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/reparse.h"
  58 #include "wimlib/resource.h"
  59 #include "wimlib/security.h"
  60 #include "wimlib/unix_data.h"
  61 #include "wimlib/wildcard.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]->tmp_flag) {
 526                         /* Found distinct dentry.  */
 527                         trees[i]->tmp_flag = 1;
 528                         trees[j++] = trees[i];
 529                 }
 530         }
 531         for (i = 0; i < j; i++)
 532                 trees[i]->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]->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->tmp_flag)
 552                                 goto tree_contained;
 553                 }
 554                 trees[j++] = trees[i];
 555                 continue;
 556
 557         tree_contained:
 558                 trees[i]->tmp_flag = 0;
 559         }
 560
 561         for (i = 0; i < j; i++)
 562                 trees[i]->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         list_del(&dentry->d_extraction_list_node);
 584         dentry_reset_extraction_list_node(dentry);
 585         return 0;
 586 }
 587
 588 /*
 589  * Build the preliminary list of dentries to be extracted.
 590  *
 591  * The list maintains the invariant that if d1 and d2 are in the list and d1 is
 592  * an ancestor of d2, then d1 appears before d2 in the list.
 593  */
 594 static void
 595 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
 596                   size_t num_trees, bool add_ancestors)
 597 {
 598         INIT_LIST_HEAD(dentry_list);
 599
 600         /* Add the trees recursively.  */
 601         for (size_t i = 0; i < num_trees; i++)
 602                 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
 603
 604         /* If requested, add ancestors of the trees.  */
 605         if (add_ancestors) {
 606                 for (size_t i = 0; i < num_trees; i++) {
 607                         struct wim_dentry *dentry = trees[i];
 608                         struct wim_dentry *ancestor;
 609                         struct list_head *place_after;
 610
 611                         if (dentry_is_root(dentry))
 612                                 continue;
 613
 614                         place_after = dentry_list;
 615                         ancestor = dentry;
 616                         do {
 617                                 ancestor = ancestor->d_parent;
 618                                 if (will_extract_dentry(ancestor)) {
 619                                         place_after = &ancestor->d_extraction_list_node;
 620                                         break;
 621                                 }
 622                         } while (!dentry_is_root(ancestor));
 623
 624                         ancestor = dentry;
 625                         do {
 626                                 ancestor = ancestor->d_parent;
 627                                 if (will_extract_dentry(ancestor))
 628                                         break;
 629                                 list_add(&ancestor->d_extraction_list_node, place_after);
 630                         } while (!dentry_is_root(ancestor));
 631                 }
 632         }
 633 }
 634
 635 static void
 636 destroy_dentry_list(struct list_head *dentry_list)
 637 {
 638         struct wim_dentry *dentry, *tmp;
 639         struct wim_inode *inode;
 640
 641         list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
 642                 inode = dentry->d_inode;
 643                 dentry_reset_extraction_list_node(dentry);
 644                 inode->i_visited = 0;
 645                 inode->i_can_externally_back = 0;
 646                 if ((void *)dentry->d_extraction_name != (void *)dentry->file_name)
 647                         FREE(dentry->d_extraction_name);
 648                 dentry->d_extraction_name = NULL;
 649                 dentry->d_extraction_name_nchars = 0;
 650         }
 651 }
 652
 653 static void
 654 destroy_blob_list(struct list_head *blob_list)
 655 {
 656         struct blob_descriptor *blob;
 657
 658         list_for_each_entry(blob, blob_list, extraction_list)
 659                 if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets))
 660                         FREE(blob->blob_extraction_targets);
 661 }
 662
 663 #ifdef __WIN32__
 664 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
 665 #else
 666 static const utf16lechar replacement_char = cpu_to_le16('?');
 667 #endif
 668
 669 static bool
 670 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
 671 {
 672         size_t i;
 673
 674         if (num_chars == 0)
 675                 return true;
 676         for (i = 0; i < num_chars; i++) {
 677                 switch (name[i]) {
 678         #ifdef __WIN32__
 679                 case cpu_to_le16('\\'):
 680                 case cpu_to_le16(':'):
 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         #endif
 688                 case cpu_to_le16('/'):
 689                 case cpu_to_le16('\0'):
 690                         if (fix)
 691                                 name[i] = replacement_char;
 692                         else
 693                                 return false;
 694                 }
 695         }
 696
 697 #ifdef __WIN32__
 698         if (name[num_chars - 1] == cpu_to_le16(' ') ||
 699             name[num_chars - 1] == cpu_to_le16('.'))
