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 Eric Biggers
  10  *
  11  * This file is part of wimlib, a library for working with WIM files.
  12  *
  13  * wimlib is free software; you can redistribute it and/or modify it under the
  14  * terms of the GNU General Public License as published by the Free
  15  * Software Foundation; either version 3 of the License, or (at your option)
  16  * any later version.
  17  *
  18  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  19  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  20  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  21  * details.
  22  *
  23  * You should have received a copy of the GNU General Public License
  24  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  25  */
  26
  27 /*
  28  * This file provides the API functions wimlib_extract_image(),
  29  * wimlib_extract_image_from_pipe(), wimlib_extract_paths(), and
  30  * wimlib_extract_pathlist().  Internally, all end up calling
  31  * do_wimlib_extract_paths() and extract_trees().
  32  *
  33  * Although wimlib supports multiple extraction modes/backends (NTFS-3g, UNIX,
  34  * Win32), this file does not itself have code to extract files or directories
  35  * to any specific target; instead, it handles generic functionality and relies
  36  * on lower-level callback functions declared in `struct apply_operations' to do
  37  * the actual extraction.
  38  */
  39
  40 #ifdef HAVE_CONFIG_H
  41 #  include "config.h"
  42 #endif
  43
  44 #include "wimlib/apply.h"
  45 #include "wimlib/dentry.h"
  46 #include "wimlib/encoding.h"
  47 #include "wimlib/endianness.h"
  48 #include "wimlib/error.h"
  49 #include "wimlib/lookup_table.h"
  50 #include "wimlib/metadata.h"
  51 #include "wimlib/pathlist.h"
  52 #include "wimlib/paths.h"
  53 #include "wimlib/reparse.h"
  54 #include "wimlib/resource.h"
  55 #include "wimlib/security.h"
  56 #include "wimlib/unix_data.h"
  57 #ifdef __WIN32__
  58 #  include "wimlib/win32.h" /* for realpath() equivalent */
  59 #endif
  60 #include "wimlib/xml.h"
  61 #include "wimlib/wildcard.h"
  62 #include "wimlib/wim.h"
  63
  64 #include <errno.h>
  65 #include <fcntl.h>
  66 #include <stdlib.h>
  67 #include <sys/stat.h>
  68 #include <unistd.h>
  69
  70 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x80000000
  71 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x40000000
  72
  73 /* Keep in sync with wimlib.h  */
  74 #define WIMLIB_EXTRACT_MASK_PUBLIC                              \
  75         (WIMLIB_EXTRACT_FLAG_NTFS                       |       \
  76          WIMLIB_EXTRACT_FLAG_UNIX_DATA                  |       \
  77          WIMLIB_EXTRACT_FLAG_NO_ACLS                    |       \
  78          WIMLIB_EXTRACT_FLAG_STRICT_ACLS                |       \
  79          WIMLIB_EXTRACT_FLAG_RPFIX                      |       \
  80          WIMLIB_EXTRACT_FLAG_NORPFIX                    |       \
  81          WIMLIB_EXTRACT_FLAG_TO_STDOUT                  |       \
  82          WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES  |       \
  83          WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS         |       \
  84          WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS          |       \
  85          WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES         |       \
  86          WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS            |       \
  87          WIMLIB_EXTRACT_FLAG_GLOB_PATHS                 |       \
  88          WIMLIB_EXTRACT_FLAG_STRICT_GLOB                |       \
  89          WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES              |       \
  90          WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE  |       \
  91          WIMLIB_EXTRACT_FLAG_WIMBOOT)
  92
  93 /* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or
  94  * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA.  */
  95 int
  96 do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
  97 {
  98         ctx->count_until_file_progress = 500;  /* Arbitrary value to limit calls  */
  99         return extract_progress(ctx, msg);
 100 }
 101
 102 static int
 103 start_file_phase(struct apply_ctx *ctx, uint64_t end_file_count, enum wimlib_progress_msg msg)
 104 {
 105         ctx->progress.extract.current_file_count = 0;
 106         ctx->progress.extract.end_file_count = end_file_count;
 107         return do_file_extract_progress(ctx, msg);
 108 }
 109
 110 int
 111 start_file_structure_phase(struct apply_ctx *ctx, uint64_t end_file_count)
 112 {
 113         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 114 }
 115
 116 int
 117 start_file_metadata_phase(struct apply_ctx *ctx, uint64_t end_file_count)
 118 {
 119         return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 120 }
 121
 122 static int
 123 end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
 124 {
 125         ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count;
 126         return do_file_extract_progress(ctx, msg);
 127 }
 128
 129 int
 130 end_file_structure_phase(struct apply_ctx *ctx)
 131 {
 132         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 133 }
 134
 135 int
 136 end_file_metadata_phase(struct apply_ctx *ctx)
 137 {
 138         return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 139 }
 140
 141 /* Check whether the extraction of a dentry should be skipped completely.  */
 142 static bool
 143 dentry_is_supported(struct wim_dentry *dentry,
 144                     const struct wim_features *supported_features)
 145 {
 146         struct wim_inode *inode = dentry->d_inode;
 147
 148         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 149                 return supported_features->reparse_points ||
 150                         (inode_is_symlink(inode) &&
 151                          supported_features->symlink_reparse_points);
 152         }
 153         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
 154                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
 155                         return supported_features->encrypted_directories != 0;
 156                 else
 157                         return supported_features->encrypted_files != 0;
 158         }
 159         return true;
 160 }
 161
 162
 163 #define PWM_ALLOW_WIM_HDR 0x00001
 164
 165 /* Read the header from a stream in a pipable WIM.  */
 166 static int
 167 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
 168                        struct wim_resource_spec *rspec,
 169                        int flags, struct wim_header_disk *hdr_ret)
 170 {
 171         union {
 172                 struct pwm_stream_hdr stream_hdr;
 173                 struct wim_header_disk pwm_hdr;
 174         } buf;
 175         struct wim_reshdr reshdr;
 176         int ret;
 177
 178         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
 179         if (ret)
 180                 goto read_error;
 181
 182         if ((flags & PWM_ALLOW_WIM_HDR) &&
 183             le64_to_cpu(buf.stream_hdr.magic) == PWM_MAGIC)
 184         {
 185                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
 186                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
 187                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
 188
 189                 if (ret)
 190                         goto read_error;
 191                 lte->resource_location = RESOURCE_NONEXISTENT;
 192                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
 193                 return 0;
 194         }
 195
 196         if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
 197                 ERROR("Data read on pipe is invalid (expected stream header).");
 198                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 199         }
 200
 201         copy_hash(lte->hash, buf.stream_hdr.hash);
 202
 203         reshdr.size_in_wim = 0;
 204         reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
 205         reshdr.offset_in_wim = pwm->in_fd.offset;
 206         reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
 207         wim_res_hdr_to_spec(&reshdr, pwm, rspec);
 208         lte_bind_wim_resource_spec(lte, rspec);
 209         lte->flags = rspec->flags;
 210         lte->size = rspec->uncompressed_size;
 211         lte->offset_in_res = 0;
 212         return 0;
 213
 214 read_error:
 215         ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
 216         return ret;
 217 }
 218
 219 static int
 220 load_streams_from_pipe(struct apply_ctx *ctx,
 221                        const struct read_stream_list_callbacks *cbs)
 222 {
 223         struct wim_lookup_table_entry *found_lte = NULL;
 224         struct wim_resource_spec *rspec = NULL;
 225         struct wim_lookup_table *lookup_table;
 226         int ret;
 227
 228         ret = WIMLIB_ERR_NOMEM;
 229         found_lte = new_lookup_table_entry();
 230         if (!found_lte)
 231                 goto out;
 232
 233         rspec = MALLOC(sizeof(struct wim_resource_spec));
 234         if (!rspec)
 235                 goto out;
 236
 237         lookup_table = ctx->wim->lookup_table;
 238         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GUID_LEN);
 239         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
 240         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
 241         ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 242         if (ret)
 243                 goto out;
 244
 245         while (ctx->num_streams_remaining) {
 246                 struct wim_header_disk pwm_hdr;
 247                 struct wim_lookup_table_entry *needed_lte;
 248
 249                 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
 250                         lte_unbind_wim_resource_spec(found_lte);
