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 #ifdef __WIN32__
  57 #  include "wimlib/win32.h" /* for realpath() equivalent */
  58 #endif
  59 #include "wimlib/xml.h"
  60 #include "wimlib/wildcard.h"
  61 #include "wimlib/wim.h"
  62
  63 #include <errno.h>
  64 #include <fcntl.h>
  65 #include <stdlib.h>
  66 #include <sys/stat.h>
  67 #include <unistd.h>
  68
  69 #define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
  70 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x40000000
  71 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x20000000
  72
  73 /* Keep in sync with wimlib.h  */
  74 #define WIMLIB_EXTRACT_MASK_PUBLIC                              \
  75         (WIMLIB_EXTRACT_FLAG_NTFS                       |       \
  76          WIMLIB_EXTRACT_FLAG_HARDLINK                   |       \
  77          WIMLIB_EXTRACT_FLAG_SYMLINK                    |       \
  78          WIMLIB_EXTRACT_FLAG_VERBOSE                    |       \
  79          WIMLIB_EXTRACT_FLAG_SEQUENTIAL                 |       \
  80          WIMLIB_EXTRACT_FLAG_UNIX_DATA                  |       \
  81          WIMLIB_EXTRACT_FLAG_NO_ACLS                    |       \
  82          WIMLIB_EXTRACT_FLAG_STRICT_ACLS                |       \
  83          WIMLIB_EXTRACT_FLAG_RPFIX                      |       \
  84          WIMLIB_EXTRACT_FLAG_NORPFIX                    |       \
  85          WIMLIB_EXTRACT_FLAG_TO_STDOUT                  |       \
  86          WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES  |       \
  87          WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS         |       \
  88          WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS          |       \
  89          WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES         |       \
  90          WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS            |       \
  91          WIMLIB_EXTRACT_FLAG_RESUME                     |       \
  92          WIMLIB_EXTRACT_FLAG_FILE_ORDER                 |       \
  93          WIMLIB_EXTRACT_FLAG_GLOB_PATHS                 |       \
  94          WIMLIB_EXTRACT_FLAG_STRICT_GLOB                |       \
  95          WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES              |       \
  96          WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE  |       \
  97          WIMLIB_EXTRACT_FLAG_WIMBOOT)
  98
  99 static bool
 100 dentry_in_list(const struct wim_dentry *dentry)
 101 {
 102         return dentry->extraction_list.next != NULL;
 103 }
 104
 105 static inline bool
 106 is_linked_extraction(const struct apply_ctx *ctx)
 107 {
 108         return 0 != (ctx->extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
 109                                            WIMLIB_EXTRACT_FLAG_SYMLINK));
 110 }
 111
 112 static inline bool
 113 can_extract_named_data_streams(const struct apply_ctx *ctx)
 114 {
 115         return ctx->supported_features.named_data_streams &&
 116                 !is_linked_extraction(ctx);
 117 }
 118 /* Inform library user of progress of stream extraction following the successful
 119  * extraction of a copy of the stream specified by @lte.  */
 120 static void
 121 update_extract_progress(struct apply_ctx *ctx,
 122                         const struct wim_lookup_table_entry *lte)
 123 {
 124         wimlib_progress_func_t progress_func = ctx->progress_func;
 125         union wimlib_progress_info *progress = &ctx->progress;
 126
 127         progress->extract.completed_bytes += lte->size;
 128         if (progress_func &&
 129             progress->extract.completed_bytes >= ctx->next_progress)
 130         {
 131                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, progress);
 132                 if (progress->extract.completed_bytes >=
 133                     progress->extract.total_bytes)
 134                 {
 135                         ctx->next_progress = ~0ULL;
 136                 } else {
 137                         ctx->next_progress += progress->extract.total_bytes / 128;
 138                         if (ctx->next_progress > progress->extract.total_bytes)
 139                                 ctx->next_progress = progress->extract.total_bytes;
 140                 }
 141         }
 142 }
 143
 144 #ifndef __WIN32__
 145 /* Extract a symbolic link (not directly as reparse data), handling fixing up
 146  * the target of absolute symbolic links and updating the extract progress.
 147  *
 148  * @inode must specify the WIM inode for a symbolic link or junction reparse
 149  * point.
