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

   1 /*
   2  * extract.c
   3  *
   4  * Support for extracting WIM images, or files or directories contained in a WIM
   5  * image.
   6  */
   7
   8 /*
   9  * Copyright (C) 2012, 2013, 2014, 2015 Eric Biggers
  10  *
  11  * This file is free software; you can redistribute it and/or modify it under
  12  * the terms of the GNU Lesser General Public License as published by the Free
  13  * Software Foundation; either version 3 of the License, or (at your option) any
  14  * later version.
  15  *
  16  * This file is distributed in the hope that it will be useful, but WITHOUT
  17  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  18  * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  19  * details.
  20  *
  21  * You should have received a copy of the GNU Lesser General Public License
  22  * along with this file; if not, see http://www.gnu.org/licenses/.
  23  */
  24
  25 /*
  26  * This file provides the API functions wimlib_extract_image(),
  27  * wimlib_extract_image_from_pipe(), wimlib_extract_paths(), and
  28  * wimlib_extract_pathlist().  Internally, all end up calling
  29  * do_wimlib_extract_paths() and extract_trees().
  30  *
  31  * Although wimlib supports multiple extraction modes/backends (NTFS-3g, UNIX,
  32  * Win32), this file does not itself have code to extract files or directories
  33  * to any specific target; instead, it handles generic functionality and relies
  34  * on lower-level callback functions declared in `struct apply_operations' to do
  35  * the actual extraction.
  36  */
  37
  38 #ifdef HAVE_CONFIG_H
  39 #  include "config.h"
  40 #endif
  41
  42 #include <errno.h>
  43 #include <fcntl.h>
  44 #include <sys/stat.h>
  45 #include <unistd.h>
  46
  47 #include "wimlib/apply.h"
  48 #include "wimlib/assert.h"
  49 #include "wimlib/blob_table.h"
  50 #include "wimlib/dentry.h"
  51 #include "wimlib/encoding.h"
  52 #include "wimlib/endianness.h"
  53 #include "wimlib/error.h"
  54 #include "wimlib/metadata.h"
  55 #include "wimlib/pathlist.h"
  56 #include "wimlib/paths.h"
  57 #include "wimlib/pattern.h"
  58 #include "wimlib/reparse.h"
  59 #include "wimlib/resource.h"
  60 #include "wimlib/security.h"
  61 #include "wimlib/unix_data.h"
  62 #include "wimlib/wim.h"
  63 #include "wimlib/win32.h" /* for realpath() equivalent */
  64 #include "wimlib/xml.h"
  65
  66 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x80000000
  67 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x40000000
  68
  69 /* Keep in sync with wimlib.h  */
  70 #define WIMLIB_EXTRACT_MASK_PUBLIC                              \
  71         (WIMLIB_EXTRACT_FLAG_NTFS                       |       \
  72          WIMLIB_EXTRACT_FLAG_UNIX_DATA                  |       \
  73          WIMLIB_EXTRACT_FLAG_NO_ACLS                    |       \
  74          WIMLIB_EXTRACT_FLAG_STRICT_ACLS                |       \
  75          WIMLIB_EXTRACT_FLAG_RPFIX                      |       \
  76          WIMLIB_EXTRACT_FLAG_NORPFIX                    |       \
  77          WIMLIB_EXTRACT_FLAG_TO_STDOUT                  |       \
  78          WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES  |       \
  79          WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS         |       \
  80          WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS          |       \
  81          WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES         |       \
  82          WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS            |       \
  83          WIMLIB_EXTRACT_FLAG_GLOB_PATHS                 |       \
  84          WIMLIB_EXTRACT_FLAG_STRICT_GLOB                |       \
  85          WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES              |       \
  86          WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE  |       \
  87          WIMLIB_EXTRACT_FLAG_WIMBOOT)
  88
  89 /* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or
  90  * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA.  */
  91 int
  92 do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
  93 {
  94         ctx->count_until_file_progress = 500;  /* Arbitrary value to limit calls  */
  95         return extract_progress(ctx, msg);
  96 }
  97
  98 static int
  99 start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg)
 100 {
 101         ctx->progress.extract.current_file_count = 0;
 102         ctx->progress.extract.end_file_count = end_file_count;
 103         return do_file_extract_progress(ctx, msg);
 104 }
 105
 106 int
 107 start_file_structure_phase(struct apply_ctx *ctx, u64 end_file_count)
 108 {
 109         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 110 }
 111
 112 int
 113 start_file_metadata_phase(struct apply_ctx *ctx, u64 end_file_count)
 114 {
 115         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 116 }
 117
 118 static int
 119 end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
 120 {
 121         ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count;
 122         return do_file_extract_progress(ctx, msg);
 123 }
 124
 125 int
 126 end_file_structure_phase(struct apply_ctx *ctx)
 127 {
 128         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 129 }
 130
 131 int
 132 end_file_metadata_phase(struct apply_ctx *ctx)
 133 {
 134         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 135 }
 136
 137 #define PWM_FOUND_WIM_HDR (-1)
 138
 139 /* Read the header for a blob in a pipable WIM.  If @pwm_hdr_ret is not NULL,
 140  * also look for a pipable WIM header and return PWM_FOUND_WIM_HDR if found.  */
 141 static int
 142 read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE],
 143                      struct wim_reshdr *reshdr_ret,
 144                      struct wim_header_disk *pwm_hdr_ret)
 145 {
 146         int ret;
 147         struct pwm_blob_hdr blob_hdr;
 148         u64 magic;
 149
 150         ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr));
 151         if (unlikely(ret))
 152                 goto read_error;
 153
 154         magic = le64_to_cpu(blob_hdr.magic);
 155
 156         if (magic == PWM_MAGIC && pwm_hdr_ret != NULL) {
 157                 memcpy(pwm_hdr_ret, &blob_hdr, sizeof(blob_hdr));
 158                 ret = full_read(&pwm->in_fd,
 159                                 (u8 *)pwm_hdr_ret + sizeof(blob_hdr),
 160                                 sizeof(*pwm_hdr_ret) - sizeof(blob_hdr));
 161                 if (unlikely(ret))
 162                         goto read_error;
 163                 return PWM_FOUND_WIM_HDR;
 164         }
 165
 166         if (unlikely(magic != PWM_BLOB_MAGIC)) {
 167                 ERROR("Data read on pipe is invalid (expected blob header)");
 168                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 169         }
 170
 171         copy_hash(hash_ret, blob_hdr.hash);
 172
 173         reshdr_ret->size_in_wim = 0; /* Not available  */
 174         reshdr_ret->flags = le32_to_cpu(blob_hdr.flags);
 175         reshdr_ret->offset_in_wim = pwm->in_fd.offset;
 176         reshdr_ret->uncompressed_size = le64_to_cpu(blob_hdr.uncompressed_size);
 177
 178         if (unlikely(reshdr_ret->uncompressed_size == 0)) {
 179                 ERROR("Data read on pipe is invalid (resource is of 0 size)");
 180                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 181         }
 182
 183         return 0;
 184
 185 read_error:
 186         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE)
 187                 ERROR("The pipe ended before all needed data was sent!");
 188         else
 189                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
 190         return ret;
 191 }
 192
 193 static int
 194 read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 195 {
 196         int ret;
 197         u8 hash[SHA1_HASH_SIZE];
 198         struct wim_reshdr reshdr;
 199         struct wim_header_disk pwm_hdr;
 200         struct wim_resource_descriptor rdesc;
 201         struct blob_descriptor *blob;
 202
 203         copy_guid(ctx->progress.extract.guid, ctx->wim->hdr.guid);
 204         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
 205         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
 206         ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 207         if (ret)
 208                 return ret;
 209
 210         while (ctx->num_blobs_remaining) {
 211
 212                 ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr);
 213
 214                 if (ret == PWM_FOUND_WIM_HDR) {
 215                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
 216                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
 217
 218                         if (part_number == ctx->progress.extract.part_number &&
 219                             total_parts == ctx->progress.extract.total_parts &&
 220                             guids_equal(pwm_hdr.guid, ctx->progress.extract.guid))
 221                                 continue;
 222
 223                         copy_guid(ctx->progress.extract.guid, pwm_hdr.guid);
 224                         ctx->progress.extract.part_number = part_number;
 225                         ctx->progress.extract.total_parts = total_parts;