 700         {
 701                 if (fix)
 702                         name[num_chars - 1] = replacement_char;
 703                 else
 704                         return false;
 705         }
 706 #endif
 707         return true;
 708 }
 709
 710 static int
 711 dentry_calculate_extraction_name(struct wim_dentry *dentry,
 712                                  struct apply_ctx *ctx)
 713 {
 714         int ret;
 715
 716         if (dentry_is_root(dentry))
 717                 return 0;
 718
 719 #ifdef WITH_NTFS_3G
 720         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 721                 dentry->d_extraction_name = dentry->file_name;
 722                 dentry->d_extraction_name_nchars = dentry->file_name_nbytes /
 723                                                    sizeof(utf16lechar);
 724                 return 0;
 725         }
 726 #endif
 727
 728         if (!ctx->supported_features.case_sensitive_filenames) {
 729                 struct wim_dentry *other;
 730                 list_for_each_entry(other, &dentry->d_ci_conflict_list,
 731                                     d_ci_conflict_list)
 732                 {
 733                         if (will_extract_dentry(other)) {
 734                                 if (ctx->extract_flags &
 735                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
 736                                         WARNING("\"%"TS"\" has the same "
 737                                                 "case-insensitive name as "
 738                                                 "\"%"TS"\"; extracting "
 739                                                 "dummy name instead",
 740                                                 dentry_full_path(dentry),
 741                                                 dentry_full_path(other));
 742                                         goto out_replace;
 743                                 } else {
 744                                         WARNING("Not extracting \"%"TS"\": "
 745                                                 "has same case-insensitive "
 746                                                 "name as \"%"TS"\"",
 747                                                 dentry_full_path(dentry),
 748                                                 dentry_full_path(other));
 749                                         goto skip_dentry;
 750                                 }
 751                         }
 752                 }
 753         }
 754
 755         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
 756                 ret = utf16le_get_tstr(dentry->file_name,
 757                                        dentry->file_name_nbytes,
 758                                        (const tchar **)&dentry->d_extraction_name,
 759                                        &dentry->d_extraction_name_nchars);
 760                 dentry->d_extraction_name_nchars /= sizeof(tchar);
 761                 return ret;
 762         } else {
 763                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
 764                 {
 765                         WARNING("\"%"TS"\" has an invalid filename "
 766                                 "that is not supported on this platform; "
 767                                 "extracting dummy name instead",
 768                                 dentry_full_path(dentry));
 769                         goto out_replace;
 770                 } else {
 771                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
 772                                 "that is not supported on this platform",
 773                                 dentry_full_path(dentry));
 774                         goto skip_dentry;
 775                 }
 776         }
 777
 778 out_replace:
 779         {
 780                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
 781
 782                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
 783                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
 784
 785                 const tchar *tchar_name;
 786                 size_t tchar_nchars;
 787
 788                 ret = utf16le_get_tstr(utf16_name_copy,
 789                                        dentry->file_name_nbytes,
 790                                        &tchar_name, &tchar_nchars);
 791                 if (ret)
 792                         return ret;
 793
 794                 tchar_nchars /= sizeof(tchar);
 795
 796                 size_t fixed_name_num_chars = tchar_nchars;
 797                 tchar fixed_name[tchar_nchars + 50];
 798
 799                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
 800                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
 801                                                  T(" (invalid filename #%lu)"),
 802                                                  ++ctx->invalid_sequence);
 803
 804                 utf16le_put_tstr(tchar_name);
 805
 806                 dentry->d_extraction_name = TSTRDUP(fixed_name);
 807                 if (!dentry->d_extraction_name)
 808                         return WIMLIB_ERR_NOMEM;
 809                 dentry->d_extraction_name_nchars = fixed_name_num_chars;
 810         }
 811         return 0;
 812
 813 skip_dentry:
 814         for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
 815         return 0;
 816 }
 817
 818 /*
 819  * Calculate the actual filename component at which each WIM dentry will be
 820  * extracted, with special handling for dentries that are unsupported by the
 821  * extraction backend or have invalid names.
 822  *
 823  * ctx->supported_features must be filled in.
 824  *
 825  * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
 826  */
 827 static int
 828 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
 829                                        struct apply_ctx *ctx)
 830 {
 831         struct list_head *prev, *cur;
 832
 833         /* Can't use list_for_each_entry() because a call to
 834          * dentry_calculate_extraction_name() may delete the current dentry and
 835          * its children from the list.  */
 836
 837         prev = dentry_list;
 838         for (;;) {
 839                 struct wim_dentry *dentry;
 840                 int ret;
 841
 842                 cur = prev->next;
 843                 if (cur == dentry_list)
 844                         break;
 845
 846                 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
 847
 848                 ret = dentry_calculate_extraction_name(dentry, ctx);
 849                 if (ret)
 850                         return ret;
 851
 852                 if (prev->next == cur)
 853                         prev = cur;
 854                 else
 855                         ; /* Current dentry and its children (which follow in
 856                              the list) were deleted.  prev stays the same.  */
 857         }
 858         return 0;
 859 }
 860
 861 static int
 862 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
 863                        struct blob_table *blob_table)
 864 {
 865         struct wim_inode *inode = dentry->d_inode;
 866         struct blob_descriptor *blob;
 867         int ret;
 868         bool force = false;
 869
 870         /* Special case:  when extracting from a pipe, the WIM blob table is
 871          * initially empty, so "resolving" an inode's streams is initially not
 872          * possible.  However, we still need to keep track of which blobs,
 873          * identified by SHA-1 message digests, need to be extracted, so we
 874          * "resolve" the inode's streams anyway by allocating a 'struct
 875          * blob_descriptor' for each one.  */
 876         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
 877                 force = true;
 878         ret = inode_resolve_streams(inode, blob_table, force);
 879         if (ret)
 880                 return ret;
 881         for (unsigned i = 0; i < inode->i_num_streams; i++) {
 882                 blob = stream_blob_resolved(&inode->i_streams[i]);