 150  *
 151  * @lte_override overrides the resource used as the reparse data for the
 152  * symbolic link.  */
 153 static int
 154 extract_symlink(const tchar *path, struct apply_ctx *ctx,
 155                 struct wim_inode *inode,
 156                 struct wim_lookup_table_entry *lte_override)
 157 {
 158         ssize_t bufsize = ctx->ops->path_max;
 159         tchar target[bufsize];
 160         tchar *buf = target;
 161         tchar *fixed_target;
 162         ssize_t sret;
 163         int ret;
 164
 165         /* If absolute symbolic link fixups requested, reserve space in the link
 166          * target buffer for the absolute path of the target directory.  */
 167         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)
 168         {
 169                 buf += ctx->realtarget_nchars;
 170                 bufsize -= ctx->realtarget_nchars;
 171         }
 172
 173         /* Translate the WIM inode's reparse data into the link target.  */
 174         sret = wim_inode_readlink(inode, buf, bufsize - 1, lte_override);
 175         if (sret < 0) {
 176                 errno = -sret;
 177                 return WIMLIB_ERR_READLINK;
 178         }
 179         buf[sret] = '\0';
 180
 181         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
 182             buf[0] == '/')
 183         {
 184                 /* Fix absolute symbolic link target to point into the
 185                  * actual extraction destination.  */
 186                 tmemcpy(target, ctx->realtarget, ctx->realtarget_nchars);
 187                 fixed_target = target;
 188         } else {
 189                 /* Keep same link target.  */
 190                 fixed_target = buf;
 191         }
 192
 193         /* Call into the apply_operations to create the symbolic link.  */
 194         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
 195               path, fixed_target);
 196         ret = ctx->ops->create_symlink(fixed_target, path, ctx);
 197         if (ret) {
 198                 ERROR_WITH_ERRNO("Failed to create symlink "
 199                                  "\"%"TS"\" => \"%"TS"\"", path, fixed_target);
 200                 return ret;
 201         }
 202
 203         /* Account for reparse data consumed.  */
 204         update_extract_progress(ctx,
 205                                 (lte_override ? lte_override :
 206                                       inode_unnamed_lte_resolved(inode)));
 207         return 0;
 208 }
 209 #endif /* !__WIN32__ */
 210
 211 /* Create a file, directory, or symbolic link.  */
 212 static int
 213 extract_inode(const tchar *path, struct apply_ctx *ctx, struct wim_inode *inode)
 214 {
 215         int ret;
 216
 217 #ifndef __WIN32__
 218         if (ctx->supported_features.symlink_reparse_points &&
 219             !ctx->supported_features.reparse_points &&
 220             inode_is_symlink(inode))
 221         {
 222                 ret = extract_symlink(path, ctx, inode, NULL);
 223         } else
 224 #endif /* !__WIN32__ */
 225         if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
 226                 ret = ctx->ops->create_directory(path, ctx, &inode->extract_cookie);
 227                 if (ret) {
 228                         ERROR_WITH_ERRNO("Failed to create the directory "
 229                                          "\"%"TS"\"", path);
 230                 }
 231         } else if ((inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) &&
 232                     ctx->ops->extract_encrypted_stream_creates_file &&
 233                     ctx->supported_features.encrypted_files) {
 234                 ret = ctx->ops->extract_encrypted_stream(
 235                                 path, inode_unnamed_lte_resolved(inode), ctx);
 236                 if (ret) {
 237                         ERROR_WITH_ERRNO("Failed to create and extract "
 238                                          "encrypted file \"%"TS"\"", path);
 239                 }
 240         } else {
 241                 ret = ctx->ops->create_file(path, ctx, &inode->extract_cookie);
 242                 if (ret) {
 243                         ERROR_WITH_ERRNO("Failed to create the file "
 244                                          "\"%"TS"\"", path);
 245                 }
 246         }
 247         return ret;
 248 }
 249
 250 static int
 251 extract_hardlink(const tchar *oldpath, const tchar *newpath,
 252                  struct apply_ctx *ctx)
 253 {
 254         int ret;
 255
 256         DEBUG("Creating hardlink \"%"TS"\" => \"%"TS"\"", newpath, oldpath);
 257         ret = ctx->ops->create_hardlink(oldpath, newpath, ctx);
 258         if (ret) {
 259                 ERROR_WITH_ERRNO("Failed to create hardlink "
 260                                  "\"%"TS"\" => \"%"TS"\"",
 261                                  newpath, oldpath);
 262         }
 263         return ret;
 264 }
 265
 266 #ifdef __WIN32__
 267 static int
 268 try_extract_rpfix(u8 *rpbuf,
 269                   u16 *rpbuflen_p,
 270                   const wchar_t *extract_root_realpath,
 271                   unsigned extract_root_realpath_nchars)
 272 {
 273         struct reparse_data rpdata;
 274         wchar_t *target;
 275         size_t target_nchars;
 276         size_t stripped_nchars;
 277         wchar_t *stripped_target;
 278         wchar_t stripped_target_nchars;
 279         int ret;
 280
 281         utf16lechar *new_target;
 282         utf16lechar *new_print_name;
 283         size_t new_target_nchars;
 284         size_t new_print_name_nchars;
 285         utf16lechar *p;
 286
 287         ret = parse_reparse_data(rpbuf, *rpbuflen_p, &rpdata);
 288         if (ret)
 289                 return ret;
 290
 291         if (extract_root_realpath[0] == L'\0' ||
 292             extract_root_realpath[1] != L':' ||