 251                 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
 252                                              PWM_ALLOW_WIM_HDR, &pwm_hdr);
 253                 if (ret)
 254                         goto out;
 255
 256                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
 257                     && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
 258                     && (needed_lte = lookup_stream(lookup_table, found_lte->hash))
 259                     && (needed_lte->out_refcnt))
 260                 {
 261                         needed_lte->offset_in_res = found_lte->offset_in_res;
 262                         needed_lte->flags = found_lte->flags;
 263                         needed_lte->size = found_lte->size;
 264
 265                         lte_unbind_wim_resource_spec(found_lte);
 266                         lte_bind_wim_resource_spec(needed_lte, rspec);
 267
 268                         ret = (*cbs->begin_stream)(needed_lte,
 269                                                    cbs->begin_stream_ctx);
 270                         if (ret) {
 271                                 lte_unbind_wim_resource_spec(needed_lte);
 272                                 goto out;
 273                         }
 274
 275                         ret = extract_stream(needed_lte, needed_lte->size,
 276                                              cbs->consume_chunk,
 277                                              cbs->consume_chunk_ctx);
 278
 279                         ret = (*cbs->end_stream)(needed_lte, ret,
 280                                                  cbs->end_stream_ctx);
 281                         lte_unbind_wim_resource_spec(needed_lte);
 282                         if (ret)
 283                                 goto out;
 284                         ctx->num_streams_remaining--;
 285                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
 286                         ret = skip_wim_stream(found_lte);
 287                         if (ret)
 288                                 goto out;
 289                 } else {
 290                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
 291                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
 292
 293                         if (part_number != ctx->progress.extract.part_number ||
 294                             total_parts != ctx->progress.extract.total_parts ||
 295                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
 296                                    WIM_GUID_LEN))
 297                         {
 298                                 ctx->progress.extract.part_number = part_number;
 299                                 ctx->progress.extract.total_parts = total_parts;
 300                                 memcpy(ctx->progress.extract.guid,
 301                                        pwm_hdr.guid, WIM_GUID_LEN);
 302                                 ret = extract_progress(ctx,
 303                                                        WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 304                                 if (ret)
 305                                         goto out;
 306                         }
 307                 }
 308         }
 309         ret = 0;
 310 out:
 311         if (found_lte && found_lte->resource_location != RESOURCE_IN_WIM)
 312                 FREE(rspec);
 313         free_lookup_table_entry(found_lte);
 314         return ret;
 315 }
 316
 317 /* Creates a temporary file opened for writing.  The open file descriptor is
 318  * returned in @fd_ret and its name is returned in @name_ret (dynamically
 319  * allocated).  */
 320 static int
 321 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
 322 {
 323         tchar *name;
 324         int open_flags;
 325         int raw_fd;
 326
 327 retry:
 328         name = ttempnam(NULL, T("wimlib"));
 329         if (!name) {
 330                 ERROR_WITH_ERRNO("Failed to create temporary filename");
 331                 return WIMLIB_ERR_NOMEM;
 332         }
 333
 334         open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
 335 #ifdef __WIN32__
 336         open_flags |= _O_SHORT_LIVED;
 337 #endif
 338         raw_fd = topen(name, open_flags, 0600);
 339
 340         if (raw_fd < 0) {
 341                 if (errno == EEXIST) {
 342                         FREE(name);
 343                         goto retry;
 344                 }
 345                 ERROR_WITH_ERRNO("Failed to create temporary file "
 346                                  "\"%"TS"\"", name);
 347                 FREE(name);
 348                 return WIMLIB_ERR_OPEN;
 349         }
 350
 351         filedes_init(fd_ret, raw_fd);
 352         *name_ret = name;
 353         return 0;
 354 }
 355
 356 static int
 357 begin_extract_stream_wrapper(struct wim_lookup_table_entry *lte, void *_ctx)
 358 {
 359         struct apply_ctx *ctx = _ctx;
 360
 361         ctx->cur_stream = lte;
 362         ctx->cur_stream_offset = 0;
 363
 364         if (unlikely(lte->out_refcnt > MAX_OPEN_STREAMS))
 365                 return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
 366         else
 367                 return (*ctx->saved_cbs->begin_stream)(lte, ctx->saved_cbs->begin_stream_ctx);
 368 }
 369
 370 static int
 371 extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
 372 {
 373         struct apply_ctx *ctx = _ctx;
 374         union wimlib_progress_info *progress = &ctx->progress;
 375         int ret;
 376
 377         ctx->cur_stream_offset += size;
 378
 379         if (likely(ctx->supported_features.hard_links)) {
 380                 progress->extract.completed_bytes +=
 381                         (u64)size * ctx->cur_stream->out_refcnt;
 382                 if (ctx->cur_stream_offset == ctx->cur_stream->size)
 383                         progress->extract.completed_streams += ctx->cur_stream->out_refcnt;
 384         } else {
 385                 const struct stream_owner *owners = stream_owners(ctx->cur_stream);
 386                 for (u32 i = 0; i < ctx->cur_stream->out_refcnt; i++) {
 387                         const struct wim_inode *inode = owners[i].inode;
 388                         const struct wim_dentry *dentry;
 389
 390                         list_for_each_entry(dentry,
 391                                             &inode->i_extraction_aliases,
 392                                             d_extraction_alias_node)
 393                         {
 394                                 progress->extract.completed_bytes += size;
 395                                 if (ctx->cur_stream_offset == ctx->cur_stream->size)
 396                                         progress->extract.completed_streams++;
 397                         }
 398                 }
 399         }
 400         if (progress->extract.completed_bytes >= ctx->next_progress) {
 401
 402                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
 403                 if (ret)
 404                         return ret;
 405
 406                 if (progress->extract.completed_bytes >=
 407                     progress->extract.total_bytes)
 408                 {
 409                         ctx->next_progress = UINT64_MAX;
 410                 } else {
 411                         /* Send new message as soon as another 1/128 of the
 412                          * total has been extracted.  (Arbitrary number.)  */
 413                         ctx->next_progress =
 414                                 progress->extract.completed_bytes +
 415                                         progress->extract.total_bytes / 128;
 416
 417                         /* ... Unless that would be more than 5000000 bytes, in
 418                          * which case send the next after the next 5000000
 419                          * bytes.  (Another arbitrary number.)  */
 420                         if (progress->extract.completed_bytes + 5000000 <
 421                             ctx->next_progress)
 422                                 ctx->next_progress =
 423                                         progress->extract.completed_bytes + 5000000;
 424
 425                         /* ... But always send a message as soon as we're
 426                          * completely done.  */
 427                         if (progress->extract.total_bytes < ctx->next_progress)
 428                                 ctx->next_progress = progress->extract.total_bytes;
 429                 }
 430         }
 431
 432         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 433                 /* Just extracting to temporary file for now.  */
 434                 ret = full_write(&ctx->tmpfile_fd, chunk, size);
 435                 if (ret) {
 436                         ERROR_WITH_ERRNO("Error writing data to "
 437                                          "temporary file \"%"TS"\"",
 438                                          ctx->tmpfile_name);
 439                 }
 440                 return ret;
 441         } else {
 442                 return (*ctx->saved_cbs->consume_chunk)(chunk, size,
 443                                                         ctx->saved_cbs->consume_chunk_ctx);
 444         }
 445 }
 446
 447 static int
 448 extract_from_tmpfile(const tchar *tmpfile_name, struct apply_ctx *ctx)
 449 {
 450         struct wim_lookup_table_entry tmpfile_lte;
 451         struct wim_lookup_table_entry *orig_lte = ctx->cur_stream;
 452         const struct read_stream_list_callbacks *cbs = ctx->saved_cbs;
 453         int ret;
 454         const u32 orig_refcnt = orig_lte->out_refcnt;
 455
 456         BUILD_BUG_ON(MAX_OPEN_STREAMS < ARRAY_LEN(orig_lte->inline_stream_owners));
 457
 458         struct stream_owner *owners = orig_lte->stream_owners;
 459
 460         /* Copy the stream's data from the temporary file to each of its
 461          * destinations.