 883                 if (blob)
 884                         blob->out_refcnt = 0;
 885         }
 886         return 0;
 887 }
 888
 889 /*
 890  * For each dentry to be extracted, resolve all streams in the corresponding
 891  * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
 892  *
 893  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
 894  */
 895 static int
 896 dentry_list_resolve_streams(struct list_head *dentry_list,
 897                             struct apply_ctx *ctx)
 898 {
 899         struct wim_dentry *dentry;
 900         int ret;
 901
 902         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
 903                 ret = dentry_resolve_streams(dentry,
 904                                              ctx->extract_flags,
 905                                              ctx->wim->blob_table);
 906                 if (ret)
 907                         return ret;
 908         }
 909         return 0;
 910 }
 911
 912 static int
 913 ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
 914            struct apply_ctx *ctx)
 915 {
 916         struct wim_inode *inode = dentry->d_inode;
 917         struct blob_descriptor *blob = stream_blob_resolved(strm);
 918         struct blob_extraction_target *targets;
 919
 920         if (!blob)
 921                 return 0;
 922
 923         /* Tally the size only for each actual extraction of the stream (not
 924          * additional hard links to the inode).  */
 925         if (inode->i_visited && ctx->supported_features.hard_links)
 926                 return 0;
 927
 928         ctx->progress.extract.total_bytes += blob->size;
 929         ctx->progress.extract.total_streams++;
 930
 931         if (inode->i_visited)
 932                 return 0;
 933
 934         /* Add each blob to 'ctx->blob_list' only one time, regardless of how
 935          * many extraction targets it will have.  */
 936         if (blob->out_refcnt == 0) {
 937                 list_add_tail(&blob->extraction_list, &ctx->blob_list);
 938                 ctx->num_blobs_remaining++;
 939         }
 940
 941         /* Set this stream as an extraction target of 'blob'.  */
 942
 943         if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 944                 targets = blob->inline_blob_extraction_targets;
 945         } else {
 946                 struct blob_extraction_target *prev_targets;
 947                 size_t alloc_blob_extraction_targets;
 948
 949                 if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) {
 950                         prev_targets = NULL;
 951                         alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets);
 952                 } else {
 953                         prev_targets = blob->blob_extraction_targets;
 954                         alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
 955                 }
 956
 957                 if (blob->out_refcnt == alloc_blob_extraction_targets) {
 958                         alloc_blob_extraction_targets *= 2;
 959                         targets = REALLOC(prev_targets,
 960                                           alloc_blob_extraction_targets *
 961                                           sizeof(targets[0]));
 962                         if (!targets)
 963                                 return WIMLIB_ERR_NOMEM;
 964                         if (!prev_targets) {
 965                                 memcpy(targets,
 966                                        blob->inline_blob_extraction_targets,
 967                                        sizeof(blob->inline_blob_extraction_targets));
 968                         }
 969                         blob->blob_extraction_targets = targets;
 970                         blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
 971                 }
 972                 targets = blob->blob_extraction_targets;
 973         }
 974         targets[blob->out_refcnt].inode = inode;
 975         targets[blob->out_refcnt].stream = strm;
 976         blob->out_refcnt++;
 977         return 0;
 978 }
 979
 980 static int
 981 ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode,
 982                      struct wim_inode_stream *strm, struct apply_ctx *ctx)
 983 {
 984         bool need_stream = false;
 985         switch (strm->stream_type) {
 986         case STREAM_TYPE_DATA:
 987                 if (stream_is_named(strm)) {
 988                         /* Named data stream  */
 989                         if (ctx->supported_features.named_data_streams)
 990                                 need_stream = true;
 991                 } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
 992                                                     FILE_ATTRIBUTE_ENCRYPTED))
 993                            && !(inode_is_symlink(inode)
 994                                 && !ctx->supported_features.reparse_points
 995                                 && ctx->supported_features.symlink_reparse_points))
 996                 {
 997                         /*
 998                          * Unnamed data stream.  Skip if any of the following is true:
 999                          *
1000                          * - file is a directory
1001                          * - file is encrypted
1002                          * - backend needs to create the file as UNIX symlink
1003                          * - backend will extract the stream as externally backed
1004                          */
1005                         if (ctx->apply_ops->will_externally_back) {
1006                                 int ret = (*ctx->apply_ops->will_externally_back)(dentry, ctx);
1007                                 if (ret > 0) /* Error?  */
1008                                         return ret;
1009                                 if (ret < 0) /* Won't externally back?  */
1010                                         need_stream = true;
1011                         } else {
1012                                 need_stream = true;
1013                         }
1014                 }
1015                 break;
1016         case STREAM_TYPE_REPARSE_POINT:
1017                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT);
1018                 if (ctx->supported_features.reparse_points ||
1019                     (inode_is_symlink(inode) &&
1020                      ctx->supported_features.symlink_reparse_points))
1021                         need_stream = true;
1022                 break;
1023         case STREAM_TYPE_EFSRPC_RAW_DATA:
1024                 wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED);
1025                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
1026                         if (ctx->supported_features.encrypted_directories)
1027                                 need_stream = true;
1028                 } else {
1029                         if (ctx->supported_features.encrypted_files)
1030                                 need_stream = true;
1031                 }
1032                 break;
1033         }
1034         if (need_stream)
1035                 return ref_stream(strm, dentry, ctx);
1036         return 0;
1037 }
1038
1039 static int
1040 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
1041 {
1042         struct wim_inode *inode = dentry->d_inode;
1043         for (unsigned i = 0; i < inode->i_num_streams; i++) {
1044                 int ret = ref_stream_if_needed(dentry, inode,
1045                                                &inode->i_streams[i], ctx);
1046                 if (ret)
1047                         return ret;
1048         }
1049         inode->i_visited = 1;
1050         return 0;
1051 }
1052
1053 /*
1054  * Given a list of dentries to be extracted, build the list of blobs that need
1055  * to be extracted, and for each blob determine the streams to which that blob
1056  * will be extracted.
1057  *
1058  * This also initializes the extract progress info with byte and blob
1059  * information.
1060  *
1061  * ctx->supported_features must be filled in.