 293             extract_root_realpath[2] != L'\\')
 294                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
 295
 296         ret = parse_substitute_name(rpdata.substitute_name,
 297                                     rpdata.substitute_name_nbytes,
 298                                     rpdata.rptag);
 299         if (ret < 0)
 300                 return 0;
 301         stripped_nchars = ret;
 302         target = rpdata.substitute_name;
 303         target_nchars = rpdata.substitute_name_nbytes / sizeof(utf16lechar);
 304         stripped_target = target + stripped_nchars;
 305         stripped_target_nchars = target_nchars - stripped_nchars;
 306
 307         new_target = alloca((6 + extract_root_realpath_nchars +
 308                              stripped_target_nchars) * sizeof(utf16lechar));
 309
 310         p = new_target;
 311         if (stripped_nchars == 6) {
 312                 /* Include \??\ prefix if it was present before */
 313                 p = wmempcpy(p, L"\\??\\", 4);
 314         }
 315
 316         /* Print name excludes the \??\ if present. */
 317         new_print_name = p;
 318         if (stripped_nchars != 0) {
 319                 /* Get drive letter from real path to extract root, if a drive
 320                  * letter was present before. */
 321                 *p++ = extract_root_realpath[0];
 322                 *p++ = extract_root_realpath[1];
 323         }
 324         /* Copy the rest of the extract root */
 325         p = wmempcpy(p, extract_root_realpath + 2, extract_root_realpath_nchars - 2);
 326
 327         /* Append the stripped target */
 328         p = wmempcpy(p, stripped_target, stripped_target_nchars);
 329         new_target_nchars = p - new_target;
 330         new_print_name_nchars = p - new_print_name;
 331
 332         if (new_target_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE ||
 333             new_print_name_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE)
 334                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
 335
 336         rpdata.substitute_name = new_target;
 337         rpdata.substitute_name_nbytes = new_target_nchars * sizeof(utf16lechar);
 338         rpdata.print_name = new_print_name;
 339         rpdata.print_name_nbytes = new_print_name_nchars * sizeof(utf16lechar);
 340         return make_reparse_buffer(&rpdata, rpbuf, rpbuflen_p);
 341 }
 342 #endif /* __WIN32__ */
 343
 344 /* Set reparse data on extracted file or directory that has
 345  * FILE_ATTRIBUTE_REPARSE_POINT set.  */
 346 static int
 347 extract_reparse_data(const tchar *path, struct apply_ctx *ctx,
 348                      struct wim_inode *inode,
 349                      struct wim_lookup_table_entry *lte_override)
 350 {
 351         int ret;
 352         u8 rpbuf[REPARSE_POINT_MAX_SIZE];
 353         u16 rpbuflen;
 354
 355         ret = wim_inode_get_reparse_data(inode, rpbuf, &rpbuflen, lte_override);
 356         if (ret)
 357                 goto error;
 358
 359 #ifdef __WIN32__
 360         /* Fix up target of absolute symbolic link or junction points so
 361          * that they point into the actual extraction target.  */
 362         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
 363             (inode->i_reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK ||
 364              inode->i_reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT) &&
 365             !inode->i_not_rpfixed)
 366         {
 367                 ret = try_extract_rpfix(rpbuf, &rpbuflen, ctx->realtarget,
 368                                         ctx->realtarget_nchars);
 369                 if (ret && !(ctx->extract_flags &
 370                              WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
 371                 {
 372                         WARNING("Reparse point fixup of \"%"TS"\" "
 373                                 "failed", path);
 374                         ret = 0;
 375                 }
 376                 if (ret)
 377                         goto error;
 378         }
 379 #endif
 380
 381         ret = ctx->ops->set_reparse_data(path, rpbuf, rpbuflen, ctx);
 382
 383         /* On Windows, the SeCreateSymbolicLink privilege is required to create
 384          * symbolic links.  To be more friendly towards non-Administrator users,
 385          * we merely warn the user if symbolic links cannot be created due to
 386          * insufficient permissions or privileges, unless
 387          * WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS was provided.  */
 388 #ifdef __WIN32__
 389         if (ret && inode_is_symlink(inode) &&
 390             (errno == EACCES || errno == EPERM) &&
 391             !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
 392         {
 393                 WARNING("Can't set reparse data on \"%"TS"\": "
 394                         "Access denied!\n"
 395                         "          You may be trying to "
 396                         "extract a symbolic link without the\n"
 397                         "          SeCreateSymbolicLink privilege, "
 398                         "which by default non-Administrator\n"
 399                         "          accounts do not have.",
 400                         path);
 401                 ret = 0;
 402         }
 403 #endif
 404         if (ret)
 405                 goto error;
 406
 407         /* Account for reparse data consumed.  */
 408         update_extract_progress(ctx,
 409                                 (lte_override ? lte_override :
 410                                       inode_unnamed_lte_resolved(inode)));
 411         return 0;
 412
 413 error:
 414         ERROR_WITH_ERRNO("Failed to set reparse data on \"%"TS"\"", path);
 415         return ret;
 416 }
 417
 418 /*
 419  * Extract zero or more streams to a file.