 226                         ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 227                         if (ret)
 228                                 return ret;
 229
 230                         continue;
 231                 }
 232
 233                 if (ret)
 234                         return ret;
 235
 236                 if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)
 237                     && (blob = lookup_blob(ctx->wim->blob_table, hash))
 238                     && (blob->out_refcnt))
 239                 {
 240                         wim_reshdr_to_desc_and_blob(&reshdr, ctx->wim, &rdesc, blob);
 241                         ret = read_blob_with_sha1(blob, cbs);
 242                         blob_unset_is_located_in_wim_resource(blob);
 243                         if (ret)
 244                                 return ret;
 245                         ctx->num_blobs_remaining--;
 246                 } else {
 247                         wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc);
 248                         ret = skip_wim_resource(&rdesc);
 249                         if (ret)
 250                                 return ret;
 251                 }
 252         }
 253
 254         return 0;
 255 }
 256
 257 /* Creates a temporary file opened for writing.  The open file descriptor is
 258  * returned in @fd_ret and its name is returned in @name_ret (dynamically
 259  * allocated).  */
 260 static int
 261 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
 262 {
 263         tchar *name;
 264         int open_flags;
 265         int raw_fd;
 266
 267 retry:
 268         name = ttempnam(NULL, T("wimlib"));
 269         if (!name) {
 270                 ERROR_WITH_ERRNO("Failed to create temporary filename");
 271                 return WIMLIB_ERR_NOMEM;
 272         }
 273
 274         open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
 275 #ifdef __WIN32__
 276         open_flags |= _O_SHORT_LIVED;
 277 #endif
 278         raw_fd = topen(name, open_flags, 0600);
 279
 280         if (raw_fd < 0) {
 281                 if (errno == EEXIST) {
 282                         FREE(name);
 283                         goto retry;
 284                 }
 285                 ERROR_WITH_ERRNO("Failed to create temporary file "
 286                                  "\"%"TS"\"", name);
 287                 FREE(name);
 288                 return WIMLIB_ERR_OPEN;
 289         }
 290
 291         filedes_init(fd_ret, raw_fd);
 292         *name_ret = name;
 293         return 0;
 294 }
 295
 296 static int
 297 begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
 298 {
 299         struct apply_ctx *ctx = _ctx;
 300
 301         ctx->cur_blob = blob;
 302         ctx->cur_blob_offset = 0;
 303
 304         if (unlikely(blob->out_refcnt > MAX_OPEN_FILES))
 305                 return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
 306
 307         return call_begin_blob(blob, ctx->saved_cbs);
 308 }
 309
 310 static int
 311 extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
 312 {
 313         struct apply_ctx *ctx = _ctx;
 314         union wimlib_progress_info *progress = &ctx->progress;
 315         int ret;
 316
 317         ctx->cur_blob_offset += size;
 318
 319         if (likely(ctx->supported_features.hard_links)) {
 320                 progress->extract.completed_bytes +=
 321                         (u64)size * ctx->cur_blob->out_refcnt;
 322                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 323                         progress->extract.completed_streams += ctx->cur_blob->out_refcnt;
 324         } else {
 325                 const struct blob_extraction_target *targets =
 326                         blob_extraction_targets(ctx->cur_blob);
 327                 for (u32 i = 0; i < ctx->cur_blob->out_refcnt; i++) {
 328                         const struct wim_inode *inode = targets[i].inode;
 329                         const struct wim_dentry *dentry;
 330
 331                         inode_for_each_extraction_alias(dentry, inode) {
 332                                 progress->extract.completed_bytes += size;
 333                                 if (ctx->cur_blob_offset == ctx->cur_blob->size)
 334                                         progress->extract.completed_streams++;
 335                         }
 336                 }
 337         }
 338         if (progress->extract.completed_bytes >= ctx->next_progress) {
 339
 340                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
 341                 if (ret)
 342                         return ret;
 343
 344                 if (progress->extract.completed_bytes >=
 345                     progress->extract.total_bytes)
 346                 {
 347                         ctx->next_progress = UINT64_MAX;
 348                 } else {
 349                         /* Send new message as soon as another 1/128 of the
 350                          * total has been extracted.  (Arbitrary number.)  */
 351                         ctx->next_progress =
 352                                 progress->extract.completed_bytes +
 353                                         progress->extract.total_bytes / 128;
 354
 355                         /* ... Unless that would be more than 5000000 bytes, in
 356                          * which case send the next after the next 5000000
 357                          * bytes.  (Another arbitrary number.)  */
 358                         if (progress->extract.completed_bytes + 5000000 <
 359                             ctx->next_progress)
 360                                 ctx->next_progress =
 361                                         progress->extract.completed_bytes + 5000000;
 362
 363                         /* ... But always send a message as soon as we're
 364                          * completely done.  */
 365                         if (progress->extract.total_bytes < ctx->next_progress)
 366                                 ctx->next_progress = progress->extract.total_bytes;
 367                 }
 368         }
 369
 370         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 371                 /* Just extracting to temporary file for now.  */
 372                 ret = full_write(&ctx->tmpfile_fd, chunk, size);
 373                 if (ret) {
 374                         ERROR_WITH_ERRNO("Error writing data to "
 375                                          "temporary file \"%"TS"\"",
 376                                          ctx->tmpfile_name);
 377                 }
 378                 return ret;
 379         }
 380
 381         return call_consume_chunk(chunk, size, ctx->saved_cbs);
 382 }
 383
 384 /* Copy the blob's data from the temporary file to each of its targets.
 385  *
 386  * This is executed only in the very uncommon case that a blob is being
 387  * extracted to more than MAX_OPEN_FILES targets!  */
 388 static int
 389 extract_from_tmpfile(const tchar *tmpfile_name,
 390                      const struct blob_descriptor *orig_blob,
 391                      const struct read_blob_callbacks *cbs)
 392 {
 393         struct blob_descriptor tmpfile_blob;
 394         const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob);
 395         int ret;
 396
 397         memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor));
 398         tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK;
 399         tmpfile_blob.file_on_disk = (tchar *)tmpfile_name;
 400         tmpfile_blob.out_refcnt = 1;
 401
 402         for (u32 i = 0; i < orig_blob->out_refcnt; i++) {
 403                 tmpfile_blob.inline_blob_extraction_targets[0] = targets[i];
 404                 ret = read_blob_with_cbs(&tmpfile_blob, cbs);
 405                 if (ret)
 406                         return ret;
 407         }
 408         return 0;
 409 }
 410
 411 static int
 412 end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
 413 {
 414         struct apply_ctx *ctx = _ctx;
 415
 416         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 417                 filedes_close(&ctx->tmpfile_fd);
 418                 if (!status)
 419                         status = extract_from_tmpfile(ctx->tmpfile_name, blob,
 420                                                       ctx->saved_cbs);
 421                 filedes_invalidate(&ctx->tmpfile_fd);
 422                 tunlink(ctx->tmpfile_name);
 423                 FREE(ctx->tmpfile_name);
 424                 return status;
 425         }
 426
 427         return call_end_blob(blob, status, ctx->saved_cbs);
 428 }
 429
 430 /*
 431  * Read the list of blobs to extract and feed their data into the specified
 432  * callback functions.
 433  *
 434  * This handles checksumming each blob.
 435  *
 436  * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
 437  *
 438  * This also works if the WIM is being read from a pipe.
 439  *
 440  * This also will split up blobs that will need to be extracted to more than
 441  * MAX_OPEN_FILES locations, as measured by the 'out_refcnt' of each blob.
 442  * Therefore, the apply_operations implementation need not worry about running
 443  * out of file descriptors, unless it might open more than one file descriptor
 444  * per 'blob_extraction_target' (e.g. Win32 currently might because the
 445  * destination file system might not support hard links).