1062  */
1063 static int
1064 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
1065 {
1066         struct wim_dentry *dentry;
1067         int ret;
1068
1069         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1070                 ret = dentry_ref_streams(dentry, ctx);
1071                 if (ret)
1072                         return ret;
1073         }
1074         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1075                 dentry->d_inode->i_visited = 0;
1076         return 0;
1077 }
1078
1079 static void
1080 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
1081 {
1082         struct wim_dentry *dentry;
1083         struct wim_inode *inode;
1084
1085         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1086                 inode = dentry->d_inode;
1087                 if (!inode->i_visited)
1088                         INIT_LIST_HEAD(&inode->i_extraction_aliases);
1089                 list_add_tail(&dentry->d_extraction_alias_node,
1090                               &inode->i_extraction_aliases);
1091                 inode->i_visited = 1;
1092         }
1093         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1094                 dentry->d_inode->i_visited = 0;
1095 }
1096
1097 static void
1098 inode_tally_features(const struct wim_inode *inode,
1099                      struct wim_features *features)
1100 {
1101         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1102                 features->archive_files++;
1103         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1104                 features->hidden_files++;
1105         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1106                 features->system_files++;
1107         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1108                 features->compressed_files++;
1109         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1110                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1111                         features->encrypted_directories++;
1112                 else
1113                         features->encrypted_files++;
1114         }
1115         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1116                 features->not_context_indexed_files++;
1117         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1118                 features->sparse_files++;
1119         if (inode_has_named_data_stream(inode))
1120                 features->named_data_streams++;
1121         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1122                 features->reparse_points++;
1123                 if (inode_is_symlink(inode))
1124                         features->symlink_reparse_points++;
1125                 else
1126                         features->other_reparse_points++;
1127         }
1128         if (inode_has_security_descriptor(inode))
1129                 features->security_descriptors++;
1130         if (inode_has_unix_data(inode))
1131                 features->unix_data++;
1132 }
1133
1134 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1135 static void
1136 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
1137 {
1138         struct wim_inode *inode = dentry->d_inode;
1139
1140         if (dentry_has_short_name(dentry))
1141                 features->short_names++;
1142
1143         if (inode->i_visited) {
1144                 features->hard_links++;
1145         } else {
1146                 inode_tally_features(inode, features);
1147                 inode->i_visited = 1;
1148         }
1149 }
1150
1151 /* Tally the features necessary to extract the specified dentries.  */
1152 static void
1153 dentry_list_get_features(struct list_head *dentry_list,
1154                          struct wim_features *features)
1155 {
1156         struct wim_dentry *dentry;
1157
1158         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1159                 dentry_tally_features(dentry, features);
1160
1161         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1162                 dentry->d_inode->i_visited = 0;
1163 }
1164
1165 static int
1166 do_feature_check(const struct wim_features *required_features,
1167                  const struct wim_features *supported_features,
1168                  int extract_flags)
1169 {
1170         /* Encrypted files.  */
1171         if (required_features->encrypted_files &&
1172             !supported_features->encrypted_files)
1173                 WARNING("Ignoring EFS-encrypted data of %lu files",
1174                         required_features->encrypted_files);
1175
1176         /* Named data streams.  */
1177         if (required_features->named_data_streams &&
1178             !supported_features->named_data_streams)
1179                 WARNING("Ignoring named data streams of %lu files",
1180                         required_features->named_data_streams);
1181
1182         /* File attributes.  */
1183         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1184                 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1185                  * We're an archive program, so theoretically we can do what we
1186                  * want with it.  */
1187
1188                 if (required_features->hidden_files &&
1189                     !supported_features->hidden_files)
1190                         WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1191                                 required_features->hidden_files);
1192
1193                 if (required_features->system_files &&
1194                     !supported_features->system_files)
1195                         WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1196                                 required_features->system_files);
1197
1198                 if (required_features->compressed_files &&
1199                     !supported_features->compressed_files)
1200                         WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1201                                 required_features->compressed_files);
1202
1203                 if (required_features->not_context_indexed_files &&
1204                     !supported_features->not_context_indexed_files)
1205                         WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1206                                 required_features->not_context_indexed_files);
1207
1208                 if (required_features->sparse_files &&
1209                     !supported_features->sparse_files)
1210                         WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1211                                 required_features->sparse_files);
1212
1213                 if (required_features->encrypted_directories &&
1214                     !supported_features->encrypted_directories)
1215                         WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1216                                 required_features->encrypted_directories);
1217         }
1218
1219         /* Hard links.  */
1220         if (required_features->hard_links && !supported_features->hard_links)
1221                 WARNING("Extracting %lu hard links as independent files",
1222                         required_features->hard_links);
1223
1224         /* Symbolic links and reparse points.  */
1225         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1226             required_features->symlink_reparse_points &&
1227             !supported_features->symlink_reparse_points &&
1228             !supported_features->reparse_points)
1229         {