 462          *
 463          * This is executed only in the very uncommon case that a
 464          * single-instance stream is being extracted to more than
 465          * MAX_OPEN_STREAMS locations!  */
 466
 467         memcpy(&tmpfile_lte, orig_lte, sizeof(struct wim_lookup_table_entry));
 468         tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK;
 469         tmpfile_lte.file_on_disk = ctx->tmpfile_name;
 470         ret = 0;
 471         for (u32 i = 0; i < orig_refcnt; i++) {
 472
 473                 /* Note: it usually doesn't matter whether we pass the original
 474                  * stream entry to callbacks provided by the extraction backend
 475                  * as opposed to the tmpfile stream entry, since they shouldn't
 476                  * actually read data from the stream other than through the
 477                  * read_stream_prefix() call below.  But for
 478                  * WIMLIB_EXTRACT_FLAG_WIMBOOT mode on Windows it does matter
 479                  * because it needs the original stream location in order to
 480                  * create the external backing reference.  */
 481
 482                 orig_lte->out_refcnt = 1;
 483                 orig_lte->inline_stream_owners[0] = owners[i];
 484
 485                 ret = (*cbs->begin_stream)(orig_lte, cbs->begin_stream_ctx);
 486                 if (ret)
 487                         break;
 488
 489                 /* Extra SHA-1 isn't necessary here, but it shouldn't hurt as
 490                  * this case is very rare anyway.  */
 491                 ret = extract_stream(&tmpfile_lte, tmpfile_lte.size,
 492                                      cbs->consume_chunk,
 493                                      cbs->consume_chunk_ctx);
 494
 495                 ret = (*cbs->end_stream)(orig_lte, ret, cbs->end_stream_ctx);
 496                 if (ret)
 497                         break;
 498         }
 499         FREE(owners);
 500         orig_lte->out_refcnt = 0;
 501         return ret;
 502 }
 503
 504 static int
 505 end_extract_stream_wrapper(struct wim_lookup_table_entry *stream,
 506                            int status, void *_ctx)
 507 {
 508         struct apply_ctx *ctx = _ctx;
 509
 510         if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
 511                 filedes_close(&ctx->tmpfile_fd);
 512                 if (!status)
 513                         status = extract_from_tmpfile(ctx->tmpfile_name, ctx);
 514                 filedes_invalidate(&ctx->tmpfile_fd);
 515                 tunlink(ctx->tmpfile_name);
 516                 FREE(ctx->tmpfile_name);
 517                 return status;
 518         } else {
 519                 return (*ctx->saved_cbs->end_stream)(stream, status,
 520                                                      ctx->saved_cbs->end_stream_ctx);
 521         }
 522 }
 523
 524 /*
 525  * Read the list of single-instance streams to extract and feed their data into
 526  * the specified callback functions.
 527  *
 528  * This handles checksumming each stream.
 529  *
 530  * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
 531  *
 532  * This also works if the WIM is being read from a pipe, whereas attempting to
 533  * read streams directly (e.g. with read_full_stream_into_buf()) will not.
 534  *
 535  * This also will split up streams that will need to be extracted to more than
 536  * MAX_OPEN_STREAMS locations, as measured by the 'out_refcnt' of each stream.
 537  * Therefore, the apply_operations implementation need not worry about running
 538  * out of file descriptors, unless it might open more than one file descriptor
 539  * per nominal destination (e.g. Win32 currently might because the destination
 540  * file system might not support hard links).
 541  */
 542 int
 543 extract_stream_list(struct apply_ctx *ctx,
 544                     const struct read_stream_list_callbacks *cbs)
 545 {
 546         struct read_stream_list_callbacks wrapper_cbs = {
 547                 .begin_stream      = begin_extract_stream_wrapper,
 548                 .begin_stream_ctx  = ctx,
 549                 .consume_chunk     = extract_chunk_wrapper,
 550                 .consume_chunk_ctx = ctx,
 551                 .end_stream        = end_extract_stream_wrapper,
 552                 .end_stream_ctx    = ctx,
 553         };
 554         ctx->saved_cbs = cbs;
 555         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
 556                 return load_streams_from_pipe(ctx, &wrapper_cbs);
 557         } else {
 558                 return read_stream_list(&ctx->stream_list,
 559                                         offsetof(struct wim_lookup_table_entry,
 560                                                  extraction_list),
 561                                         &wrapper_cbs, VERIFY_STREAM_HASHES);
 562         }
 563 }
 564
 565 /* Extract a WIM dentry to standard output.
 566  *
 567  * This obviously doesn't make sense in all cases.  We return an error if the
 568  * dentry does not correspond to a regular file.  Otherwise we extract the
 569  * unnamed data stream only.  */
 570 static int
 571 extract_dentry_to_stdout(struct wim_dentry *dentry,
 572                          const struct wim_lookup_table *lookup_table)
 573 {
 574         struct wim_inode *inode = dentry->d_inode;
 575         struct wim_lookup_table_entry *lte;
 576         struct filedes _stdout;
 577
 578         if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
 579                                    FILE_ATTRIBUTE_DIRECTORY))
 580         {
 581                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
 582                       "extracted to standard output", dentry_full_path(dentry));
 583                 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
 584         }
 585
 586         lte = inode_unnamed_lte(inode, lookup_table);
 587         if (!lte) {
 588                 const u8 *hash = inode_unnamed_stream_hash(inode);
 589                 if (!is_zero_hash(hash))
 590                         return stream_not_found_error(inode, hash);
 591                 return 0;
 592         }
 593
 594         filedes_init(&_stdout, STDOUT_FILENO);
 595         return extract_full_stream_to_fd(lte, &_stdout);
 596 }
 597
 598 static int
 599 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
 600                            const struct wim_lookup_table *lookup_table)
 601 {
 602         for (size_t i = 0; i < num_dentries; i++) {
 603                 int ret = extract_dentry_to_stdout(dentries[i], lookup_table);
 604                 if (ret)
 605                         return ret;
 606         }
 607         return 0;
 608 }
 609
 610 /**********************************************************************/
 611
 612 /*
 613  * Removes duplicate dentries from the array.
 614  *
 615  * Returns the new number of dentries, packed at the front of the array.
 616  */
 617 static size_t
 618 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
 619 {
 620         size_t i, j = 0;
 621         for (i = 0; i < num_trees; i++) {
 622                 if (!trees[i]->tmp_flag) {
 623                         /* Found distinct dentry.  */
 624                         trees[i]->tmp_flag = 1;
 625                         trees[j++] = trees[i];
 626                 }
 627         }
 628         for (i = 0; i < j; i++)
 629                 trees[i]->tmp_flag = 0;
 630         return j;
 631 }
 632
 633 /*
 634  * Remove dentries that are descendants of other dentries in the array.
 635  *
 636  * Returns the new number of dentries, packed at the front of the array.
 637  */
 638 static size_t
 639 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
 640 {
 641         size_t i, j = 0;
 642         for (i = 0; i < num_trees; i++)
 643                 trees[i]->tmp_flag = 1;
 644         for (i = 0; i < num_trees; i++) {
 645                 struct wim_dentry *d = trees[i];
 646                 while (!dentry_is_root(d)) {
 647                         d = d->d_parent;
 648                         if (d->tmp_flag)
 649                                 goto tree_contained;
 650                 }
 651                 trees[j++] = trees[i];
 652                 continue;
 653
 654         tree_contained:
 655                 trees[i]->tmp_flag = 0;
 656         }
 657
 658         for (i = 0; i < j; i++)
 659                 trees[i]->tmp_flag = 0;
 660         return j;
 661 }
 662
 663 static int
 664 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
 665 {
 666         struct list_head *dentry_list = _dentry_list;
 667         list_add_tail(&dentry->d_extraction_list_node, dentry_list);
 668         return 0;
 669 }
 670
 671 static void
 672 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
 673 {
 674         dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
 675 }
 676
 677 static int
 678 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
 679 {
 680         list_del(&dentry->d_extraction_list_node);
 681         dentry_reset_extraction_list_node(dentry);
 682         return 0;
 683 }
 684
 685 /*
 686  * Build the preliminary list of dentries to be extracted.
 687  *
 688  * The list maintains the invariant that if d1 and d2 are in the list and d1 is
 689  * an ancestor of d2, then d1 appears before d2 in the list.