 446  */
 447 int
 448 extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 449 {
 450         struct read_blob_callbacks wrapper_cbs = {
 451                 .begin_blob     = begin_extract_blob_wrapper,
 452                 .consume_chunk  = extract_chunk_wrapper,
 453                 .end_blob       = end_extract_blob_wrapper,
 454                 .ctx            = ctx,
 455         };
 456         ctx->saved_cbs = cbs;
 457         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
 458                 return read_blobs_from_pipe(ctx, &wrapper_cbs);
 459         } else {
 460                 return read_blob_list(&ctx->blob_list,
 461                                       offsetof(struct blob_descriptor,
 462                                                extraction_list),
 463                                       &wrapper_cbs, VERIFY_BLOB_HASHES);
 464         }
 465 }
 466
 467 /* Extract a WIM dentry to standard output.
 468  *
 469  * This obviously doesn't make sense in all cases.  We return an error if the
 470  * dentry does not correspond to a regular file.  Otherwise we extract the
 471  * unnamed data stream only.  */
 472 static int
 473 extract_dentry_to_stdout(struct wim_dentry *dentry,
 474                          const struct blob_table *blob_table)
 475 {
 476         struct wim_inode *inode = dentry->d_inode;
 477         struct blob_descriptor *blob;
 478         struct filedes _stdout;
 479
 480         if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
 481                                    FILE_ATTRIBUTE_DIRECTORY |
 482                                    FILE_ATTRIBUTE_ENCRYPTED))
 483         {
 484                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 485                       "extracted to standard output", dentry_full_path(dentry));
 486                 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
 487         }
 488
 489         blob = inode_get_blob_for_unnamed_data_stream(inode, blob_table);
 490         if (!blob) {
 491                 const u8 *hash = inode_get_hash_of_unnamed_data_stream(inode);
 492                 if (!is_zero_hash(hash))
 493                         return blob_not_found_error(inode, hash);
 494                 return 0;
 495         }
 496
 497         filedes_init(&_stdout, STDOUT_FILENO);
 498         return extract_blob_to_fd(blob, &_stdout);
 499 }
 500
 501 static int
 502 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
 503                            const struct blob_table *blob_table)
 504 {
 505         for (size_t i = 0; i < num_dentries; i++) {
 506                 int ret = extract_dentry_to_stdout(dentries[i], blob_table);
 507                 if (ret)
 508                         return ret;
 509         }
 510         return 0;
 511 }
 512
 513 /**********************************************************************/
 514
 515 /*
 516  * Removes duplicate dentries from the array.
 517  *
 518  * Returns the new number of dentries, packed at the front of the array.
 519  */
 520 static size_t
 521 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
 522 {
 523         size_t i, j = 0;
 524         for (i = 0; i < num_trees; i++) {
 525                 if (!trees[i]->d_tmp_flag) {
 526                         /* Found distinct dentry.  */
 527                         trees[i]->d_tmp_flag = 1;
 528                         trees[j++] = trees[i];
 529                 }
 530         }
 531         for (i = 0; i < j; i++)
 532                 trees[i]->d_tmp_flag = 0;
 533         return j;
 534 }
 535
 536 /*
 537  * Remove dentries that are descendants of other dentries in the array.
 538  *
 539  * Returns the new number of dentries, packed at the front of the array.
 540  */
 541 static size_t
 542 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
 543 {
 544         size_t i, j = 0;
 545         for (i = 0; i < num_trees; i++)
 546                 trees[i]->d_tmp_flag = 1;
 547         for (i = 0; i < num_trees; i++) {
 548                 struct wim_dentry *d = trees[i];
 549                 while (!dentry_is_root(d)) {
 550                         d = d->d_parent;
 551                         if (d->d_tmp_flag)
 552                                 goto tree_contained;
 553                 }
 554                 trees[j++] = trees[i];
 555                 continue;
 556
 557         tree_contained:
 558                 trees[i]->d_tmp_flag = 0;
 559         }
 560
 561         for (i = 0; i < j; i++)
 562                 trees[i]->d_tmp_flag = 0;
 563         return j;
 564 }
 565
 566 static int
 567 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
 568 {
 569         struct list_head *dentry_list = _dentry_list;
 570         list_add_tail(&dentry->d_extraction_list_node, dentry_list);
 571         return 0;
 572 }
 573
 574 static void
 575 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
 576 {
 577         dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
 578 }
 579
 580 static int
 581 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
 582 {
 583         if (will_extract_dentry(dentry)) {
 584                 list_del(&dentry->d_extraction_list_node);
 585                 dentry_reset_extraction_list_node(dentry);
 586         }
 587         return 0;
 588 }
 589
 590 /*
 591  * Build the preliminary list of dentries to be extracted.
 592  *
 593  * The list maintains the invariant that if d1 and d2 are in the list and d1 is
 594  * an ancestor of d2, then d1 appears before d2 in the list.
 595  */
 596 static void
 597 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
 598                   size_t num_trees, bool add_ancestors)
 599 {
 600         INIT_LIST_HEAD(dentry_list);
 601
 602         /* Add the trees recursively.  */
 603         for (size_t i = 0; i < num_trees; i++)
 604                 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
 605
 606         /* If requested, add ancestors of the trees.  */
 607         if (add_ancestors) {
 608                 for (size_t i = 0; i < num_trees; i++) {
 609                         struct wim_dentry *dentry = trees[i];
 610                         struct wim_dentry *ancestor;
 611                         struct list_head *place_after;
 612
 613                         if (dentry_is_root(dentry))
 614                                 continue;
 615
 616                         place_after = dentry_list;
 617                         ancestor = dentry;
 618                         do {
 619                                 ancestor = ancestor->d_parent;
 620                                 if (will_extract_dentry(ancestor)) {
 621                                         place_after = &ancestor->d_extraction_list_node;
 622                                         break;
 623                                 }
 624                         } while (!dentry_is_root(ancestor));
 625
 626                         ancestor = dentry;
 627                         do {
 628                                 ancestor = ancestor->d_parent;
 629                                 if (will_extract_dentry(ancestor))
 630                                         break;
 631                                 list_add(&ancestor->d_extraction_list_node, place_after);
 632                         } while (!dentry_is_root(ancestor));
 633                 }
 634         }
 635 }
 636
 637 static void
 638 destroy_dentry_list(struct list_head *dentry_list)
 639 {
 640         struct wim_dentry *dentry, *tmp;
 641         struct wim_inode *inode;
 642
 643         list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
 644                 inode = dentry->d_inode;
 645                 dentry_reset_extraction_list_node(dentry);
 646                 inode->i_visited = 0;
 647                 inode->i_can_externally_back = 0;
 648                 if ((void *)dentry->d_extraction_name != (void *)dentry->d_name)
 649                         FREE(dentry->d_extraction_name);
 650                 dentry->d_extraction_name = NULL;
 651                 dentry->d_extraction_name_nchars = 0;
 652         }
 653 }
 654
 655 static void
 656 destroy_blob_list(struct list_head *blob_list)
 657 {
 658         struct blob_descriptor *blob;
 659
 660         list_for_each_entry(blob, blob_list, extraction_list)
 661                 if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets))
 662                         FREE(blob->blob_extraction_targets);
 663 }
 664
 665 #ifdef __WIN32__
 666 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
 667 #else
 668 static const utf16lechar replacement_char = cpu_to_le16('?');
 669 #endif
 670
 671 static bool
 672 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
 673 {
 674         size_t i;
 675
 676         if (num_chars == 0)
 677                 return true;
 678         for (i = 0; i < num_chars; i++) {
 679                 switch (name[i]) {
 680         #ifdef __WIN32__
 681                 case cpu_to_le16('\\'):
 682                 case cpu_to_le16(':'):
 683                 case cpu_to_le16('*'):
 684                 case cpu_to_le16('?'):
 685                 case cpu_to_le16('"'):
 686                 case cpu_to_le16('<'):
 687                 case cpu_to_le16('>'):
 688                 case cpu_to_le16('|'):
 689         #endif
 690                 case cpu_to_le16('/'):
 691                 case cpu_to_le16('\0'):
 692                         if (fix)
 693                                 name[i] = replacement_char;
 694                         else
 695                                 return false;
 696                 }
 697         }
 698
 699 #ifdef __WIN32__
 700         if (name[num_chars - 1] == cpu_to_le16(' ') ||
 701             name[num_chars - 1] == cpu_to_le16('.'))