1230                 ERROR("Extraction backend does not support symbolic links!");
1231                 return WIMLIB_ERR_UNSUPPORTED;
1232         }
1233         if (required_features->reparse_points &&
1234             !supported_features->reparse_points)
1235         {
1236                 if (supported_features->symlink_reparse_points) {
1237                         if (required_features->other_reparse_points) {
1238                                 WARNING("Ignoring reparse data of %lu non-symlink/junction files",
1239                                         required_features->other_reparse_points);
1240                         }
1241                 } else {
1242                         WARNING("Ignoring reparse data of %lu files",
1243                                 required_features->reparse_points);
1244                 }
1245         }
1246
1247         /* Security descriptors.  */
1248         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1249                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1250              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1251             required_features->security_descriptors &&
1252             !supported_features->security_descriptors)
1253         {
1254                 ERROR("Extraction backend does not support security descriptors!");
1255                 return WIMLIB_ERR_UNSUPPORTED;
1256         }
1257         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1258             required_features->security_descriptors &&
1259             !supported_features->security_descriptors)
1260                 WARNING("Ignoring Windows NT security descriptors of %lu files",
1261                         required_features->security_descriptors);
1262
1263         /* UNIX data.  */
1264         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1265             required_features->unix_data && !supported_features->unix_data)
1266         {
1267                 ERROR("Extraction backend does not support UNIX data!");
1268                 return WIMLIB_ERR_UNSUPPORTED;
1269         }
1270
1271         if (required_features->unix_data &&
1272             !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1273         {
1274                 WARNING("Ignoring UNIX metadata of %lu files",
1275                         required_features->unix_data);
1276         }
1277
1278         /* DOS Names.  */
1279         if (required_features->short_names &&
1280             !supported_features->short_names)
1281         {
1282                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1283                         ERROR("Extraction backend does not support DOS names!");
1284                         return WIMLIB_ERR_UNSUPPORTED;
1285                 }
1286                 WARNING("Ignoring DOS names of %lu files",
1287                         required_features->short_names);
1288         }
1289
1290         /* Timestamps.  */
1291         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1292             !supported_features->timestamps)
1293         {
1294                 ERROR("Extraction backend does not support timestamps!");
1295                 return WIMLIB_ERR_UNSUPPORTED;
1296         }
1297
1298         return 0;
1299 }
1300
1301 static const struct apply_operations *
1302 select_apply_operations(int extract_flags)
1303 {
1304 #ifdef WITH_NTFS_3G
1305         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1306                 return &ntfs_3g_apply_ops;
1307 #endif
1308 #ifdef __WIN32__
1309         return &win32_apply_ops;
1310 #else
1311         return &unix_apply_ops;
1312 #endif
1313 }
1314
1315 static int
1316 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1317               const tchar *target, int extract_flags)
1318 {
1319         const struct apply_operations *ops;
1320         struct apply_ctx *ctx;
1321         int ret;
1322         LIST_HEAD(dentry_list);
1323
1324         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1325                 ret = extract_dentries_to_stdout(trees, num_trees,
1326                                                  wim->blob_table);
1327                 goto out;
1328         }
1329
1330         num_trees = remove_duplicate_trees(trees, num_trees);
1331         num_trees = remove_contained_trees(trees, num_trees);
1332
1333         ops = select_apply_operations(extract_flags);
1334
1335         if (num_trees > 1 && ops->single_tree_only) {
1336                 ERROR("Extracting multiple directory trees "
1337                       "at once is not supported in %s extraction mode!",
1338                       ops->name);
1339                 ret = WIMLIB_ERR_UNSUPPORTED;
1340                 goto out;
1341         }
1342
1343         ctx = CALLOC(1, ops->context_size);
1344         if (!ctx) {
1345                 ret = WIMLIB_ERR_NOMEM;
1346                 goto out;
1347         }
1348
1349         ctx->wim = wim;
1350         ctx->target = target;
1351         ctx->target_nchars = tstrlen(target);
1352         ctx->extract_flags = extract_flags;
1353         if (ctx->wim->progfunc) {
1354                 ctx->progfunc = ctx->wim->progfunc;
1355                 ctx->progctx = ctx->wim->progctx;
1356                 ctx->progress.extract.image = wim->current_image;
1357                 ctx->progress.extract.extract_flags = (extract_flags &
1358                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
1359                 ctx->progress.extract.wimfile_name = wim->filename;
1360                 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1361                                                                          wim->current_image);
1362                 ctx->progress.extract.target = target;
1363         }
1364         INIT_LIST_HEAD(&ctx->blob_list);
1365         filedes_invalidate(&ctx->tmpfile_fd);
1366         ctx->apply_ops = ops;
1367
1368         ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1369         if (ret)
1370                 goto out_cleanup;
1371
1372         build_dentry_list(&dentry_list, trees, num_trees,
1373                           !(extract_flags &
1374                             WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1375
1376         dentry_list_get_features(&dentry_list, &ctx->required_features);
1377
1378         ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1379                                ctx->extract_flags);
1380         if (ret)
1381                 goto out_cleanup;
1382
1383         ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1384         if (ret)
1385                 goto out_cleanup;
1386
1387         if (unlikely(list_empty(&dentry_list))) {
1388                 WARNING("There is nothing to extract!");
1389                 goto out_cleanup;
1390         }
1391
1392         ret = dentry_list_resolve_streams(&dentry_list, ctx);
1393         if (ret)
1394                 goto out_cleanup;
1395
1396         dentry_list_build_inode_alias_lists(&dentry_list);
1397
1398         ret = dentry_list_ref_streams(&dentry_list, ctx);
1399         if (ret)
1400                 goto out_cleanup;
1401
1402         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1403                 /* When extracting from a pipe, the number of bytes of data to
1404                  * extract can't be determined in the normal way (examining the
1405                  * blob table), since at this point all we have is a set of
1406                  * SHA-1 message digests of blobs that need to be extracted.