 420  *
 421  * This function operates slightly differently depending on whether @lte_spec is
 422  * NULL or not.  When @lte_spec is NULL, the behavior is to extract the default
 423  * file contents (unnamed stream), and, if named data streams are supported in
 424  * the extract mode and volume, any named data streams.  When @lte_spec is not
 425  * NULL, the behavior is to extract only all copies of the stream @lte_spec, and
 426  * in addition use @lte_spec to set the reparse data or create the symbolic link
 427  * if appropriate.
 428  *
 429  * @path
 430  *      Path to file to extract (as can be passed to apply_operations
 431  *      functions).
 432  * @ctx
 433  *      Apply context.
 434  * @dentry
 435  *      WIM dentry that corresponds to the file being extracted.
 436  * @lte_spec
 437  *      If non-NULL, specifies the lookup table entry for a stream to extract,
 438  *      and only that stream will be extracted (although there may be more than
 439  *      one instance of it).
 440  * @lte_override
 441  *      Used only if @lte_spec != NULL; it is passed to the extraction functions
 442  *      rather than @lte_spec, allowing the location of the stream to be
 443  *      overridden.  (This is used when the WIM is being read from a nonseekable
 444  *      file, such as a pipe, when streams need to be used more than once; each
 445  *      such stream is extracted to a temporary file.)
 446  */
 447 static int
 448 extract_streams(const tchar *path, struct apply_ctx *ctx,
 449                 struct wim_dentry *dentry,
 450                 struct wim_lookup_table_entry *lte_spec,
 451                 struct wim_lookup_table_entry *lte_override)
 452 {
 453         struct wim_inode *inode = dentry->d_inode;
 454         struct wim_lookup_table_entry *lte;
 455         file_spec_t file_spec;
 456         int ret;
 457
 458         if (dentry->was_linked)
 459                 return 0;
 460
 461 #ifdef ENABLE_DEBUG
 462         if (lte_spec) {
 463                 char sha1_str[100];
 464                 char *p = sha1_str;
 465                 for (unsigned i = 0; i < SHA1_HASH_SIZE; i++)
 466                         p += sprintf(p, "%02x", lte_override->hash[i]);
 467                 DEBUG("Extracting stream SHA1=%s to \"%"TS"\"",
 468                       sha1_str, path, inode->i_ino);
 469         } else {
 470                 DEBUG("Extracting streams to \"%"TS"\"", path, inode->i_ino);
 471         }
 472 #endif
 473
 474         if (ctx->ops->uses_cookies)
 475                 file_spec.cookie = inode->extract_cookie;
 476         else
 477                 file_spec.path = path;
 478
 479         /* Unnamed data stream.  */
 480         lte = inode_unnamed_lte_resolved(inode);
 481         if (lte && (!lte_spec || lte == lte_spec)) {
 482                 if (lte_spec)
 483                         lte = lte_override;
 484                 if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
 485                                              FILE_ATTRIBUTE_REPARSE_POINT)))
 486                 {
 487                         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED &&
 488                             ctx->supported_features.encrypted_files) {
 489                                 if (!ctx->ops->extract_encrypted_stream_creates_file) {
 490                                         ret = ctx->ops->extract_encrypted_stream(
 491                                                                 path, lte, ctx);
 492                                         if (ret)
 493                                                 goto error;
 494                                 }
 495                         } else {
 496                                 ret = ctx->ops->extract_unnamed_stream(
 497                                                         file_spec, lte, ctx,
 498                                                         dentry);
 499                                 if (ret)
 500                                         goto error;
 501                         }
 502                         update_extract_progress(ctx, lte);
 503                 }
 504                 else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)
 505                 {
 506                         ret = 0;
 507                         if (ctx->supported_features.reparse_points)
 508                                 ret = extract_reparse_data(path, ctx, inode, lte);
 509                 #ifndef __WIN32__
 510                         else if ((inode_is_symlink(inode) &&
 511                                   ctx->supported_features.symlink_reparse_points))
 512                                 ret = extract_symlink(path, ctx, inode, lte);
 513                 #endif
 514                         if (ret)
 515                                 return ret;
 516                 }
 517         }
 518
 519         /* Named data streams.  */
 520         if (can_extract_named_data_streams(ctx)) {
 521                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 522                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
 523