 690  */
 691 static void
 692 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
 693                   size_t num_trees, bool add_ancestors)
 694 {
 695         INIT_LIST_HEAD(dentry_list);
 696
 697         /* Add the trees recursively.  */
 698         for (size_t i = 0; i < num_trees; i++)
 699                 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
 700
 701         /* If requested, add ancestors of the trees.  */
 702         if (add_ancestors) {
 703                 for (size_t i = 0; i < num_trees; i++) {
 704                         struct wim_dentry *dentry = trees[i];
 705                         struct wim_dentry *ancestor;
 706                         struct list_head *place_after;
 707
 708                         if (dentry_is_root(dentry))
 709                                 continue;
 710
 711                         place_after = dentry_list;
 712                         ancestor = dentry;
 713                         do {
 714                                 ancestor = ancestor->d_parent;
 715                                 if (will_extract_dentry(ancestor)) {
 716                                         place_after = &ancestor->d_extraction_list_node;
 717                                         break;
 718                                 }
 719                         } while (!dentry_is_root(ancestor));
 720
 721                         ancestor = dentry;
 722                         do {
 723                                 ancestor = ancestor->d_parent;
 724                                 if (will_extract_dentry(ancestor))
 725                                         break;
 726                                 list_add(&ancestor->d_extraction_list_node, place_after);
 727                         } while (!dentry_is_root(ancestor));
 728                 }
 729         }
 730 }
 731
 732 static void
 733 destroy_dentry_list(struct list_head *dentry_list)
 734 {
 735         struct wim_dentry *dentry, *tmp;
 736         struct wim_inode *inode;
 737
 738         list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
 739                 inode = dentry->d_inode;
 740                 dentry_reset_extraction_list_node(dentry);
 741                 inode->i_visited = 0;
 742                 if ((void *)dentry->d_extraction_name != (void *)dentry->file_name)
 743                         FREE(dentry->d_extraction_name);
 744                 dentry->d_extraction_name = NULL;
 745                 dentry->d_extraction_name_nchars = 0;
 746         }
 747 }
 748
 749 static void
 750 destroy_stream_list(struct list_head *stream_list)
 751 {
 752         struct wim_lookup_table_entry *lte;
 753
 754         list_for_each_entry(lte, stream_list, extraction_list)
 755                 if (lte->out_refcnt > ARRAY_LEN(lte->inline_stream_owners))
 756                         FREE(lte->stream_owners);
 757 }
 758
 759 #ifdef __WIN32__
 760 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
 761 #else
 762 static const utf16lechar replacement_char = cpu_to_le16('?');
 763 #endif
 764
 765 static bool
 766 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
 767 {
 768         size_t i;
 769
 770         if (num_chars == 0)
 771                 return true;
 772         for (i = 0; i < num_chars; i++) {
 773                 switch (name[i]) {
 774         #ifdef __WIN32__
 775                 case cpu_to_le16('\\'):
 776                 case cpu_to_le16(':'):
 777                 case cpu_to_le16('*'):
 778                 case cpu_to_le16('?'):
 779                 case cpu_to_le16('"'):
 780                 case cpu_to_le16('<'):
 781                 case cpu_to_le16('>'):
 782                 case cpu_to_le16('|'):
 783         #endif
 784                 case cpu_to_le16('/'):
 785                 case cpu_to_le16('\0'):
 786                         if (fix)
 787                                 name[i] = replacement_char;
 788                         else
 789                                 return false;
 790                 }
 791         }
 792
 793 #ifdef __WIN32__
 794         if (name[num_chars - 1] == cpu_to_le16(' ') ||
 795             name[num_chars - 1] == cpu_to_le16('.'))
 796         {
 797                 if (fix)
 798                         name[num_chars - 1] = replacement_char;
 799                 else
 800                         return false;
 801         }
 802 #endif
 803         return true;
 804 }
 805
 806 static int
 807 dentry_calculate_extraction_name(struct wim_dentry *dentry,
 808                                  struct apply_ctx *ctx)
 809 {
 810         int ret;
 811
 812         if (!dentry_is_supported(dentry, &ctx->supported_features))
 813                 goto skip_dentry;
 814
 815         if (dentry_is_root(dentry))
 816                 return 0;
 817
 818 #ifdef WITH_NTFS_3G
 819         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 820                 dentry->d_extraction_name = dentry->file_name;
 821                 dentry->d_extraction_name_nchars = dentry->file_name_nbytes /
 822                                                    sizeof(utf16lechar);
 823                 return 0;
 824         }
 825 #endif
 826
 827         if (!ctx->supported_features.case_sensitive_filenames) {
 828                 struct wim_dentry *other;
 829                 list_for_each_entry(other, &dentry->d_ci_conflict_list,
 830                                     d_ci_conflict_list)
 831                 {
 832                         if (will_extract_dentry(other)) {
 833                                 if (ctx->extract_flags &
 834                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
 835                                         WARNING("\"%"TS"\" has the same "
 836                                                 "case-insensitive name as "
 837                                                 "\"%"TS"\"; extracting "
 838                                                 "dummy name instead",
 839                                                 dentry_full_path(dentry),
 840                                                 dentry_full_path(other));
 841                                         goto out_replace;
 842                                 } else {
 843                                         WARNING("Not extracting \"%"TS"\": "
 844                                                 "has same case-insensitive "
 845                                                 "name as \"%"TS"\"",
 846                                                 dentry_full_path(dentry),
 847                                                 dentry_full_path(other));
 848                                         goto skip_dentry;
 849                                 }
 850                         }
 851                 }
 852         }
 853
 854         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
 855                 ret = utf16le_get_tstr(dentry->file_name,
 856                                        dentry->file_name_nbytes,
 857                                        (const tchar **)&dentry->d_extraction_name,
 858                                        &dentry->d_extraction_name_nchars);
 859                 dentry->d_extraction_name_nchars /= sizeof(tchar);
 860                 return ret;
 861         } else {
 862                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
 863                 {
 864                         WARNING("\"%"TS"\" has an invalid filename "
 865                                 "that is not supported on this platform; "
 866                                 "extracting dummy name instead",
 867                                 dentry_full_path(dentry));
 868                         goto out_replace;
 869                 } else {
 870                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
 871                                 "that is not supported on this platform",
 872                                 dentry_full_path(dentry));
 873                         goto skip_dentry;
 874                 }
 875         }
 876
 877 out_replace:
 878         {
 879                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
 880
 881                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
 882                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
 883
 884                 const tchar *tchar_name;
 885                 size_t tchar_nchars;
 886
 887                 ret = utf16le_get_tstr(utf16_name_copy,
 888                                        dentry->file_name_nbytes,
 889                                        &tchar_name, &tchar_nchars);
 890                 if (ret)
 891                         return ret;
 892
 893                 tchar_nchars /= sizeof(tchar);
 894
 895                 size_t fixed_name_num_chars = tchar_nchars;
 896                 tchar fixed_name[tchar_nchars + 50];
 897
 898                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
 899                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
 900                                                  T(" (invalid filename #%lu)"),
 901                                                  ++ctx->invalid_sequence);
 902
 903                 utf16le_put_tstr(tchar_name);
 904
 905                 dentry->d_extraction_name = memdup(fixed_name,
 906                                                    2 * fixed_name_num_chars + 2);
 907                 if (!dentry->d_extraction_name)
 908                         return WIMLIB_ERR_NOMEM;
 909                 dentry->d_extraction_name_nchars = fixed_name_num_chars;
 910         }
 911         return 0;
 912
 913 skip_dentry:
 914         for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
 915         return 0;
 916 }
 917
 918 /*
 919  * Calculate the actual filename component at which each WIM dentry will be
 920  * extracted, with special handling for dentries that are unsupported by the
 921  * extraction backend or have invalid names.
 922  *
 923  * ctx->supported_features must be filled in.
 924  *
 925  * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
 926  */
 927 static int
 928 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
 929                                        struct apply_ctx *ctx)
 930 {
 931         struct list_head *prev, *cur;
 932
 933         /* Can't use list_for_each_entry() because a call to
 934          * dentry_calculate_extraction_name() may delete the current dentry and
 935          * its children from the list.  */
 936
 937         prev = dentry_list;
 938         for (;;) {
 939                 struct wim_dentry *dentry;
 940                 int ret;
 941
 942                 cur = prev->next;
 943                 if (cur == dentry_list)
 944                         break;
 945
 946                 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
 947
 948                 ret = dentry_calculate_extraction_name(dentry, ctx);
 949                 if (ret)
 950                         return ret;
 951
 952                 if (prev->next == cur)
 953                         prev = cur;
 954                 else
 955                         ; /* Current dentry and its children (which follow in
 956                              the list) were deleted.  prev stays the same.  */
 957         }
 958         return 0;
 959 }
 960
 961 static int
 962 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
 963                        struct wim_lookup_table *lookup_table)
 964 {
 965         struct wim_inode *inode = dentry->d_inode;
 966         struct wim_lookup_table_entry *lte;
 967         int ret;
 968         bool force = false;
 969
 970         /* Special case:  when extracting from a pipe, the WIM lookup table is
 971          * initially empty, so "resolving" an inode's streams is initially not
 972          * possible.  However, we still need to keep track of which streams,
 973          * identified by SHA1 message digests, need to be extracted, so we
 974          * "resolve" the inode's streams anyway by allocating new entries.  */
 975         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
 976                 force = true;
 977         ret = inode_resolve_streams(inode, lookup_table, force);
 978         if (ret)
 979                 return ret;
 980         for (u32 i = 0; i <= inode->i_num_ads; i++) {
 981                 lte = inode_stream_lte_resolved(inode, i);
 982                 if (lte)
 983                         lte->out_refcnt = 0;
 984         }
 985         return 0;
 986 }
 987
 988 /*
 989  * For each dentry to be extracted, resolve all streams in the corresponding
 990  * inode and set 'out_refcnt' in each to 0.