 702         {
 703                 if (fix)
 704                         name[num_chars - 1] = replacement_char;
 705                 else
 706                         return false;
 707         }
 708 #endif
 709         return true;
 710 }
 711
 712 static int
 713 dentry_calculate_extraction_name(struct wim_dentry *dentry,
 714                                  struct apply_ctx *ctx)
 715 {
 716         int ret;
 717
 718         if (dentry_is_root(dentry))
 719                 return 0;
 720
 721 #ifdef WITH_NTFS_3G
 722         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 723                 dentry->d_extraction_name = dentry->d_name;
 724                 dentry->d_extraction_name_nchars = dentry->d_name_nbytes /
 725                                                    sizeof(utf16lechar);
 726                 return 0;
 727         }
 728 #endif
 729
 730         if (!ctx->supported_features.case_sensitive_filenames) {
 731                 struct wim_dentry *other;
 732                 list_for_each_entry(other, &dentry->d_ci_conflict_list,
 733                                     d_ci_conflict_list)
 734                 {
 735                         if (will_extract_dentry(other)) {
 736                                 if (ctx->extract_flags &
 737                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
 738                                         WARNING("\"%"TS"\" has the same "
 739                                                 "case-insensitive name as "
 740                                                 "\"%"TS"\"; extracting "
 741                                                 "dummy name instead",
 742                                                 dentry_full_path(dentry),
 743                                                 dentry_full_path(other));
 744                                         goto out_replace;
 745                                 } else {
 746                                         WARNING("Not extracting \"%"TS"\": "
 747                                                 "has same case-insensitive "
 748                                                 "name as \"%"TS"\"",
 749                                                 dentry_full_path(dentry),
 750                                                 dentry_full_path(other));
 751                                         goto skip_dentry;
 752                                 }
 753                         }
 754                 }
 755         }
 756
 757         if (file_name_valid(dentry->d_name, dentry->d_name_nbytes / 2, false)) {
 758                 size_t nbytes = 0;
 759                 ret = utf16le_get_tstr(dentry->d_name,
 760                                        dentry->d_name_nbytes,
 761                                        (const tchar **)&dentry->d_extraction_name,
 762                                        &nbytes);
 763                 dentry->d_extraction_name_nchars = nbytes / sizeof(tchar);
 764                 return ret;
 765         } else {
 766                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
 767                 {
 768                         WARNING("\"%"TS"\" has an invalid filename "
 769                                 "that is not supported on this platform; "
 770                                 "extracting dummy name instead",
 771                                 dentry_full_path(dentry));
 772                         goto out_replace;
 773                 } else {
 774                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
 775                                 "that is not supported on this platform",
 776                                 dentry_full_path(dentry));
 777                         goto skip_dentry;
 778                 }
 779         }
 780
 781 out_replace:
 782         {
 783                 utf16lechar utf16_name_copy[dentry->d_name_nbytes / 2];
 784
 785                 memcpy(utf16_name_copy, dentry->d_name, dentry->d_name_nbytes);
 786                 file_name_valid(utf16_name_copy, dentry->d_name_nbytes / 2, true);
 787
 788                 const tchar *tchar_name;
 789                 size_t tchar_nchars;
 790
 791                 ret = utf16le_get_tstr(utf16_name_copy,
 792                                        dentry->d_name_nbytes,
 793                                        &tchar_name, &tchar_nchars);
 794                 if (ret)
 795                         return ret;
 796
 797                 tchar_nchars /= sizeof(tchar);
 798
 799                 size_t fixed_name_num_chars = tchar_nchars;
 800                 tchar fixed_name[tchar_nchars + 50];
 801
 802                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
 803                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
 804                                                  T(" (invalid filename #%lu)"),
 805                                                  ++ctx->invalid_sequence);
 806
 807                 utf16le_put_tstr(tchar_name);
 808
 809                 dentry->d_extraction_name = TSTRDUP(fixed_name);
 810                 if (!dentry->d_extraction_name)
 811                         return WIMLIB_ERR_NOMEM;
 812                 dentry->d_extraction_name_nchars = fixed_name_num_chars;
 813         }
 814         return 0;
 815
 816 skip_dentry:
 817         for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
 818         return 0;
 819 }
 820
 821 /*
 822  * Calculate the actual filename component at which each WIM dentry will be
 823  * extracted, with special handling for dentries that are unsupported by the
 824  * extraction backend or have invalid names.
 825  *
 826  * ctx->supported_features must be filled in.
 827  *
 828  * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
 829  */
 830 static int
 831 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
 832                                        struct apply_ctx *ctx)
 833 {
 834         struct list_head *prev, *cur;
 835
 836         /* Can't use list_for_each_entry() because a call to
 837          * dentry_calculate_extraction_name() may delete the current dentry and
 838          * its children from the list.  */
 839
 840         prev = dentry_list;
 841         for (;;) {
 842                 struct wim_dentry *dentry;
 843                 int ret;
 844
 845                 cur = prev->next;
 846                 if (cur == dentry_list)
 847                         break;
 848
 849                 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
 850
 851                 ret = dentry_calculate_extraction_name(dentry, ctx);
 852                 if (ret)
 853                         return ret;
 854
 855                 if (prev->next == cur)
 856                         prev = cur;
 857                 else
 858                         ; /* Current dentry and its children (which follow in
 859                              the list) were deleted.  prev stays the same.  */
 860         }
 861         return 0;
 862 }
 863
 864 static int
 865 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
 866                        struct blob_table *blob_table)
 867 {
 868         struct wim_inode *inode = dentry->d_inode;
 869         struct blob_descriptor *blob;
 870         int ret;
 871         bool force = false;
 872
 873         /* Special case:  when extracting from a pipe, the WIM blob table is
 874          * initially empty, so "resolving" an inode's streams is initially not
 875          * possible.  However, we still need to keep track of which blobs,
 876          * identified by SHA-1 message digests, need to be extracted, so we
 877          * "resolve" the inode's streams anyway by allocating a 'struct
 878          * blob_descriptor' for each one.  */
 879         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
 880                 force = true;
 881         ret = inode_resolve_streams(inode, blob_table, force);
 882         if (ret)
 883                 return ret;
 884         for (unsigned i = 0; i < inode->i_num_streams; i++) {
 885                 blob = stream_blob_resolved(&inode->i_streams[i]);
 886                 if (blob)
 887                         blob->out_refcnt = 0;
 888         }
 889         return 0;
 890 }
 891
 892 /*
 893  * For each dentry to be extracted, resolve all streams in the corresponding
 894  * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
 895  *
 896  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
 897  */
 898 static int
 899 dentry_list_resolve_streams(struct list_head *dentry_list,
 900                             struct apply_ctx *ctx)
 901 {
 902         struct wim_dentry *dentry;
 903         int ret;
 904
 905         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
 906                 ret = dentry_resolve_streams(dentry,
 907                                              ctx->extract_flags,
 908                                              ctx->wim->blob_table);
 909                 if (ret)
 910                         return ret;
 911         }
 912         return 0;
 913 }
 914
 915 static int
 916 ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
 917            struct apply_ctx *ctx)
 918 {
 919         struct wim_inode *inode = dentry->d_inode;
 920         struct blob_descriptor *blob = stream_blob_resolved(strm);
 921         struct blob_extraction_target *targets;
 922
 923         if (!blob)
 924                 return 0;
 925
 926         /* Tally the size only for each actual extraction of the stream (not
 927          * additional hard links to the inode).  */
 928         if (inode->i_visited && ctx->supported_features.hard_links)
 929                 return 0;
 930
 931         ctx->progress.extract.total_bytes += blob->size;
 932         ctx->progress.extract.total_streams++;
 933
 934         if (inode->i_visited)
 935                 return 0;
 936
 937         /* Add each blob to 'ctx->blob_list' only one time, regardless of how
 938          * many extraction targets it will have.  */
 939         if (blob->out_refcnt == 0) {
 940                 list_add_tail(&blob->extraction_list, &ctx->blob_list);
 941                 ctx->num_blobs_remaining++;
 942         }
 943
 944         /* Set this stream as an extraction target of 'blob'.  */
 945