 524                         if (!ads_entry_is_named_stream(entry))
 525                                 continue;
 526                         lte = entry->lte;
 527                         if (!lte)
 528                                 continue;
 529                         if (lte_spec && lte_spec != lte)
 530                                 continue;
 531                         if (lte_spec)
 532                                 lte = lte_override;
 533                         ret = ctx->ops->extract_named_stream(file_spec, entry->stream_name,
 534                                                              entry->stream_name_nbytes / 2,
 535                                                              lte, ctx);
 536                         if (ret)
 537                                 goto error;
 538                         update_extract_progress(ctx, lte);
 539                 }
 540         }
 541         return 0;
 542
 543 error:
 544         ERROR_WITH_ERRNO("Failed to extract data of \"%"TS"\"", path);
 545         return ret;
 546 }
 547
 548 /* Set attributes on an extracted file or directory if supported by the
 549  * extraction mode.  */
 550 static int
 551 extract_file_attributes(const tchar *path, struct apply_ctx *ctx,
 552                         struct wim_dentry *dentry, unsigned pass)
 553 {
 554         int ret;
 555
 556         if (ctx->ops->set_file_attributes &&
 557             !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES) &&
 558             !(dentry == ctx->target_dentry && ctx->root_dentry_is_special)) {
 559                 u32 attributes = dentry->d_inode->i_attributes;
 560
 561                 /* Clear unsupported attributes.  */
 562                 attributes &= ctx->supported_attributes_mask;
 563
 564                 if ((attributes & FILE_ATTRIBUTE_DIRECTORY &&
 565                      !ctx->supported_features.encrypted_directories) ||
 566                     (!(attributes & FILE_ATTRIBUTE_DIRECTORY) &&
 567                      !ctx->supported_features.encrypted_files))
 568                 {
 569                         attributes &= ~FILE_ATTRIBUTE_ENCRYPTED;
 570                 }
 571
 572                 if (attributes == 0)
 573                         attributes = FILE_ATTRIBUTE_NORMAL;
 574
 575                 ret = ctx->ops->set_file_attributes(path, attributes, ctx, pass);
 576                 if (ret) {
 577                         ERROR_WITH_ERRNO("Failed to set attributes on "
 578                                          "\"%"TS"\"", path);
 579                         return ret;
 580                 }
 581         }
 582         return 0;
 583 }
 584
 585
 586 /* Set or remove the short (DOS) name on an extracted file or directory if
 587  * supported by the extraction mode.  Since DOS names are unimportant and it's
 588  * easy to run into problems setting them on Windows (SetFileShortName()
 589  * requires SE_RESTORE privilege, which only the Administrator can request, and
 590  * also requires DELETE access to the file), failure is ignored unless
 591  * WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES is set.  */
 592 static int
 593 extract_short_name(const tchar *path, struct apply_ctx *ctx,
 594                    struct wim_dentry *dentry)
 595 {
 596         int ret;
 597
 598         /* The root of the dentry tree being extracted may not be extracted to
 599          * its original name, so its short name should be ignored.  */
 600         if (dentry == ctx->target_dentry)
 601                 return 0;
 602
 603         if (ctx->supported_features.short_names) {
 604                 ret = ctx->ops->set_short_name(path,
 605                                                dentry->short_name,
 606                                                dentry->short_name_nbytes / 2,
 607                                                ctx);
 608                 if (ret && (ctx->extract_flags &
 609                             WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES))
 610                 {
 611                         ERROR_WITH_ERRNO("Failed to set short name of "
 612                                          "\"%"TS"\"", path);
 613                         return ret;
 614                 }
 615         }
 616         return 0;
 617 }
 618
 619 /* Set security descriptor, UNIX data, or neither on an extracted file, taking
 620  * into account the current extraction mode and flags.  */
 621 static int
 622 extract_security(const tchar *path, struct apply_ctx *ctx,
 623                  struct wim_dentry *dentry)
 624 {
 625         int ret;
 626         struct wim_inode *inode = dentry->d_inode;
 627
 628         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS)
 629                 return 0;
 630
 631         if ((ctx->target_dentry == dentry) && ctx->root_dentry_is_special)
 632                 return 0;
 633
 634 #ifndef __WIN32__
 635         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 636                 struct wimlib_unix_data data;
 637
 638                 ret = inode_get_unix_data(inode, &data, NULL);
 639                 if (ret < 0)
