = WIMLIB_ERR_INVALID_IMAGE;
1940                         goto out_wimlib_free;
1941                 } else if (image == WIMLIB_ALL_IMAGES) {
1942                         ERROR("Applying all images from a pipe is not supported!");
1943                         ret = WIMLIB_ERR_INVALID_IMAGE;
1944                         goto out_wimlib_free;
1945                 }
1946         } else {
1947                 if (pwm->hdr.image_count != 1) {
1948                         ERROR("No image was specified, but the pipable WIM "
1949                               "did not contain exactly 1 image");
1950                         ret = WIMLIB_ERR_INVALID_IMAGE;
1951                         goto out_wimlib_free;
1952                 }
1953                 image = 1;
1954         }
1955
1956         /* Load the needed metadata resource.  */
1957         for (i = 1; i <= pwm->hdr.image_count; i++) {
1958                 ret = handle_pwm_metadata_resource(pwm, i, i == image);
1959                 if (ret)
1960                         goto out_wimlib_free;
1961         }
1962         /* Extract the image.  */
1963         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
1964         ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
1965         /* Clean up and return.  */
1966 out_wimlib_free:
1967         wimlib_free(pwm);
1968         return ret;
1969 }
1970
1971
1972 WIMLIBAPI int
1973 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
1974                                const tchar *target, int extract_flags)
1975 {
1976         return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
1977                                                             image_num_or_name,
1978                                                             target,
1979                                                             extract_flags,
1980                                                             NULL,
1981                                                             NULL);
1982 }
1983
1984 WIMLIBAPI int
1985 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1986                      int extract_flags)
1987 {
1988         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1989                 return WIMLIB_ERR_INVALID_PARAM;
1990         return do_wimlib_extract_image(wim, image, target, extract_flags);
1991 }