 991  *
 992  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
 993  */
 994 static int
 995 dentry_list_resolve_streams(struct list_head *dentry_list,
 996                             struct apply_ctx *ctx)
 997 {
 998         struct wim_dentry *dentry;
 999         int ret;
1000
1001         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1002                 ret = dentry_resolve_streams(dentry,
1003                                              ctx->extract_flags,
1004                                              ctx->wim->lookup_table);
1005                 if (ret)
1006                         return ret;
1007         }
1008         return 0;
1009 }
1010
1011 static int
1012 ref_stream(struct wim_lookup_table_entry *lte, u32 stream_idx,
1013            struct wim_dentry *dentry, struct apply_ctx *ctx)
1014 {
1015         struct wim_inode *inode = dentry->d_inode;
1016         struct stream_owner *stream_owners;
1017
1018         if (!lte)
1019                 return 0;
1020
1021         /* Tally the size only for each extraction of the stream (not hard
1022          * links).  */
1023         if (inode->i_visited && ctx->supported_features.hard_links)
1024                 return 0;
1025
1026         ctx->progress.extract.total_bytes += lte->size;
1027         ctx->progress.extract.total_streams++;
1028
1029         if (inode->i_visited)
1030                 return 0;
1031
1032         /* Add stream to the dentry_list only one time, even if it's going
1033          * to be extracted to multiple inodes.  */
1034         if (lte->out_refcnt == 0) {
1035                 list_add_tail(&lte->extraction_list, &ctx->stream_list);
1036                 ctx->num_streams_remaining++;
1037         }
1038
1039         /* If inode not yet been visited, append it to the stream_owners array.  */
1040         if (lte->out_refcnt < ARRAY_LEN(lte->inline_stream_owners)) {
1041                 stream_owners = lte->inline_stream_owners;
1042         } else {
1043                 struct stream_owner *prev_stream_owners;
1044                 size_t alloc_stream_owners;
1045
1046                 if (lte->out_refcnt == ARRAY_LEN(lte->inline_stream_owners)) {
1047                         prev_stream_owners = NULL;
1048                         alloc_stream_owners = ARRAY_LEN(lte->inline_stream_owners);
1049                 } else {
1050                         prev_stream_owners = lte->stream_owners;
1051                         alloc_stream_owners = lte->alloc_stream_owners;
1052                 }
1053
1054                 if (lte->out_refcnt == alloc_stream_owners) {
1055                         alloc_stream_owners *= 2;
1056                         stream_owners = REALLOC(prev_stream_owners,
1057                                                alloc_stream_owners *
1058                                                 sizeof(stream_owners[0]));
1059                         if (!stream_owners)
1060                                 return WIMLIB_ERR_NOMEM;
1061                         if (!prev_stream_owners) {
1062                                 memcpy(stream_owners,
1063                                        lte->inline_stream_owners,
1064                                        sizeof(lte->inline_stream_owners));
1065                         }
1066                         lte->stream_owners = stream_owners;
1067                         lte->alloc_stream_owners = alloc_stream_owners;
1068                 }
1069                 stream_owners = lte->stream_owners;
1070         }
1071         stream_owners[lte->out_refcnt].inode = inode;
1072         if (stream_idx == 0) {
1073                 stream_owners[lte->out_refcnt].stream_name = NULL;
1074         } else {
1075                 stream_owners[lte->out_refcnt].stream_name =
1076                         inode->i_ads_entries[stream_idx - 1].stream_name;
1077         }
1078         lte->out_refcnt++;
1079         return 0;
1080 }
1081
1082 static int
1083 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
1084 {
1085         struct wim_inode *inode = dentry->d_inode;
1086         int ret;
1087
1088         /* The unnamed data stream will always be extracted, except in an
1089          * unlikely case.  */
1090         if (!inode_is_encrypted_directory(inode)) {
1091                 u16 stream_idx;
1092                 struct wim_lookup_table_entry *stream;
1093
1094                 stream = inode_unnamed_stream_resolved(inode, &stream_idx);
1095                 ret = ref_stream(stream, stream_idx, dentry, ctx);
1096                 if (ret)
1097                         return ret;
1098         }
1099
1100         /* Named data streams will be extracted only if supported in the current
1101          * extraction mode and volume, and to avoid complications, if not doing
1102          * a linked extraction.  */
1103         if (ctx->supported_features.named_data_streams) {
1104                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1105                         if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
1106                                 continue;
1107                         ret = ref_stream(inode->i_ads_entries[i].lte, i + 1,
1108                                          dentry, ctx);
1109                         if (ret)
1110                                 return ret;
1111                 }
1112         }
1113         inode->i_visited = 1;
1114         return 0;
1115 }
1116
1117 /*
1118  * For each dentry to be extracted, iterate through the data streams of the
1119  * corresponding inode.  For each such stream that is not to be ignored due to
1120  * the supported features or extraction flags, add it to the list of streams to
1121  * be extracted (ctx->stream_list) if not already done so.
1122  *
1123  * Also builds a mapping from each stream to the inodes referencing it.
1124  *
1125  * This also initializes the extract progress info with byte and stream
1126  * information.
1127  *
1128  * ctx->supported_features must be filled in.
1129  *
1130  * Possible error codes: WIMLIB_ERR_NOMEM.
1131  */
1132 static int
1133 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
1134 {
1135         struct wim_dentry *dentry;
1136         int ret;
1137
1138         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1139                 ret = dentry_ref_streams(dentry, ctx);
1140                 if (ret)
1141                         return ret;
1142         }
1143         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1144                 dentry->d_inode->i_visited = 0;
1145         return 0;
1146 }
1147
1148 static void
1149 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
1150 {
1151         struct wim_dentry *dentry;
1152         struct wim_inode *inode;
1153
1154         list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
1155                 inode = dentry->d_inode;
1156                 if (!inode->i_visited)
1157                         INIT_LIST_HEAD(&inode->i_extraction_aliases);
1158                 list_add_tail(&dentry->d_extraction_alias_node,
1159                               &inode->i_extraction_aliases);
1160                 inode->i_visited = 1;
1161         }
1162         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1163                 dentry->d_inode->i_visited = 0;
1164 }
1165
1166 static void
1167 inode_tally_features(const struct wim_inode *inode,
1168                      struct wim_features *features)
1169 {
1170         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1171                 features->archive_files++;
1172         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1173                 features->hidden_files++;
1174         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1175                 features->system_files++;
1176         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1177                 features->compressed_files++;
1178         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1179                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1180                         features->encrypted_directories++;
1181                 else
1182                         features->encrypted_files++;
1183         }
1184         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1185                 features->not_context_indexed_files++;
1186         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1187                 features->sparse_files++;
1188         if (inode_has_named_stream(inode))
1189                 features->named_data_streams++;
1190         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1191                 features->reparse_points++;
1192                 if (inode_is_symlink(inode))
1193                         features->symlink_reparse_points++;
1194                 else
1195                         features->other_reparse_points++;
1196         }
1197         if (inode->i_security_id != -1)
1198                 features->security_descriptors++;
1199         if (inode_has_unix_data(inode))
1200                 features->unix_data++;
1201 }
1202
1203 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1204 static void
1205 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
1206 {
1207         struct wim_inode *inode = dentry->d_inode;
1208
1209         if (dentry_has_short_name(dentry))
1210                 features->short_names++;
1211
1212         if (inode->i_visited) {
1213                 features->hard_links++;
1214         } else {
1215                 inode_tally_features(inode, features);
1216                 inode->i_visited = 1;
1217         }
1218 }
1219
1220 /* Tally the features necessary to extract the specified dentries.  */
1221 static void
1222 dentry_list_get_features(struct list_head *dentry_list,
1223                          struct wim_features *features)
1224 {
1225         struct wim_dentry *dentry;
1226
1227         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1228                 dentry_tally_features(dentry, features);
1229
1230         list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1231                 dentry->d_inode->i_visited = 0;
1232 }
1233
1234 static int
1235 do_feature_check(const struct wim_features *required_features,
1236                  const struct wim_features *supported_features,
1237                  int extract_flags)
1238 {
1239         /* File attributes.  */