 640                         ret = 0;
 641                 else if (ret == 0)
 642                         ret = ctx->ops->set_unix_data(path, &data, ctx);
 643                 if (ret) {
 644                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
 645                                 ERROR_WITH_ERRNO("Failed to set UNIX owner, "
 646                                                  "group, and/or mode on "
 647                                                  "\"%"TS"\"", path);
 648                                 return ret;
 649                         } else {
 650                                 WARNING_WITH_ERRNO("Failed to set UNIX owner, "
 651                                                    "group, and/or/mode on "
 652                                                    "\"%"TS"\"", path);
 653                         }
 654                 }
 655         }
 656         else
 657 #endif /* __WIN32__ */
 658         if (ctx->supported_features.security_descriptors &&
 659             inode->i_security_id != -1)
 660         {
 661                 const struct wim_security_data *sd;
 662                 const u8 *desc;
 663                 size_t desc_size;
 664
 665                 sd = wim_const_security_data(ctx->wim);
 666                 desc = sd->descriptors[inode->i_security_id];
 667                 desc_size = sd->sizes[inode->i_security_id];
 668
 669                 ret = ctx->ops->set_security_descriptor(path, desc,
 670                                                         desc_size, ctx);
 671                 if (ret) {
 672                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
 673                                 ERROR_WITH_ERRNO("Failed to set security "
 674                                                  "descriptor on \"%"TS"\"", path);
 675                                 return ret;
 676                         } else {
 677                         #if 0
 678                                 if (errno != EACCES) {
 679                                         WARNING_WITH_ERRNO("Failed to set "
 680                                                            "security descriptor "
 681                                                            "on \"%"TS"\"", path);
 682                                 }
 683                         #endif
 684                                 ctx->no_security_descriptors++;
 685                         }
 686                 }
 687         }
 688         return 0;
 689 }
 690
 691 /* Set timestamps on an extracted file.  Failure is warning-only unless
 692  * WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS is set.  */
 693 static int
 694 extract_timestamps(const tchar *path, struct apply_ctx *ctx,
 695                    struct wim_dentry *dentry)
 696 {
 697         struct wim_inode *inode = dentry->d_inode;
 698         int ret;
 699
 700         if ((ctx->target_dentry == dentry) && ctx->root_dentry_is_special)
 701                 return 0;
 702
 703         if (ctx->ops->set_timestamps) {
 704                 ret = ctx->ops->set_timestamps(path,
 705                                                inode->i_creation_time,
 706                                                inode->i_last_write_time,
 707                                                inode->i_last_access_time,
 708                                                ctx);
 709                 if (ret) {
 710                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) {
 711                                 ERROR_WITH_ERRNO("Failed to set timestamps "
 712                                                  "on \"%"TS"\"", path);
 713                                 return ret;
 714                         } else {
 715                                 WARNING_WITH_ERRNO("Failed to set timestamps "
 716                                                    "on \"%"TS"\"", path);
 717                         }
 718                 }
 719         }
 720         return 0;
 721 }
 722
 723 /* Check whether the extraction of a dentry should be skipped completely.  */
 724 static bool
 725 dentry_is_supported(struct wim_dentry *dentry,
 726                     const struct wim_features *supported_features)
 727 {
 728         struct wim_inode *inode = dentry->d_inode;
 729
 730         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 731                 return supported_features->reparse_points ||
 732                         (inode_is_symlink(inode) &&
 733                          supported_features->symlink_reparse_points);
 734         }
 735         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
 736                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
 737                         return supported_features->encrypted_directories != 0;
 738                 else
 739                         return supported_features->encrypted_files != 0;
 740         }
 741         return true;
 742 }
 743
 744 /* Given a WIM dentry to extract, build the path to which to extract it, in the
 745  * format understood by the callbacks in the apply_operations being used.