 946         if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 947                 targets = blob->inline_blob_extraction_targets;
 948         } else {
 949                 struct blob_extraction_target *prev_targets;
 950                 size_t alloc_blob_extraction_targets;
 951
 952                 if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 953                         prev_targets = NULL;
 954                         alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets);
 955                 } else {
 956                         prev_targets = blob->blob_extraction_targets;
 957                         alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
 958                 }
 959
 960                 if (blob->out_refcnt == alloc_blob_extraction_targets) {
 961                         alloc_blob_extraction_targets *= 2;
 962                         targets = REALLOC(prev_targets,
 963                                           alloc_blob_extraction_targets *
 964                                           sizeof(targets[0]));
 965                         if (!targets)
 966                                 return WIMLIB_ERR_NOMEM;
 967                         if (!prev_targets) {
 968                                 memcpy(targets,
 969                                        blob->inline_blob_extraction_targets,
 970                                        sizeof(blob->inline_blob_extraction_targets));
 971                         }
 972                         blob->blob_extraction_targets = targets;
 973                         blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
 974                 }
 975                 targets = blob->blob_extraction_targets;
 976         }
 977         targets[blob->out_refcnt].inode = inode;
 978         targets[blob->out_refcnt].stream = strm;
 979         blob->out_refcnt++;
 980         return 0;
 981 }
 982
 983 static int
 984 ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode,
 985                      struct wim_inode_stream *strm, struct apply_ctx *ctx)
 986 {
 987         bool need_stream = false;
 988         switch (strm->stream_type) {
 989         case STREAM_TYPE_DATA:
 990                 if (stream_is_named(strm)) {
 991                         /* Named data stream  */
 992                         if (ctx->supported_features.named_data_streams)
 993                                 need_stream = true;
 994                 } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
 995                                                     FILE_ATTRIBUTE_ENCRYPTED))
 996                            && !(inode_is_symlink(inode)
 997                                 && !ctx->supported_features.reparse_points
 998                                 && ctx->supported_features.symlink_reparse_points))
 999                 {
1000                         /*
1001                          * Unnamed data stream.  Skip if any of the following is true:
1002                          *
1003                          * - file is a directory
1004                          * - file is encrypted
1005                          * - backend needs to create the file as UNIX symlink
1006                          * - backend will extract the stream as externally backed
1007                          */
1008                         if (ctx->apply_ops->will_externally_back) {
1009                                 int ret = (*ctx->apply_ops->will_externally_back)(dentry, ctx);
1010                                 if (ret > 0) /* Error?  */
1011                                         return ret;
1012                                 if (ret < 0) /* Won't externally back?  */
1013                                         need_stream = true;
1014                         } else {
1015                                 need_stream = true;
1016                         }
1017                 }
1018                 break;
1019         case STREAM_TYPE_REPARSE_POINT:
1020                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT);
1021                 if (ctx->supported_features.reparse_points ||
1022                     (inode_is_symlink(inode) &&
1023                      ctx->supported_features.symlink_reparse_points))
1024                         need_stream = true;
1025                 break;
1026         case STREAM_TYPE_EFSRPC_RAW_DATA:
1027                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED);
1028                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
1029                         if (ctx->supported_features.encrypted_directories)
1030                                 need_stream = true;
1031                 } else {
1032                         if (ctx->supported_features.encrypted_files)
1033                                 need_stream = true;
1034                 }
1035                 break;
1036         }
1037         if (need_stream)
1038                 return ref_stream(strm, dentry, ctx);
1039         return 0;
1040 }
1041
1042 static int
1043 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
1044 {
1045         struct wim_inode *inode = dentry->d_inode;
1046         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1047                 int ret = ref_stream_if_needed(dentry, inode,
1048                                                &inode->i_streams[i], ctx);
1049                 if (ret)
1050                         return ret;
1051         }
1052         inode->i_visited = 1;
1053         return 0;
1054 }
1055
1056 /*
1057  * Given a list of dentries to be extracted, build the list of blobs that need
1058  * to be extracted, and for each blob determine the streams to which that blob
1059  * will be extracted.
1060  *
1061  * This also initializes the extract progress info with byte and blob
1062  * information.
1063  *
1064  * ctx->supported_features must be filled in.
1065  */
1066 static int
1067 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
1068 {
1069         struct wim_dentry *dentry;
1070         int ret;
1071
1072         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1073                 ret = dentry_ref_streams(dentry, ctx);
1074                 if (ret)
1075                         return ret;
1076         }
1077         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1078                 dentry->d_inode->i_visited = 0;
1079         return 0;
1080 }
1081
1082 static void
1083 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
1084 {
1085         struct wim_dentry *dentry;
1086
1087         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1088                 dentry->d_inode->i_first_extraction_alias = NULL;
1089
1090         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1091                 dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias;
1092                 dentry->d_inode->i_first_extraction_alias = dentry;
1093         }
1094 }
1095
1096 static void
1097 inode_tally_features(const struct wim_inode *inode,
1098                      struct wim_features *features)
1099 {
1100         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1101                 features->archive_files++;
1102         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1103                 features->hidden_files++;
1104         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1105                 features->system_files++;
1106         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1107                 features->compressed_files++;
1108         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1109                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1110                         features->encrypted_directories++;
1111                 else
1112                         features->encrypted_files++;
1113         }
1114         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1115                 features->not_context_indexed_files++;
1116         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1117                 features->sparse_files++;
1118         if (inode_has_named_data_stream(inode))
1119                 features->named_data_streams++;
1120         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1121                 features->reparse_points++;
1122                 if (inode_is_symlink(inode))
1123                         features->symlink_reparse_points++;
1124                 else
1125                         features->other_reparse_points++;
1126         }
1127         if (inode_has_security_descriptor(inode))
1128                 features->security_descriptors++;
1129         if (inode_has_unix_data(inode))
1130                 features->unix_data++;
1131 }
1132
1133 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1134 static void
1135 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
1136 {
1137         struct wim_inode *inode = dentry->d_inode;
1138
1139         if (dentry_has_short_name(dentry))
1140                 features->short_names++;
1141
1142         if (inode->i_visited) {
1143                 features->hard_links++;
1144         } else {
1145                 inode_tally_features(inode, features);
1146                 inode->i_visited = 1;
1147         }
1148 }
1149
1150 /* Tally the features necessary to extract the specified dentries.  */
1151 static void
1152 dentry_list_get_features(struct list_head *dentry_list,
1153                          struct wim_features *features)
1154 {
1155         struct wim_dentry *dentry;
1156
1157         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1158                 dentry_tally_features(dentry, features);
1159
1160         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1161                 dentry->d_inode->i_visited = 0;
1162 }
1163
1164 static int
1165 do_feature_check(const struct wim_features *required_features,
1166                  const struct wim_features *supported_features,
1167                  int extract_flags)
1168 {
1169         /* Encrypted files.  */
1170         if (required_features->encrypted_files &&
1171             !supported_features->encrypted_files)
1172                 WARNING("Ignoring EFS-encrypted data of %lu files",
1173                         required_features->encrypted_files);
1174
1175         /* Named data streams.  */
1176         if (required_features->named_data_streams &&
1177             !supported_features->named_data_streams)