1240         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1241                 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1242                  * We're an archive program, so theoretically we can do what we
1243                  * want with it.  */
1244
1245                 if (required_features->hidden_files &&
1246                     !supported_features->hidden_files)
1247                         WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1248                                 required_features->hidden_files);
1249
1250                 if (required_features->system_files &&
1251                     !supported_features->system_files)
1252                         WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1253                                 required_features->system_files);
1254
1255                 if (required_features->compressed_files &&
1256                     !supported_features->compressed_files)
1257                         WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1258                                 required_features->compressed_files);
1259
1260                 if (required_features->not_context_indexed_files &&
1261                     !supported_features->not_context_indexed_files)
1262                         WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1263                                 required_features->not_context_indexed_files);
1264
1265                 if (required_features->sparse_files &&
1266                     !supported_features->sparse_files)
1267                         WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1268                                 required_features->sparse_files);
1269
1270                 if (required_features->encrypted_directories &&
1271                     !supported_features->encrypted_directories)
1272                         WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1273                                 required_features->encrypted_directories);
1274         }
1275
1276         /* Encrypted files.  */
1277         if (required_features->encrypted_files &&
1278             !supported_features->encrypted_files)
1279                 WARNING("Ignoring %lu encrypted files",
1280                         required_features->encrypted_files);
1281
1282         /* Named data streams.  */
1283         if (required_features->named_data_streams &&
1284             (!supported_features->named_data_streams))
1285                 WARNING("Ignoring named data streams of %lu files",
1286                         required_features->named_data_streams);
1287
1288         /* Hard links.  */
1289         if (required_features->hard_links && !supported_features->hard_links)
1290                 WARNING("Extracting %lu hard links as independent files",
1291                         required_features->hard_links);
1292
1293         /* Symbolic links and reparse points.  */
1294         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1295             required_features->symlink_reparse_points &&
1296             !supported_features->symlink_reparse_points &&
1297             !supported_features->reparse_points)
1298         {
1299                 ERROR("Extraction backend does not support symbolic links!");
1300                 return WIMLIB_ERR_UNSUPPORTED;
1301         }
1302         if (required_features->reparse_points &&
1303             !supported_features->reparse_points)
1304         {
1305                 if (supported_features->symlink_reparse_points) {
1306                         if (required_features->other_reparse_points) {
1307                                 WARNING("Ignoring %lu non-symlink/junction "
1308                                         "reparse point files",
1309                                         required_features->other_reparse_points);
1310                         }
1311                 } else {
1312                         WARNING("Ignoring %lu reparse point files",
1313                                 required_features->reparse_points);
1314                 }
1315         }
1316
1317         /* Security descriptors.  */
1318         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1319                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1320              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1321             required_features->security_descriptors &&
1322             !supported_features->security_descriptors)
1323         {
1324                 ERROR("Extraction backend does not support security descriptors!");
1325                 return WIMLIB_ERR_UNSUPPORTED;
1326         }
1327         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1328             required_features->security_descriptors &&
1329             !supported_features->security_descriptors)
1330                 WARNING("Ignoring Windows NT security descriptors of %lu files",
1331                         required_features->security_descriptors);
1332
1333         /* UNIX data.  */
1334         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1335             required_features->unix_data && !supported_features->unix_data)
1336         {
1337                 ERROR("Extraction backend does not support UNIX data!");
1338                 return WIMLIB_ERR_UNSUPPORTED;
1339         }
1340
1341         if (required_features->unix_data &&
1342             !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1343         {
1344                 WARNING("Ignoring UNIX metadata of %lu files",
1345                         required_features->unix_data);
1346         }
1347
1348         /* DOS Names.  */
1349         if (required_features->short_names &&
1350             !supported_features->short_names)
1351         {
1352                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1353                         ERROR("Extraction backend does not support DOS names!");
1354                         return WIMLIB_ERR_UNSUPPORTED;
1355                 }
1356                 WARNING("Ignoring DOS names of %lu files",
1357                         required_features->short_names);
1358         }
1359
1360         /* Timestamps.  */
1361         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1362             !supported_features->timestamps)
1363         {
1364                 ERROR("Extraction backend does not support timestamps!");
1365                 return WIMLIB_ERR_UNSUPPORTED;
1366         }
1367
1368         return 0;
1369 }
1370
1371 static const struct apply_operations *
1372 select_apply_operations(int extract_flags)
1373 {
1374 #ifdef WITH_NTFS_3G
1375         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1376                 return &ntfs_3g_apply_ops;
1377 #endif
1378 #ifdef __WIN32__
1379         return &win32_apply_ops;
1380 #else
1381         return &unix_apply_ops;
1382 #endif
1383 }
1384
1385 static int
1386 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1387               const tchar *target, int extract_flags)
1388 {
1389         const struct apply_operations *ops;
1390         struct apply_ctx *ctx;
1391         int ret;
1392         LIST_HEAD(dentry_list);
1393
1394         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1395                 ret = extract_dentries_to_stdout(trees, num_trees,
1396                                                  wim->lookup_table);
1397                 goto out;
1398         }
1399
1400         num_trees = remove_duplicate_trees(trees, num_trees);
1401         num_trees = remove_contained_trees(trees, num_trees);
1402
1403         ops = select_apply_operations(extract_flags);
1404
1405         if (num_trees > 1 && ops->single_tree_only) {
1406                 ERROR("Extracting multiple directory trees "
1407                       "at once is not supported in %s extraction mode!",
1408                       ops->name);
1409                 ret = WIMLIB_ERR_UNSUPPORTED;
1410                 goto out;
1411         }
1412
1413         ctx = CALLOC(1, ops->context_size);
1414         if (!ctx) {
1415                 ret = WIMLIB_ERR_NOMEM;
1416                 goto out;
1417         }
1418
1419         ctx->wim = wim;
1420         ctx->target = target;
1421         ctx->target_nchars = tstrlen(target);
1422         ctx->extract_flags = extract_flags;
1423         if (ctx->wim->progfunc) {
1424                 ctx->progfunc = ctx->wim->progfunc;
1425                 ctx->progctx = ctx->wim->progctx;
1426                 ctx->progress.extract.image = wim->current_image;
1427                 ctx->progress.extract.extract_flags = (extract_flags &
1428                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
1429                 ctx->progress.extract.wimfile_name = wim->filename;
1430                 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1431                                                                          wim->current_image);
1432                 ctx->progress.extract.target = target;
1433         }
1434         INIT_LIST_HEAD(&ctx->stream_list);
1435         filedes_invalidate(&ctx->tmpfile_fd);
1436
1437         ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1438         if (ret)
1439                 goto out_cleanup;
1440
1441         build_dentry_list(&dentry_list, trees, num_trees,
1442                           !(extract_flags &
1443                             WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1444
1445         dentry_list_get_features(&dentry_list, &ctx->required_features);
1446
1447         ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1448                                ctx->extract_flags);
1449         if (ret)
1450                 goto out_cleanup;
1451
1452         ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1453         if (ret)
1454                 goto out_cleanup;
1455
1456         ret = dentry_list_resolve_streams(&dentry_list, ctx);
1457         if (ret)
1458                 goto out_cleanup;
1459
1460         ret = dentry_list_ref_streams(&dentry_list, ctx);
1461         if (ret)
1462                 goto out_cleanup;
1463
1464         dentry_list_build_inode_alias_lists(&dentry_list);
1465
1466         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1467                 /* When extracting from a pipe, the number of bytes of data to
1468                  * extract can't be determined in the normal way (examining the
1469                  * lookup table), since at this point all we have is a set of
1470                  * SHA1 message digests of streams that need to be extracted.