 746  *
 747  * Write the resulting path into @path, which must have room for at least
 748  * ctx->ops->path_max characters.
 749  *
 750  * Return %true if successful; %false if this WIM dentry doesn't actually need
 751  * to be extracted or if the calculated path exceeds ctx->ops->max_path
 752  * characters.
 753  *
 754  * This function clobbers the tmp_list member of @dentry and its ancestors up
 755  * until the extraction root.  */
 756 static bool
 757 build_extraction_path(tchar path[], struct wim_dentry *dentry,
 758                       const struct apply_ctx *ctx)
 759 {
 760         size_t path_nchars;
 761         LIST_HEAD(ancestor_list);
 762         tchar *p = path;
 763         const tchar *target_prefix;
 764         size_t target_prefix_nchars;
 765         struct wim_dentry *d;
 766
 767         path_nchars = ctx->ops->path_prefix_nchars;
 768
 769         if (ctx->ops->requires_realtarget_in_paths) {
 770                 target_prefix        = ctx->realtarget;
 771                 target_prefix_nchars = ctx->realtarget_nchars;
 772         } else if (ctx->ops->requires_target_in_paths) {
 773                 target_prefix        = ctx->target;
 774                 target_prefix_nchars = ctx->target_nchars;
 775         } else {
 776                 target_prefix        = NULL;
 777                 target_prefix_nchars = 0;
 778         }
 779         path_nchars += target_prefix_nchars;
 780
 781         for (d = dentry; d != ctx->target_dentry; d = d->parent) {
 782                 if (!dentry_in_list(d))
 783                         break;
 784
 785                 path_nchars += d->extraction_name_nchars + 1;
 786                 list_add(&d->tmp_list, &ancestor_list);
 787         }
 788
 789         path_nchars++; /* null terminator */
 790
 791         if (path_nchars > ctx->ops->path_max) {
 792                 WARNING("\"%"TS"\": Path too long to extract",
 793                         dentry_full_path(dentry));
 794                 return false;
 795         }
 796
 797         p = tmempcpy(p, ctx->ops->path_prefix, ctx->ops->path_prefix_nchars);
 798         p = tmempcpy(p, target_prefix, target_prefix_nchars);
 799         list_for_each_entry(d, &ancestor_list, tmp_list) {
 800                 *p++ = ctx->ops->path_separator;
 801                 p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars);
 802         }
 803         *p++ = T('\0');
 804         wimlib_assert(p - path == path_nchars);
 805         return true;
 806 }
 807
 808 static unsigned
 809 get_num_path_components(const tchar *path, tchar path_separator)
 810 {
 811         unsigned num_components = 0;
 812 #ifdef __WIN32__
 813         /* Ignore drive letter.  */
 814         if (path[0] != L'\0' && path[1] == L':')
 815                 path += 2;
 816 #endif
 817
 818         while (*path) {
 819                 while (*path == path_separator)
 820                         path++;
 821                 if (*path)
 822                         num_components++;
 823                 while (*path && *path != path_separator)
 824                         path++;
 825         }
 826         return num_components;
 827 }
 828
 829 static int
 830 extract_multiimage_symlink(const tchar *oldpath, const tchar *newpath,
 831                            struct apply_ctx *ctx, struct wim_dentry *dentry)
 832 {
 833         size_t num_raw_path_components;
 834         const struct wim_dentry *d;
 835         size_t num_target_path_components;
 836         tchar *p;
 837         const tchar *p_old;
 838         int ret;
 839
 840         num_raw_path_components = 0;
 841         for (d = dentry; d != ctx->target_dentry; d = d->parent)
 842                 num_raw_path_components++;
 843
 844         if (ctx->ops->requires_realtarget_in_paths)
 845                 num_target_path_components = get_num_path_components(ctx->realtarget,
 846                                                                      ctx->ops->path_separator);
 847         else if (ctx->ops->requires_target_in_paths)
 848                 num_target_path_components = get_num_path_components(ctx->target,
 849                                                                      ctx->ops->path_separator);
 850         else
 851                 num_target_path_components = 0;
 852
 853         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
 854                 wimlib_assert(num_target_path_components > 0);
 855                 num_raw_path_components++;
 856                 num_target_path_components--;
 857         }
 858
 859         p_old = oldpath + ctx->ops->path_prefix_nchars;
 860 #ifdef __WIN32__
 861         if (p_old[0] != L'\0' && p_old[1] == ':')
 862                 p_old += 2;
 863 #endif
 864         while (*p_old == ctx->ops->path_separator)
 865                 p_old++;
 866         while (--num_target_path_components) {
 867                 while (*p_old != ctx->ops->path_separator)
 868                         p_old++;
 869                 while (*p_old == ctx->ops->path_separator)
 870                         p_old++;
 871         }
 872
 873         tchar symlink_target[tstrlen(p_old) + 3 * num_raw_path_components + 1];
 874
 875         p = &symlink_target[0];