1178                 WARNING("Ignoring named data streams of %lu files",
1179                         required_features->named_data_streams);
1180
1181         /* File attributes.  */
1182         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1183                 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1184                  * We're an archive program, so theoretically we can do what we
1185                  * want with it.  */
1186
1187                 if (required_features->hidden_files &&
1188                     !supported_features->hidden_files)
1189                         WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1190                                 required_features->hidden_files);
1191
1192                 if (required_features->system_files &&
1193                     !supported_features->system_files)
1194                         WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1195                                 required_features->system_files);
1196
1197                 if (required_features->compressed_files &&
1198                     !supported_features->compressed_files)
1199                         WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1200                                 required_features->compressed_files);
1201
1202                 if (required_features->not_context_indexed_files &&
1203                     !supported_features->not_context_indexed_files)
1204                         WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1205                                 required_features->not_context_indexed_files);
1206
1207                 if (required_features->sparse_files &&
1208                     !supported_features->sparse_files)
1209                         WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1210                                 required_features->sparse_files);
1211
1212                 if (required_features->encrypted_directories &&
1213                     !supported_features->encrypted_directories)
1214                         WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1215                                 required_features->encrypted_directories);
1216         }
1217
1218         /* Hard links.  */
1219         if (required_features->hard_links && !supported_features->hard_links)
1220                 WARNING("Extracting %lu hard links as independent files",
1221                         required_features->hard_links);
1222
1223         /* Symbolic links and reparse points.  */
1224         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1225             required_features->symlink_reparse_points &&
1226             !supported_features->symlink_reparse_points &&
1227             !supported_features->reparse_points)
1228         {
1229                 ERROR("Extraction backend does not support symbolic links!");
1230                 return WIMLIB_ERR_UNSUPPORTED;
1231         }
1232         if (required_features->reparse_points &&
1233             !supported_features->reparse_points)
1234         {
1235                 if (supported_features->symlink_reparse_points) {
1236                         if (required_features->other_reparse_points) {
1237                                 WARNING("Ignoring reparse data of %lu non-symlink/junction files",
1238                                         required_features->other_reparse_points);
1239                         }
1240                 } else {
1241                         WARNING("Ignoring reparse data of %lu files",
1242                                 required_features->reparse_points);
1243                 }
1244         }
1245
1246         /* Security descriptors.  */
1247         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1248                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1249              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1250             required_features->security_descriptors &&
1251             !supported_features->security_descriptors)
1252         {
1253                 ERROR("Extraction backend does not support security descriptors!");
1254                 return WIMLIB_ERR_UNSUPPORTED;
1255         }
1256         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1257             required_features->security_descriptors &&
1258             !supported_features->security_descriptors)
1259                 WARNING("Ignoring Windows NT security descriptors of %lu files",
1260                         required_features->security_descriptors);
1261
1262         /* UNIX data.  */
1263         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1264             required_features->unix_data && !supported_features->unix_data)
1265         {
1266                 ERROR("Extraction backend does not support UNIX data!");
1267                 return WIMLIB_ERR_UNSUPPORTED;
1268         }
1269
1270         if (required_features->unix_data &&
1271             !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1272         {
1273                 WARNING("Ignoring UNIX metadata of %lu files",
1274                         required_features->unix_data);
1275         }
1276
1277         /* DOS Names.  */
1278         if (required_features->short_names &&
1279             !supported_features->short_names)
1280         {
1281                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1282                         ERROR("Extraction backend does not support DOS names!");
1283                         return WIMLIB_ERR_UNSUPPORTED;
1284                 }
1285                 WARNING("Ignoring DOS names of %lu files",
1286                         required_features->short_names);
1287         }
1288
1289         /* Timestamps.  */
1290         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1291             !supported_features->timestamps)
1292         {
1293                 ERROR("Extraction backend does not support timestamps!");
1294                 return WIMLIB_ERR_UNSUPPORTED;
1295         }
1296
1297         return 0;
1298 }
1299
1300 static const struct apply_operations *
1301 select_apply_operations(int extract_flags)
1302 {
1303 #ifdef WITH_NTFS_3G
1304         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1305                 return &ntfs_3g_apply_ops;
1306 #endif
1307 #ifdef __WIN32__
1308         return &win32_apply_ops;
1309 #else
1310         return &unix_apply_ops;
1311 #endif
1312 }
1313
1314 static int
1315 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1316               const tchar *target, int extract_flags)
1317 {
1318         const struct apply_operations *ops;
1319         struct apply_ctx *ctx;
1320         int ret;
1321         LIST_HEAD(dentry_list);
1322
1323         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1324                 ret = extract_dentries_to_stdout(trees, num_trees,
1325                                                  wim->blob_table);
1326                 goto out;
1327         }
1328
1329         num_trees = remove_duplicate_trees(trees, num_trees);
1330         num_trees = remove_contained_trees(trees, num_trees);
1331
1332         ops = select_apply_operations(extract_flags);
1333
1334         if (num_trees > 1 && ops->single_tree_only) {
1335                 ERROR("Extracting multiple directory trees "
1336                       "at once is not supported in %s extraction mode!",
1337                       ops->name);
1338                 ret = WIMLIB_ERR_UNSUPPORTED;
1339                 goto out;
1340         }
1341
1342         ctx = CALLOC(1, ops->context_size);
1343         if (!ctx) {
1344                 ret = WIMLIB_ERR_NOMEM;
1345                 goto out;
1346         }
1347
1348         ctx->wim = wim;
1349         ctx->target = target;
1350         ctx->target_nchars = tstrlen(target);
1351         ctx->extract_flags = extract_flags;
1352         if (ctx->wim->progfunc) {
1353                 ctx->progfunc = ctx->wim->progfunc;
1354                 ctx->progctx = ctx->wim->progctx;
1355                 ctx->progress.extract.image = wim->current_image;
1356                 ctx->progress.extract.extract_flags = (extract_flags &
1357                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
1358                 ctx->progress.extract.wimfile_name = wim->filename;
1359                 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1360                                                                          wim->current_image);
1361                 ctx->progress.extract.target = target;
1362         }
1363         INIT_LIST_HEAD(&ctx->blob_list);
1364         filedes_invalidate(&ctx->tmpfile_fd);
1365         ctx->apply_ops = ops;
1366
1367         ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1368         if (ret)
1369                 goto out_cleanup;
1370
1371         build_dentry_list(&dentry_list, trees, num_trees,
1372                           !(extract_flags &
1373                             WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1374
1375         dentry_list_get_features(&dentry_list, &ctx->required_features);
1376
1377         ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1378                                ctx->extract_flags);
1379         if (ret)
1380                 goto out_cleanup;
1381
1382         ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1383         if (ret)
1384                 goto out_cleanup;
1385
1386         if (unlikely(list_empty(&dentry_list))) {
1387                 WARNING("There is nothing to extract!");
1388                 goto out_cleanup;
1389         }
1390
1391         ret = dentry_list_resolve_streams(&dentry_list, ctx);
1392         if (ret)
1393                 goto out_cleanup;
1394
1395         dentry_list_build_inode_alias_lists(&dentry_list);
1396
1397         ret = dentry_list_ref_streams(&dentry_list, ctx);
1398         if (ret)
1399                 goto out_cleanup;
1400
1401         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1402                 /* When extracting from a pipe, the number of bytes of data to
1403                  * extract can't be determined in the normal way (examining the
1404                  * blob table), since at this point all we have is a set of
1405                  * SHA-1 message digests of blobs that need to be extracted.