1471                  * However, we can get a reasonably accurate estimate by taking
1472                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1473                  * data.  This does assume that a full image is being extracted,
1474                  * but currently there is no API for doing otherwise.  (Also,
1475                  * subtract <HARDLINKBYTES> from this if hard links are
1476                  * supported by the extraction mode.)  */
1477                 ctx->progress.extract.total_bytes =
1478                         wim_info_get_image_total_bytes(wim->wim_info,
1479                                                        wim->current_image);
1480                 if (ctx->supported_features.hard_links) {
1481                         ctx->progress.extract.total_bytes -=
1482                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
1483                                                                    wim->current_image);
1484                 }
1485         }
1486
1487         ret = extract_progress(ctx,
1488                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1489                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
1490                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
1491         if (ret)
1492                 goto out_cleanup;
1493
1494         ret = (*ops->extract)(&dentry_list, ctx);
1495         if (ret)
1496                 goto out_cleanup;
1497
1498         if (ctx->progress.extract.completed_bytes <
1499             ctx->progress.extract.total_bytes)
1500         {
1501                 ctx->progress.extract.completed_bytes =
1502                         ctx->progress.extract.total_bytes;
1503                 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
1504                 if (ret)
1505                         goto out_cleanup;
1506         }
1507
1508         ret = extract_progress(ctx,
1509                                ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1510                                        WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
1511                                        WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
1512 out_cleanup:
1513         destroy_stream_list(&ctx->stream_list);
1514         destroy_dentry_list(&dentry_list);
1515         FREE(ctx);
1516 out:
1517         return ret;
1518 }
1519
1520 static int
1521 mkdir_if_needed(const tchar *target)
1522 {
1523         if (!tmkdir(target, 0755))
1524                 return 0;
1525
1526         if (errno == EEXIST)
1527                 return 0;
1528
1529 #ifdef __WIN32__
1530         /* _wmkdir() fails with EACCES if called on a drive root directory.  */
1531         if (errno == EACCES)
1532                 return 0;
1533 #endif
1534
1535         ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
1536         return WIMLIB_ERR_MKDIR;
1537 }
1538
1539 /* Make sure the extraction flags make sense, and update them if needed.  */
1540 static int
1541 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1542 {
1543         int extract_flags = *extract_flags_p;
1544
1545         /* Check for invalid flag combinations  */
1546
1547         if ((extract_flags &
1548              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1549               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1550                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1551                 return WIMLIB_ERR_INVALID_PARAM;
1552
1553         if ((extract_flags &
1554              (WIMLIB_EXTRACT_FLAG_RPFIX |
1555               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1556                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
1557                 return WIMLIB_ERR_INVALID_PARAM;
1558
1559 #ifndef WITH_NTFS_3G
1560         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1561                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1562                       "        it cannot apply a WIM image directly to an NTFS volume.");
1563                 return WIMLIB_ERR_UNSUPPORTED;
1564         }
1565 #endif
1566
1567         if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1568 #ifdef __WIN32__
1569                 if (!wim->filename)
1570                         return WIMLIB_ERR_NO_FILENAME;
1571 #else
1572                 ERROR("WIMBoot extraction is only supported on Windows!");
1573                 return WIMLIB_ERR_UNSUPPORTED;
1574 #endif
1575         }
1576
1577
1578         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1579                               WIMLIB_EXTRACT_FLAG_NORPFIX |
1580                               WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1581                                         WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1582         {
1583                 /* For full-image extraction, do reparse point fixups by default
1584                  * if the WIM header says they are enabled.  */
1585                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1586                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1587         }
1588
1589         *extract_flags_p = extract_flags;
1590         return 0;
1591 }
1592
1593 static u32
1594 get_wildcard_flags(int extract_flags)
1595 {
1596         u32 wildcard_flags = 0;
1597
1598         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
1599                 wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
1600         else
1601                 wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
1602
1603         if (default_ignore_case)
1604                 wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
1605
1606         return wildcard_flags;
1607 }
1608
1609 struct append_dentry_ctx {
1610         struct wim_dentry **dentries;
1611         size_t num_dentries;
1612         size_t num_alloc_dentries;
1613 };
1614
1615 static int
1616 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1617 {
1618         struct append_dentry_ctx *ctx = _ctx;
1619
1620         if (ctx->num_dentries == ctx->num_alloc_dentries) {
1621                 struct wim_dentry **new_dentries;
1622                 size_t new_length;
1623
1624                 new_length = max(ctx->num_alloc_dentries + 8,
1625                                  ctx->num_alloc_dentries * 3 / 2);
1626                 new_dentries = REALLOC(ctx->dentries,
1627                                        new_length * sizeof(ctx->dentries[0]));
1628                 if (new_dentries == NULL)
1629                         return WIMLIB_ERR_NOMEM;
1630                 ctx->dentries = new_dentries;
1631                 ctx->num_alloc_dentries = new_length;
1632         }
1633         ctx->dentries[ctx->num_dentries++] = dentry;
1634         return 0;
1635 }
1636
1637 static int
1638 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1639                         const tchar * const *paths, size_t num_paths,
1640                         int extract_flags)
1641 {
1642         int ret;
1643         struct wim_dentry **trees;
1644         size_t num_trees;
1645
1646         if (wim == NULL || target == NULL || target[0] == T('\0') ||
1647             (num_paths != 0 && paths == NULL))
1648                 return WIMLIB_ERR_INVALID_PARAM;
1649
1650         ret = check_extract_flags(wim, &extract_flags);
1651         if (ret)
1652                 return ret;
1653
1654         ret = select_wim_image(wim, image);
1655         if (ret)
1656                 return ret;
1657
1658         ret = wim_checksum_unhashed_streams(wim);
1659         if (ret)
1660                 return ret;
1661
1662         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1663                               WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1664             (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1665         {
1666                 ret = mkdir_if_needed(target);
1667                 if (ret)
1668                         return ret;
1669         }
1670
1671         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1672
1673                 struct append_dentry_ctx append_dentry_ctx = {
1674                         .dentries = NULL,
1675                         .num_dentries = 0,
1676                         .num_alloc_dentries = 0,
1677                 };
1678
1679                 u32 wildcard_flags = get_wildcard_flags(extract_flags);
1680
1681                 for (size_t i = 0; i < num_paths; i++) {
1682                         tchar *path = canonicalize_wim_path(paths[i]);
1683                         if (path == NULL) {
1684                                 ret = WIMLIB_ERR_NOMEM;
1685                                 trees = append_dentry_ctx.dentries;
1686                                 goto out_free_trees;
1687                         }
1688                         ret = expand_wildcard(wim, path,
1689                                               append_dentry_cb,
1690                                               &append_dentry_ctx,
1691                                               wildcard_flags);
1692                         FREE(path);
1693                         if (ret) {
1694                                 trees = append_dentry_ctx.dentries;
1695                                 goto out_free_trees;
1696                         }
1697                 }
1698                 trees = append_dentry_ctx.dentries;
1699                 num_trees = append_dentry_ctx.num_dentries;
1700         } else {
1701                 trees = MALLOC(num_paths * sizeof(trees[0]));
1702                 if (trees == NULL)
1703                         return WIMLIB_ERR_NOMEM;
1704
1705                 for (size_t i = 0; i < num_paths; i++) {
1706
1707                         tchar *path = canonicalize_wim_path(paths[i]);
1708                         if (path == NULL) {
1709                                 ret = WIMLIB_ERR_NOMEM;
1710                                 goto out_free_trees;
1711                         }
1712
1713                         trees[i] = get_dentry(wim, path,
1714                                               WIMLIB_CASE_PLATFORM_DEFAULT);
1715                         FREE(path);
1716                         if (trees[i] == NULL) {
1717                                   ERROR("Path \"%"TS"\" does not exist "
1718                                         "in WIM image %d",
1719                                         paths[i], wim->current_image);
1720                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1721                                   goto out_free_trees;
1722                         }
1723                 }
1724                 num_trees = num_paths;
1725         }
1726
1727         if (num_trees == 0) {
1728                 ret = 0;
1729                 goto out_free_trees;
1730         }
1731
1732         ret = extract_trees(wim, trees, num_trees, target, extract_flags);
1733 out_free_trees:
1734         FREE(trees);
1735         return ret;
1736 }
1737
1738 static int
1739 extract_single_image(WIMStruct *wim, int image,
1740                      const tchar *target, int extract_flags)
1741 {
1742         const tchar *path = WIMLIB_WIM_ROOT_PATH;
1743         extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1744         return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
1745 }
1746
1747 static const tchar * const filename_forbidden_chars =
1748 T(
1749 #ifdef __WIN32__
1750 "<>:\"/\\|?*"
1751 #else
1752 "/"
1753 #endif
1754 );
1755
1756 /* This function checks if it is okay to use a WIM image's name as a directory
1757  * name.  */
1758 static bool
1759 image_name_ok_as_dir(const tchar *image_name)
1760 {
1761         return image_name && *image_name &&
1762                 !tstrpbrk(image_name, filename_forbidden_chars) &&
1763                 tstrcmp(image_name, T(".")) &&
1764                 tstrcmp(image_name, T(".."));
1765 }
1766
1767 /* Extracts all images from the WIM to the directory @target, with the images
1768  * placed in subdirectories named by their image names. */
1769 static int
1770 extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
1771 {
1772         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
1773         size_t output_path_len = tstrlen(target);
1774         tchar buf[output_path_len + 1 + image_name_max_len + 1];
1775         int ret;
1776         int image;
1777         const tchar *image_name;
1778
1779         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1780                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1781                 return WIMLIB_ERR_INVALID_PARAM;
1782         }
1783
1784         ret = mkdir_if_needed(target);
1785         if (ret)
1786                 return ret;
1787         tmemcpy(buf, target, output_path_len);
1788         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1789         for (image = 1; image <= wim->hdr.image_count; image++) {
1790                 image_name = wimlib_get_image_name(wim, image);
1791                 if (image_name_ok_as_dir(image_name)) {
1792                         tstrcpy(buf + output_path_len + 1, image_name);
1793                 } else {
1794                         /* Image name is empty or contains forbidden characters.