1407                  * However, we can get a reasonably accurate estimate by taking
1408                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1409                  * data.  This does assume that a full image is being extracted,
1410                  * but currently there is no API for doing otherwise.  (Also,
1411                  * subtract <HARDLINKBYTES> from this if hard links are
1412                  * supported by the extraction mode.)  */
1413                 ctx->progress.extract.total_bytes =
1414                         wim_info_get_image_total_bytes(wim->wim_info,
1415                                                        wim->current_image);
1416                 if (ctx->supported_features.hard_links) {
1417                         ctx->progress.extract.total_bytes -=
1418                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
1419                                                                    wim->current_image);
1420                 }
1421         }
1422
1423         ret = extract_progress(ctx,
1424                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1425                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
1426                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
1427         if (ret)
1428                 goto out_cleanup;
1429
1430         ret = (*ops->extract)(&dentry_list, ctx);
1431         if (ret)
1432                 goto out_cleanup;
1433
1434         if (ctx->progress.extract.completed_bytes <
1435             ctx->progress.extract.total_bytes)
1436         {
1437                 ctx->progress.extract.completed_bytes =
1438                         ctx->progress.extract.total_bytes;
1439                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
1440                 if (ret)
1441                         goto out_cleanup;
1442         }
1443
1444         ret = extract_progress(ctx,
1445                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1446                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
1447                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
1448 out_cleanup:
1449         destroy_blob_list(&ctx->blob_list);
1450         destroy_dentry_list(&dentry_list);
1451         FREE(ctx);
1452 out:
1453         return ret;
1454 }
1455
1456 static int
1457 mkdir_if_needed(const tchar *target)
1458 {
1459         if (!tmkdir(target, 0755))
1460                 return 0;
1461
1462         if (errno == EEXIST)
1463                 return 0;
1464
1465 #ifdef __WIN32__
1466         /* _wmkdir() fails with EACCES if called on a drive root directory.  */
1467         if (errno == EACCES)
1468                 return 0;
1469 #endif
1470
1471         ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
1472         return WIMLIB_ERR_MKDIR;
1473 }
1474
1475 /* Make sure the extraction flags make sense, and update them if needed.  */
1476 static int
1477 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1478 {
1479         int extract_flags = *extract_flags_p;
1480
1481         /* Check for invalid flag combinations  */
1482
1483         if ((extract_flags &
1484              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1485               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1486                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1487                 return WIMLIB_ERR_INVALID_PARAM;
1488
1489         if ((extract_flags &
1490              (WIMLIB_EXTRACT_FLAG_RPFIX |
1491               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1492                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
1493                 return WIMLIB_ERR_INVALID_PARAM;
1494
1495 #ifndef WITH_NTFS_3G
1496         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1497                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1498                       "        it cannot apply a WIM image directly to an NTFS volume.");
1499                 return WIMLIB_ERR_UNSUPPORTED;
1500         }
1501 #endif
1502
1503         if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1504 #ifdef __WIN32__
1505                 if (!wim->filename)
1506                         return WIMLIB_ERR_NO_FILENAME;
1507 #else
1508                 ERROR("WIMBoot extraction is only supported on Windows!");
1509                 return WIMLIB_ERR_UNSUPPORTED;
1510 #endif
1511         }
1512
1513
1514         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1515                               WIMLIB_EXTRACT_FLAG_NORPFIX |
1516                               WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1517                                         WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1518         {
1519                 /* For full-image extraction, do reparse point fixups by default
1520                  * if the WIM header says they are enabled.  */
1521                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1522                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1523         }
1524
1525         *extract_flags_p = extract_flags;
1526         return 0;
1527 }
1528
1529 static u32
1530 get_wildcard_flags(int extract_flags)
1531 {
1532         u32 wildcard_flags = 0;
1533
1534         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
1535                 wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
1536         else
1537                 wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
1538
1539         if (default_ignore_case)
1540                 wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
1541
1542         return wildcard_flags;
1543 }
1544
1545 struct append_dentry_ctx {
1546         struct wim_dentry **dentries;
1547         size_t num_dentries;
1548         size_t num_alloc_dentries;
1549 };
1550
1551 static int
1552 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1553 {
1554         struct append_dentry_ctx *ctx = _ctx;
1555
1556         if (ctx->num_dentries == ctx->num_alloc_dentries) {
1557                 struct wim_dentry **new_dentries;
1558                 size_t new_length;
1559
1560                 new_length = max(ctx->num_alloc_dentries + 8,
1561                                  ctx->num_alloc_dentries * 3 / 2);
1562                 new_dentries = REALLOC(ctx->dentries,
1563                                        new_length * sizeof(ctx->dentries[0]));
1564                 if (new_dentries == NULL)
1565                         return WIMLIB_ERR_NOMEM;
1566                 ctx->dentries = new_dentries;
1567                 ctx->num_alloc_dentries = new_length;
1568         }
1569         ctx->dentries[ctx->num_dentries++] = dentry;
1570         return 0;
1571 }
1572
1573 static int
1574 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1575                         const tchar * const *paths, size_t num_paths,
1576                         int extract_flags)
1577 {
1578         int ret;
1579         struct wim_dentry **trees;
1580         size_t num_trees;
1581
1582         if (wim == NULL || target == NULL || target[0] == T('\0') ||
1583             (num_paths != 0 && paths == NULL))
1584                 return WIMLIB_ERR_INVALID_PARAM;
1585
1586         ret = check_extract_flags(wim, &extract_flags);
1587         if (ret)
1588                 return ret;
1589
1590         ret = select_wim_image(wim, image);
1591         if (ret)
1592                 return ret;
1593
1594         ret = wim_checksum_unhashed_blobs(wim);
1595         if (ret)
1596                 return ret;
1597
1598         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1599                               WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1600             (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1601         {
1602                 ret = mkdir_if_needed(target);
1603                 if (ret)
1604                         return ret;
1605         }
1606
1607         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1608