1406                  * However, we can get a reasonably accurate estimate by taking
1407                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1408                  * data.  This does assume that a full image is being extracted,
1409                  * but currently there is no API for doing otherwise.  (Also,
1410                  * subtract <HARDLINKBYTES> from this if hard links are
1411                  * supported by the extraction mode.)  */
1412                 ctx->progress.extract.total_bytes =
1413                         wim_info_get_image_total_bytes(wim->wim_info,
1414                                                        wim->current_image);
1415                 if (ctx->supported_features.hard_links) {
1416                         ctx->progress.extract.total_bytes -=
1417                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
1418                                                                    wim->current_image);
1419                 }
1420         }
1421
1422         ret = extract_progress(ctx,
1423                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1424                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
1425                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
1426         if (ret)
1427                 goto out_cleanup;
1428
1429         ret = (*ops->extract)(&dentry_list, ctx);
1430         if (ret)
1431                 goto out_cleanup;
1432
1433         if (ctx->progress.extract.completed_bytes <
1434             ctx->progress.extract.total_bytes)
1435         {
1436                 ctx->progress.extract.completed_bytes =
1437                         ctx->progress.extract.total_bytes;
1438                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
1439                 if (ret)
1440                         goto out_cleanup;
1441         }
1442
1443         ret = extract_progress(ctx,
1444                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1445                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
1446                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
1447 out_cleanup:
1448         destroy_blob_list(&ctx->blob_list);
1449         destroy_dentry_list(&dentry_list);
1450         FREE(ctx);
1451 out:
1452         return ret;
1453 }
1454
1455 static int
1456 mkdir_if_needed(const tchar *target)
1457 {
1458         if (!tmkdir(target, 0755))
1459                 return 0;
1460
1461         if (errno == EEXIST)
1462                 return 0;
1463
1464 #ifdef __WIN32__
1465         /* _wmkdir() fails with EACCES if called on a drive root directory.  */
1466         if (errno == EACCES)
1467                 return 0;
1468 #endif
1469
1470         ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
1471         return WIMLIB_ERR_MKDIR;
1472 }
1473
1474 /* Make sure the extraction flags make sense, and update them if needed.  */
1475 static int
1476 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1477 {
1478         int extract_flags = *extract_flags_p;
1479
1480         /* Check for invalid flag combinations  */
1481
1482         if ((extract_flags &
1483              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1484               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1485                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1486                 return WIMLIB_ERR_INVALID_PARAM;
1487
1488         if ((extract_flags &
1489              (WIMLIB_EXTRACT_FLAG_RPFIX |
1490               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1491                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
1492                 return WIMLIB_ERR_INVALID_PARAM;
1493
1494 #ifndef WITH_NTFS_3G
1495         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1496                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1497                       "        it cannot apply a WIM image directly to an NTFS volume.");
1498                 return WIMLIB_ERR_UNSUPPORTED;
1499         }
1500 #endif
1501
1502         if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1503 #ifdef __WIN32__
1504                 if (!wim->filename)
1505                         return WIMLIB_ERR_NO_FILENAME;
1506 #else
1507                 ERROR("WIMBoot extraction is only supported on Windows!");
1508                 return WIMLIB_ERR_UNSUPPORTED;
1509 #endif
1510         }
1511
1512
1513         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1514                               WIMLIB_EXTRACT_FLAG_NORPFIX |
1515                               WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1516                                         WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1517         {
1518                 /* For full-image extraction, do reparse point fixups by default
1519                  * if the WIM header says they are enabled.  */
1520                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1521                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1522         }
1523
1524         *extract_flags_p = extract_flags;
1525         return 0;
1526 }
1527
1528 struct append_dentry_ctx {
1529         struct wim_dentry **dentries;
1530         size_t num_dentries;
1531         size_t num_alloc_dentries;
1532 };
1533
1534 static int
1535 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1536 {
1537         struct append_dentry_ctx *ctx = _ctx;
1538
1539         if (ctx->num_dentries == ctx->num_alloc_dentries) {
1540                 struct wim_dentry **new_dentries;
1541                 size_t new_length;
1542
1543                 new_length = max(ctx->num_alloc_dentries + 8,
1544                                  ctx->num_alloc_dentries * 3 / 2);
1545                 new_dentries = REALLOC(ctx->dentries,
1546                                        new_length * sizeof(ctx->dentries[0]));
1547                 if (new_dentries == NULL)
1548                         return WIMLIB_ERR_NOMEM;
1549                 ctx->dentries = new_dentries;
1550                 ctx->num_alloc_dentries = new_length;
1551         }
1552         ctx->dentries[ctx->num_dentries++] = dentry;
1553         return 0;
1554 }
1555
1556 /* Append dentries matched by a path which can contain wildcard characters.  */
1557 static int
1558 append_matched_dentries(WIMStruct *wim, const tchar *orig_pattern,
1559                         int extract_flags, struct append_dentry_ctx *ctx)
1560 {
1561         const size_t count_before = ctx->num_dentries;
1562         tchar *pattern;
1563         int ret;
1564
1565         pattern = canonicalize_wim_path(orig_pattern);
1566         if (!pattern)
1567                 return WIMLIB_ERR_NOMEM;
1568         ret = expand_path_pattern(wim_get_current_root_dentry(wim), pattern,
1569                                   append_dentry_cb, ctx);
1570         FREE(pattern);
1571         if (ret || ctx->num_dentries > count_before)
1572                 return ret;
1573         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB) {
1574                 ERROR("No matches for path pattern \"%"TS"\"", orig_pattern);
1575                 return WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1576         }
1577         WARNING("No matches for path pattern \"%"TS"\"", orig_pattern);
1578         return 0;
1579 }
1580
1581 static int
1582 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1583                         const tchar * const *paths, size_t num_paths,
1584                         int extract_flags)
1585 {
1586         int ret;
1587         struct wim_dentry **trees;
1588         size_t num_trees;
1589
1590         if (wim == NULL || target == NULL || target[0] == T('\0') ||
1591             (num_paths != 0 && paths == NULL))
1592                 return WIMLIB_ERR_INVALID_PARAM;
1593
1594         ret = check_extract_flags(wim, &extract_flags);
1595         if (ret)
1596                 return ret;
1597
1598         ret = select_wim_image(wim, image);
1599         if (ret)
1600                 return ret;
1601
1602         ret = wim_checksum_unhashed_blobs(wim);
1603         if (ret)
1604                 return ret;
1605
1606         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1607                               WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1608             (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1609         {
1610                 ret = mkdir_if_needed(target);
1611                 if (ret)
1612                         return ret;
1613         }
1614
1615         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1616
1617                 struct append_dentry_ctx append_dentry_ctx = {
1618                         .dentries = NULL,
1619                         .num_dentries = 0,
1620                         .num_alloc_dentries = 0,
1621                 };
1622
1623                 for (size_t i = 0; i < num_paths; i++) {
1624                         ret = append_matched_dentries(wim, paths[i],
1625                                                       extract_flags,
1626                                                       &append_dentry_ctx);
1627                         if (ret) {
1628                                 trees = append_dentry_ctx.dentries;
1629                                 goto out_free_trees;
1630                         }
1631                 }
1632                 trees = append_dentry_ctx.dentries;
1633                 num_trees = append_dentry_ctx.num_dentries;
1634         } else {
1635                 trees = MALLOC(num_paths * sizeof(trees[0]));
1636                 if (trees == NULL)
1637                         return WIMLIB_ERR_NOMEM;
1638
1639                 for (size_t i = 0; i < num_paths; i++) {
1640
1641                         tchar *path = canonicalize_wim_path(paths[i]);
1642                         if (path == NULL) {
1643                                 ret = WIMLIB_ERR_NOMEM;
1644                                 goto out_free_trees;
1645                         }
1646
1647                         trees[i] = get_dentry(wim, path,
1648                                               WIMLIB_CASE_PLATFORM_DEFAULT);
1649                         FREE(path);
1650                         if (trees[i] == NULL) {
1651                                   ERROR("Path \"%"TS"\" does not exist "
1652                                         "in WIM image %d",
1653                                         paths[i], wim->current_image);
1654                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1655                                   goto out_free_trees;
1656                         }
1657                 }
1658                 num_trees = num_paths;
1659         }
1660
1661         if (num_trees == 0) {
1662                 ret = 0;
1663                 goto out_free_trees;
1664         }
1665
1666         ret = extract_trees(wim, trees, num_trees, target, extract_flags);
1667 out_free_trees:
1668         FREE(trees);
1669         return ret;
1670 }
1671
1672 static int
1673 extract_single_image(WIMStruct *wim, int image,
1674                      const tchar *target, int extract_flags)
1675 {
1676         const tchar *path = WIMLIB_WIM_ROOT_PATH;
1677         extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1678         return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
1679 }
1680
1681 static const tchar * const filename_forbidden_chars =
1682 T(
1683 #ifdef __WIN32__
1684 "<>:\"/\\|?*"
1685 #else
1686 "/"
1687 #endif
1688 );
1689
1690 /* This function checks if it is okay to use a WIM image's name as a directory
1691  * name.  */
1692 static bool
1693 image_name_ok_as_dir(const tchar *image_name)
1694 {
1695         return image_name && *image_name &&
1696                 !tstrpbrk(image_name, filename_forbidden_chars) &&
1697                 tstrcmp(image_name, T(".")) &&
1698                 tstrcmp(image_name, T(".."));
1699 }
1700
1701 /* Extracts all images from the WIM to the directory @target, with the images
1702  * placed in subdirectories named by their image names. */
1703 static int
1704 extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
1705 {
1706         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
1707         size_t output_path_len = tstrlen(target);
1708         tchar buf[output_path_len + 1 + image_name_max_len + 1];
1709         int ret;
1710         int image;
1711         const tchar *image_name;
1712
1713         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1714                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1715                 return WIMLIB_ERR_INVALID_PARAM;
1716         }
1717
1718         ret = mkdir_if_needed(target);
1719         if (ret)
1720                 return ret;
1721         tmemcpy(buf, target, output_path_len);
1722         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1723         for (image = 1; image <= wim->hdr.image_count; image++) {
1724                 image_name = wimlib_get_image_name(wim, image);
1725                 if (image_name_ok_as_dir(image_name)) {
1726                         tstrcpy(buf + output_path_len + 1, image_name);
1727                 } else {
1728                         /* Image name is empty or contains forbidden characters.