1795                          * Use image number instead. */
1796                         tsprintf(buf + output_path_len + 1, T("%d"), image);
1797                 }
1798                 ret = extract_single_image(wim, image, buf, extract_flags);
1799                 if (ret)
1800                         return ret;
1801         }
1802         return 0;
1803 }
1804
1805 static int
1806 do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1807                         int extract_flags)
1808 {
1809         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1810                              WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1811                              WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1812                 return WIMLIB_ERR_INVALID_PARAM;
1813
1814         if (image == WIMLIB_ALL_IMAGES)
1815                 return extract_all_images(wim, target, extract_flags);
1816         else
1817                 return extract_single_image(wim, image, target, extract_flags);
1818 }
1819
1820
1821 /****************************************************************************
1822  *                          Extraction API                                  *
1823  ****************************************************************************/
1824
1825 WIMLIBAPI int
1826 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1827                      const tchar * const *paths, size_t num_paths,
1828                      int extract_flags)
1829 {
1830         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1831                 return WIMLIB_ERR_INVALID_PARAM;
1832
1833         return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1834                                        extract_flags);
1835 }
1836
1837 WIMLIBAPI int
1838 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1839                         const tchar *path_list_file, int extract_flags)
1840 {
1841         int ret;
1842         tchar **paths;
1843         size_t num_paths;
1844         void *mem;
1845
1846         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1847         if (ret) {
1848                 ERROR("Failed to read path list file \"%"TS"\"",
1849                       path_list_file);
1850                 return ret;
1851         }
1852
1853         ret = wimlib_extract_paths(wim, image, target,
1854                                    (const tchar * const *)paths, num_paths,
1855                                    extract_flags);
1856         FREE(paths);
1857         FREE(mem);
1858         return ret;
1859 }
1860
1861 WIMLIBAPI int
1862 wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
1863                                              const tchar *image_num_or_name,
1864                                              const tchar *target,
1865                                              int extract_flags,
1866                                              wimlib_progress_func_t progfunc,
1867                                              void *progctx)
1868 {
1869         int ret;
1870         WIMStruct *pwm;
1871         struct filedes *in_fd;
1872         int image;
1873         unsigned i;
1874
1875         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1876                 return WIMLIB_ERR_INVALID_PARAM;
1877
1878         /* Read the WIM header from the pipe and get a WIMStruct to represent
1879          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
1880          * wimlib_open_wim(), getting a WIMStruct in this way will result in
1881          * an empty lookup table, no XML data read, and no filename set.  */
1882         ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
1883                                     progfunc, progctx);
1884         if (ret)
1885                 return ret;
1886
1887         /* Sanity check to make sure this is a pipable WIM.  */
1888         if (pwm->hdr.magic != PWM_MAGIC) {
1889                 ERROR("The WIM being read from file descriptor %d "
1890                       "is not pipable!", pipe_fd);
1891                 ret = WIMLIB_ERR_NOT_PIPABLE;
1892                 goto out_wimlib_free;
1893         }
1894
1895         /* Sanity check to make sure the first part of a pipable split WIM is
1896          * sent over the pipe first.  */
1897         if (pwm->hdr.part_number != 1) {
1898                 ERROR("The first part of the split WIM must be "
1899                       "sent over the pipe first.");
1900                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1901                 goto out_wimlib_free;
1902         }
1903
1904         in_fd = &pwm->in_fd;
1905         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1906
1907         /* As mentioned, the WIMStruct we created from the pipe does not have
1908          * XML data yet.  Fix this by reading the extra copy of the XML data
1909          * that directly follows the header in pipable WIMs.  (Note: see
1910          * write_pipable_wim() for more details about the format of pipable
1911          * WIMs.)  */
1912         {
1913                 struct wim_lookup_table_entry xml_lte;
1914                 struct wim_resource_spec xml_rspec;
1915                 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
1916                 if (ret)
1917                         goto out_wimlib_free;
1918
1919                 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
1920                 {
1921                         ERROR("Expected XML data, but found non-metadata "
1922                               "stream.");
1923                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1924                         goto out_wimlib_free;
1925                 }
1926
1927                 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
1928
1929                 ret = read_wim_xml_data(pwm);
1930                 if (ret)
1931                         goto out_wimlib_free;
1932
1933                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
1934                         ERROR("Image count in XML data is not the same as in WIM header.");
1935                         ret = WIMLIB_ERR_IMAGE_COUNT;
1936                         goto out_wimlib_free;
1937                 }
1938         }
1939
1940         /* Get image index (this may use the XML data that was just read to
1941          * resolve an image name).  */
1942         if (image_num_or_name) {
1943                 image = wimlib_resolve_image(pwm, image_num_or_name);
1944                 if (image == WIMLIB_NO_IMAGE) {
1945                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1946                               image_num_or_name);
1947                         ret = WIMLIB_ERR_INVALID_IMAGE;
1948                         goto out_wimlib_free;
1949                 } else if (image == WIMLIB_ALL_IMAGES) {
1950                         ERROR("Applying all images from a pipe is not supported!");
1951                         ret = WIMLIB_ERR_INVALID_IMAGE;
1952                         goto out_wimlib_free;
1953                 }
1954         } else {
1955                 if (pwm->hdr.image_count != 1) {
1956                         ERROR("No image was specified, but the pipable WIM "
1957                               "did not contain exactly 1 image");
1958                         ret = WIMLIB_ERR_INVALID_IMAGE;
1959                         goto out_wimlib_free;
1960                 }
1961                 image = 1;
1962         }
1963
1964         /* Load the needed metadata resource.  */
1965         for (i = 1; i <= pwm->hdr.image_count; i++) {
1966                 struct wim_lookup_table_entry *metadata_lte;
1967                 struct wim_image_metadata *imd;
1968                 struct wim_resource_spec *metadata_rspec;
1969
1970                 metadata_lte = new_lookup_table_entry();
1971                 if (metadata_lte == NULL) {
1972                         ret = WIMLIB_ERR_NOMEM;
1973                         goto out_wimlib_free;
1974                 }
1975                 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
1976                 if (metadata_rspec == NULL) {
1977                         ret = WIMLIB_ERR_NOMEM;
1978                         free_lookup_table_entry(metadata_lte);
1979                         goto out_wimlib_free;
1980                 }
1981
1982                 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
1983                 imd = pwm->image_metadata[i - 1];
1984                 imd->metadata_lte = metadata_lte;
1985                 if (ret) {
1986                         FREE(metadata_rspec);
1987                         goto out_wimlib_free;
1988                 }
1989
1990                 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
1991                         ERROR("Expected metadata resource, but found "
1992                               "non-metadata stream.");
1993                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1994                         goto out_wimlib_free;
1995                 }
1996
1997                 if (i == image) {
1998                         /* Metadata resource is for the image being extracted.
1999                          * Parse it and save the metadata in memory.  */
2000                         ret = read_metadata_resource(pwm, imd);
2001                         if (ret)
2002                                 goto out_wimlib_free;
2003                         imd->modified = 1;
2004                 } else {
2005                         /* Metadata resource is not for the image being
2006                          * extracted.  Skip over it.  */
2007                         ret = skip_wim_stream(metadata_lte);
2008                         if (ret)
2009                                 goto out_wimlib_free;
2010                 }
2011         }
2012         /* Extract the image.  */
2013         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
2014         ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
2015         /* Clean up and return.  */
2016 out_wimlib_free:
2017         wimlib_free(pwm);
2018         return ret;
2019 }
2020
2021
2022 WIMLIBAPI int
2023 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
2024                                const tchar *target, int extract_flags)
2025 {
2026         return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
2027                                                             image_num_or_name,
2028                                                             target,
2029                                                             extract_flags,
2030                                                             NULL,
2031                                                             NULL);
2032 }
2033
2034 WIMLIBAPI int
2035 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
2036                      int extract_flags)
2037 {
2038         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
2039                 return WIMLIB_ERR_INVALID_PARAM;
2040         return do_wimlib_extract_image(wim, image, target, extract_flags);
2041 }