1609                 struct append_dentry_ctx append_dentry_ctx = {
1610                         .dentries = NULL,
1611                         .num_dentries = 0,
1612                         .num_alloc_dentries = 0,
1613                 };
1614
1615                 u32 wildcard_flags = get_wildcard_flags(extract_flags);
1616
1617                 for (size_t i = 0; i < num_paths; i++) {
1618                         tchar *path = canonicalize_wim_path(paths[i]);
1619                         if (path == NULL) {
1620                                 ret = WIMLIB_ERR_NOMEM;
1621                                 trees = append_dentry_ctx.dentries;
1622                                 goto out_free_trees;
1623                         }
1624                         ret = expand_wildcard(wim, path,
1625                                               append_dentry_cb,
1626                                               &append_dentry_ctx,
1627                                               wildcard_flags);
1628                         FREE(path);
1629                         if (ret) {
1630                                 trees = append_dentry_ctx.dentries;
1631                                 goto out_free_trees;
1632                         }
1633                 }
1634                 trees = append_dentry_ctx.dentries;
1635                 num_trees = append_dentry_ctx.num_dentries;
1636         } else {
1637                 trees = MALLOC(num_paths * sizeof(trees[0]));
1638                 if (trees == NULL)
1639                         return WIMLIB_ERR_NOMEM;
1640
1641                 for (size_t i = 0; i < num_paths; i++) {
1642
1643                         tchar *path = canonicalize_wim_path(paths[i]);
1644                         if (path == NULL) {
1645                                 ret = WIMLIB_ERR_NOMEM;
1646                                 goto out_free_trees;
1647                         }
1648
1649                         trees[i] = get_dentry(wim, path,
1650                                               WIMLIB_CASE_PLATFORM_DEFAULT);
1651                         FREE(path);
1652                         if (trees[i] == NULL) {
1653                                   ERROR("Path \"%"TS"\" does not exist "
1654                                         "in WIM image %d",
1655                                         paths[i], wim->current_image);
1656                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1657                                   goto out_free_trees;
1658                         }
1659                 }
1660                 num_trees = num_paths;
1661         }
1662
1663         if (num_trees == 0) {
1664                 ret = 0;
1665                 goto out_free_trees;
1666         }
1667
1668         ret = extract_trees(wim, trees, num_trees, target, extract_flags);
1669 out_free_trees:
1670         FREE(trees);
1671         return ret;
1672 }
1673
1674 static int
1675 extract_single_image(WIMStruct *wim, int image,
1676                      const tchar *target, int extract_flags)
1677 {
1678         const tchar *path = WIMLIB_WIM_ROOT_PATH;
1679         extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1680         return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
1681 }
1682
1683 static const tchar * const filename_forbidden_chars =
1684 T(
1685 #ifdef __WIN32__
1686 "<>:\"/\\|?*"
1687 #else
1688 "/"
1689 #endif
1690 );
1691
1692 /* This function checks if it is okay to use a WIM image's name as a directory
1693  * name.  */
1694 static bool
1695 image_name_ok_as_dir(const tchar *image_name)
1696 {
1697         return image_name && *image_name &&
1698                 !tstrpbrk(image_name, filename_forbidden_chars) &&
1699                 tstrcmp(image_name, T(".")) &&
1700                 tstrcmp(image_name, T(".."));
1701 }
1702
1703 /* Extracts all images from the WIM to the directory @target, with the images
1704  * placed in subdirectories named by their image names. */
1705 static int
1706 extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
1707 {
1708         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
1709         size_t output_path_len = tstrlen(target);
1710         tchar buf[output_path_len + 1 + image_name_max_len + 1];
1711         int ret;
1712         int image;
1713         const tchar *image_name;
1714
1715         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1716                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1717                 return WIMLIB_ERR_INVALID_PARAM;
1718         }
1719
1720         ret = mkdir_if_needed(target);
1721         if (ret)
1722                 return ret;
1723         tmemcpy(buf, target, output_path_len);
1724         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1725         for (image = 1; image <= wim->hdr.image_count; image++) {
1726                 image_name = wimlib_get_image_name(wim, image);
1727                 if (image_name_ok_as_dir(image_name)) {
1728                         tstrcpy(buf + output_path_len + 1, image_name);
1729                 } else {
1730                         /* Image name is empty or contains forbidden characters.
1731                          * Use image number instead. */
1732                         tsprintf(buf + output_path_len + 1, T("%d"), image);
1733                 }
1734                 ret = extract_single_image(wim, image, buf, extract_flags);
1735                 if (ret)
1736                         return ret;
1737         }
1738         return 0;
1739 }
1740
1741 static int
1742 do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1743                         int extract_flags)
1744 {
1745         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1746                              WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1747                              WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1748                 return WIMLIB_ERR_INVALID_PARAM;
1749
1750         if (image == WIMLIB_ALL_IMAGES)
1751                 return extract_all_images(wim, target, extract_flags);
1752         else
1753                 return extract_single_image(wim, image, target, extract_flags);
1754 }
1755
1756
1757 /****************************************************************************
1758  *                          Extraction API                                  *
1759  ****************************************************************************/
1760
1761 WIMLIBAPI int
1762 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1763                      const tchar * const *paths, size_t num_paths,
1764                      int extract_flags)
1765 {
1766         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1767                 return WIMLIB_ERR_INVALID_PARAM;
1768
1769         return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1770                                        extract_flags);
1771 }
1772
1773 WIMLIBAPI int
1774 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1775                         const tchar *path_list_file, int extract_flags)
1776 {
1777         int ret;
1778         tchar **paths;
1779         size_t num_paths;
1780         void *mem;
1781
1782         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1783         if (ret) {
1784                 ERROR("Failed to read path list file \"%"TS"\"",
1785                       path_list_file);
1786                 return ret;
1787         }
1788
1789         ret = wimlib_extract_paths(wim, image, target,
1790                                    (const tchar * const *)paths, num_paths,
1791                                    extract_flags);
1792         FREE(paths);
1793         FREE(mem);
1794         return ret;
1795 }
1796
1797 WIMLIBAPI int
1798 wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
1799                                              const tchar *image_num_or_name,
1800                                              const tchar *target,
1801                                              int extract_flags,
1802                                              wimlib_progress_func_t progfunc,