1729                          * Use image number instead. */
1730                         tsprintf(buf + output_path_len + 1, T("%d"), image);
1731                 }
1732                 ret = extract_single_image(wim, image, buf, extract_flags);
1733                 if (ret)
1734                         return ret;
1735         }
1736         return 0;
1737 }
1738
1739 static int
1740 do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1741                         int extract_flags)
1742 {
1743         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1744                              WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1745                              WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1746                 return WIMLIB_ERR_INVALID_PARAM;
1747
1748         if (image == WIMLIB_ALL_IMAGES)
1749                 return extract_all_images(wim, target, extract_flags);
1750         else
1751                 return extract_single_image(wim, image, target, extract_flags);
1752 }
1753
1754
1755 /****************************************************************************
1756  *                          Extraction API                                  *
1757  ****************************************************************************/
1758
1759 WIMLIBAPI int
1760 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1761                      const tchar * const *paths, size_t num_paths,
1762                      int extract_flags)
1763 {
1764         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1765                 return WIMLIB_ERR_INVALID_PARAM;
1766
1767         return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1768                                        extract_flags);
1769 }
1770
1771 WIMLIBAPI int
1772 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1773                         const tchar *path_list_file, int extract_flags)
1774 {
1775         int ret;
1776         tchar **paths;
1777         size_t num_paths;
1778         void *mem;
1779
1780         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1781         if (ret) {
1782                 ERROR("Failed to read path list file \"%"TS"\"",
1783                       path_list_file);
1784                 return ret;
1785         }
1786
1787         ret = wimlib_extract_paths(wim, image, target,
1788                                    (const tchar * const *)paths, num_paths,
1789                                    extract_flags);
1790         FREE(paths);
1791         FREE(mem);
1792         return ret;
1793 }
1794
1795 WIMLIBAPI int
1796 wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
1797                                              const tchar *image_num_or_name,
1798                                              const tchar *target,
1799                                              int extract_flags,
1800                                              wimlib_progress_func_t progfunc,
1801                                              void *progctx)
1802 {
1803         int ret;
1804         WIMStruct *pwm;
1805         struct filedes *in_fd;
1806         int image;
1807         unsigned i;
1808
1809         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1810                 return WIMLIB_ERR_INVALID_PARAM;
1811
1812         /* Read the WIM header from the pipe and get a WIMStruct to represent
1813          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
1814          * wimlib_open_wim(), getting a WIMStruct in this way will result in an
1815          * empty blob table, no XML data read, and no filename set.  */
1816         ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
1817                                     progfunc, progctx);
1818         if (ret)
1819                 return ret;
1820
1821         /* Sanity check to make sure this is a pipable WIM.  */
1822         if (pwm->hdr.magic != PWM_MAGIC) {
1823                 ERROR("The WIM being read from file descriptor %d "
1824                       "is not pipable!", pipe_fd);
1825                 ret = WIMLIB_ERR_NOT_PIPABLE;
1826                 goto out_wimlib_free;
1827         }
1828
1829         /* Sanity check to make sure the first part of a pipable split WIM is
1830          * sent over the pipe first.  */
1831         if (pwm->hdr.part_number != 1) {
1832                 ERROR("The first part of the split WIM must be "
1833                       "sent over the pipe first.");
1834                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1835                 goto out_wimlib_free;
1836         }
1837
1838         in_fd = &pwm->in_fd;
1839         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1840
1841         /* As mentioned, the WIMStruct we created from the pipe does not have
1842          * XML data yet.  Fix this by reading the extra copy of the XML data
1843          * that directly follows the header in pipable WIMs.  (Note: see
1844          * write_pipable_wim() for more details about the format of pipable
1845          * WIMs.)  */
1846         {
1847                 u8 hash[SHA1_HASH_SIZE];
1848
1849                 ret = read_pwm_blob_header(pwm, hash,
1850                                            &pwm->hdr.xml_data_reshdr, NULL);
1851                 if (ret)
1852                         goto out_wimlib_free;
1853
1854                 if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1855                         ERROR("Expected XML data, but found non-metadata resource.");
1856                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1857                         goto out_wimlib_free;
1858                 }
1859
1860                 ret = read_wim_xml_data(pwm);
1861                 if (ret)
1862                         goto out_wimlib_free;
1863
1864                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
1865                         ERROR("Image count in XML data is not the same as in WIM header.");
1866                         ret = WIMLIB_ERR_IMAGE_COUNT;
1867                         goto out_wimlib_free;
1868                 }
1869         }
1870
1871         /* Get image index (this may use the XML data that was just read to
1872          * resolve an image name).  */
1873         if (image_num_or_name) {
1874                 image = wimlib_resolve_image(pwm, image_num_or_name);
1875                 if (image == WIMLIB_NO_IMAGE) {
1876                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1877                               image_num_or_name);
1878                         ret = WIMLIB_ERR_INVALID_IMAGE;
1879                         goto out_wimlib_free;
1880                 } else if (image == WIMLIB_ALL_IMAGES) {
1881                         ERROR("Applying all images from a pipe is not supported!");
1882                         ret = WIMLIB_ERR_INVALID_IMAGE;
1883                         goto out_wimlib_free;
1884                 }
1885         } else {
1886                 if (pwm->hdr.image_count != 1) {
1887                         ERROR("No image was specified, but the pipable WIM "
1888                               "did not contain exactly 1 image");
1889                         ret = WIMLIB_ERR_INVALID_IMAGE;
1890                         goto out_wimlib_free;
1891                 }
1892                 image = 1;
1893         }
1894
1895         /* Load the needed metadata resource.  */
1896         for (i = 1; i <= pwm->hdr.image_count; i++) {
1897                 struct wim_image_metadata *imd;
1898                 struct wim_reshdr reshdr;
1899                 struct wim_resource_descriptor *metadata_rdesc;
1900
1901                 imd = pwm->image_metadata[i - 1];
1902
1903                 ret = WIMLIB_ERR_NOMEM;
1904                 imd->metadata_blob = new_blob_descriptor();
1905                 if (!imd->metadata_blob)
1906                         goto out_wimlib_free;
1907
1908                 imd->metadata_blob->is_metadata = 1;
1909
1910                 ret = read_pwm_blob_header(pwm, imd->metadata_blob->hash,
1911                                            &reshdr, NULL);
1912                 if (ret)
1913                         goto out_wimlib_free;
1914
1915                 if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1916                         ERROR("Expected metadata resource, but found "
1917                               "non-metadata resource");
1918                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1919                         goto out_wimlib_free;
1920                 }
1921
1922                 ret = WIMLIB_ERR_NOMEM;
1923                 metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
1924                 if (!metadata_rdesc)
1925                         goto out_wimlib_free;
1926                 wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc,
1927                                             imd->metadata_blob);
1928
1929                 if (i == image) {
1930                         /* Metadata resource is for the image being extracted.
1931                          * Parse it and save the metadata in memory.  */
1932                         ret = read_metadata_resource(imd);
1933                         if (ret)
1934                                 goto out_wimlib_free;
1935                         imd->modified = 1;
1936                 } else {
1937                         /* Metadata resource is not for the image being
1938                          * extracted.  Skip over it.  */
1939                         ret = skip_wim_resource(metadata_rdesc);
1940                         if (ret)
1941                                 goto out_wimlib_free;
1942                 }
1943         }
1944         /* Extract the image.  */
1945         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
1946         ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
1947         /* Clean up and return.  */
1948 out_wimlib_free:
1949         wimlib_free(pwm);
1950         return ret;
1951 }
1952
1953
1954 WIMLIBAPI int
1955 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
1956                                const tchar *target, int extract_flags)
1957 {
1958         return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
1959                                                             image_num_or_name,
1960                                                             target,
1961                                                             extract_flags,
1962                                                             NULL,
1963                                                             NULL);
1964 }
1965
1966 WIMLIBAPI int
1967 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1968                      int extract_flags)
1969 {
1970         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1971                 return WIMLIB_ERR_INVALID_PARAM;
1972         return do_wimlib_extract_image(wim, image, target, extract_flags);
1973 }