1803                                              void *progctx)
1804 {
1805         int ret;
1806         WIMStruct *pwm;
1807         struct filedes *in_fd;
1808         int image;
1809         unsigned i;
1810
1811         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1812                 return WIMLIB_ERR_INVALID_PARAM;
1813
1814         /* Read the WIM header from the pipe and get a WIMStruct to represent
1815          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
1816          * wimlib_open_wim(), getting a WIMStruct in this way will result in an
1817          * empty blob table, no XML data read, and no filename set.  */
1818         ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
1819                                     progfunc, progctx);
1820         if (ret)
1821                 return ret;
1822
1823         /* Sanity check to make sure this is a pipable WIM.  */
1824         if (pwm->hdr.magic != PWM_MAGIC) {
1825                 ERROR("The WIM being read from file descriptor %d "
1826                       "is not pipable!", pipe_fd);
1827                 ret = WIMLIB_ERR_NOT_PIPABLE;
1828                 goto out_wimlib_free;
1829         }
1830
1831         /* Sanity check to make sure the first part of a pipable split WIM is
1832          * sent over the pipe first.  */
1833         if (pwm->hdr.part_number != 1) {
1834                 ERROR("The first part of the split WIM must be "
1835                       "sent over the pipe first.");
1836                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1837                 goto out_wimlib_free;
1838         }
1839
1840         in_fd = &pwm->in_fd;
1841         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1842
1843         /* As mentioned, the WIMStruct we created from the pipe does not have
1844          * XML data yet.  Fix this by reading the extra copy of the XML data
1845          * that directly follows the header in pipable WIMs.  (Note: see
1846          * write_pipable_wim() for more details about the format of pipable
1847          * WIMs.)  */
1848         {
1849                 u8 hash[SHA1_HASH_SIZE];
1850
1851                 ret = read_pwm_blob_header(pwm, hash,
1852                                            &pwm->hdr.xml_data_reshdr, NULL);
1853                 if (ret)
1854                         goto out_wimlib_free;
1855
1856                 if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1857                         ERROR("Expected XML data, but found non-metadata resource.");
1858                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1859                         goto out_wimlib_free;
1860                 }
1861
1862                 ret = read_wim_xml_data(pwm);
1863                 if (ret)
1864                         goto out_wimlib_free;
1865
1866                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
1867                         ERROR("Image count in XML data is not the same as in WIM header.");
1868                         ret = WIMLIB_ERR_IMAGE_COUNT;
1869                         goto out_wimlib_free;
1870                 }
1871         }
1872
1873         /* Get image index (this may use the XML data that was just read to
1874          * resolve an image name).  */
1875         if (image_num_or_name) {
1876                 image = wimlib_resolve_image(pwm, image_num_or_name);
1877                 if (image == WIMLIB_NO_IMAGE) {
1878                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1879                               image_num_or_name);
1880                         ret = WIMLIB_ERR_INVALID_IMAGE;
1881                         goto out_wimlib_free;
1882                 } else if (image == WIMLIB_ALL_IMAGES) {
1883                         ERROR("Applying all images from a pipe is not supported!");
1884                         ret = WIMLIB_ERR_INVALID_IMAGE;
1885                         goto out_wimlib_free;
1886                 }
1887         } else {
1888                 if (pwm->hdr.image_count != 1) {
1889                         ERROR("No image was specified, but the pipable WIM "
1890                               "did not contain exactly 1 image");
1891                         ret = WIMLIB_ERR_INVALID_IMAGE;
1892                         goto out_wimlib_free;
1893                 }
1894                 image = 1;
1895         }
1896
1897         /* Load the needed metadata resource.  */
1898         for (i = 1; i <= pwm->hdr.image_count; i++) {
1899                 struct wim_image_metadata *imd;
1900                 struct wim_reshdr reshdr;
1901                 struct wim_resource_descriptor *metadata_rdesc;
1902
1903                 imd = pwm->image_metadata[i - 1];
1904
1905                 ret = WIMLIB_ERR_NOMEM;
1906                 imd->metadata_blob = new_blob_descriptor();
1907                 if (!imd->metadata_blob)
1908                         goto out_wimlib_free;
1909
1910                 imd->metadata_blob->is_metadata = 1;
1911
1912                 ret = read_pwm_blob_header(pwm, imd->metadata_blob->hash,
1913                                            &reshdr, NULL);
1914                 if (ret)
1915                         goto out_wimlib_free;
1916
1917                 if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
1918                         ERROR("Expected metadata resource, but found "
1919                               "non-metadata resource");
1920                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1921                         goto out_wimlib_free;
1922                 }
1923
1924                 ret = WIMLIB_ERR_NOMEM;
1925                 metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
1926                 if (!metadata_rdesc)
1927                         goto out_wimlib_free;
1928                 wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc,
1929                                             imd->metadata_blob);
1930
1931                 if (i == image) {
1932                         /* Metadata resource is for the image being extracted.
1933                          * Parse it and save the metadata in memory.  */
1934                         ret = read_metadata_resource(imd);
1935                         if (ret)
1936                                 goto out_wimlib_free;
1937                         imd->modified = 1;
1938                 } else {
1939                         /* Metadata resource is not for the image being
1940                          * extracted.  Skip over it.  */
1941                         ret = skip_wim_resource(metadata_rdesc);
1942                         if (ret)
1943                                 goto out_wimlib_free;
1944                 }
1945         }
1946         /* Extract the image.  */
1947         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
1948         ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
1949         /* Clean up and return.  */
1950 out_wimlib_free:
1951         wimlib_free(pwm);
1952         return ret;
1953 }
1954
1955
1956 WIMLIBAPI int
1957 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
1958                                const tchar *target, int extract_flags)
1959 {
1960         return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
1961                                                             image_num_or_name,
1962                                                             target,
1963                                                             extract_flags,
1964                                                             NULL,
1965                                                             NULL);
1966 }
1967
1968 WIMLIBAPI int
1969 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1970                      int extract_flags)
1971 {
1972         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1973                 return WIMLIB_ERR_INVALID_PARAM;
1974         return do_wimlib_extract_image(wim, image, target, extract_flags);
1975 }