 876         while (num_raw_path_components--) {
 877                 *p++ = '.';
 878                 *p++ = '.';
 879                 *p++ = ctx->ops->path_separator;
 880         }
 881         tstrcpy(p, p_old);
 882         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
 883               newpath, symlink_target);
 884         ret = ctx->ops->create_symlink(symlink_target, newpath, ctx);
 885         if (ret) {
 886                 ERROR_WITH_ERRNO("Failed to create symlink "
 887                                  "\"%"TS"\" => \"%"TS"\"",
 888                                  newpath, symlink_target);
 889         }
 890         return ret;
 891 }
 892
 893 /* Create the "skeleton" of an extracted file or directory.  Don't yet extract
 894  * data streams, reparse data (including symbolic links), timestamps, and
 895  * security descriptors.  Basically, everything that doesn't require reading
 896  * non-metadata resources from the WIM file and isn't delayed until the final
 897  * pass.  */
 898 static int
 899 do_dentry_extract_skeleton(tchar path[], struct wim_dentry *dentry,
 900                            struct apply_ctx *ctx)
 901 {
 902         struct wim_inode *inode = dentry->d_inode;
 903         int ret;
 904         const tchar *oldpath;
 905
 906         if (unlikely(is_linked_extraction(ctx))) {
 907                 struct wim_lookup_table_entry *unnamed_lte;
 908
 909                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
 910                 if (unnamed_lte && unnamed_lte->extracted_file) {
 911                         oldpath = unnamed_lte->extracted_file;
 912                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK)
 913                                 goto hardlink;
 914                         else
 915                                 goto symlink;
 916                 }
 917         }
 918
 919         /* Create hard link if this dentry corresponds to an already-extracted
 920          * inode.  */
 921         if (inode->i_extracted_file) {
 922                 oldpath = inode->i_extracted_file;
 923                 goto hardlink;
 924         }
 925
 926         /* Skip symlinks unless they can be extracted as reparse points rather
 927          * than created directly.  */
 928         if (inode_is_symlink(inode) && !ctx->supported_features.reparse_points)
 929                 return 0;
 930
 931         /* Create this file or directory unless it's the extraction root, which
 932          * was already created if necessary.  */
 933         if (dentry != ctx->target_dentry) {
 934                 ret = extract_inode(path, ctx, inode);
 935                 if (ret)
 936                         return ret;
 937         }
 938
 939         /* Create empty named data streams.  */
 940         if (can_extract_named_data_streams(ctx)) {
 941                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 942                         file_spec_t file_spec;
 943                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
 944
 945                         if (!ads_entry_is_named_stream(entry))
 946                                 continue;
 947                         if (entry->lte)
 948                                 continue;
 949                         if (ctx->ops->uses_cookies)
 950                                 file_spec.cookie = inode->extract_cookie;
 951                         else
 952                                 file_spec.path = path;
 953                         ret = ctx->ops->extract_named_stream(file_spec,
 954                                                              entry->stream_name,
 955                                                              entry->stream_name_nbytes / 2,
 956                                                              entry->lte, ctx);
 957                         if (ret) {
 958                                 ERROR_WITH_ERRNO("\"%"TS"\": failed to create "
 959                                                  "empty named data stream",
 960                                                  path);
 961                                 return ret;
 962                         }
 963                 }
 964         }
 965
 966         /* Set file attributes (if supported).  */
 967         ret = extract_file_attributes(path, ctx, dentry, 0);
 968         if (ret)
 969                 return ret;
 970
 971         /* Set or remove file short name (if supported).  */
 972         ret = extract_short_name(path, ctx, dentry);
 973         if (ret)
 974                 return ret;
 975
 976         /* If inode has multiple links and hard links are supported in this
 977          * extraction mode and volume, save the path to the extracted file in
 978          * case it's needed to create a hard link.  */
 979         if (unlikely(is_linked_extraction(ctx))) {
 980                 struct wim_lookup_table_entry *unnamed_lte;
 981
 982                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
 983                 if (unnamed_lte) {
 984                         unnamed_lte->extracted_file = TSTRDUP(path);
 985                         if (!unnamed_lte->extracted_file)
 986                                 return WIMLIB_ERR_NOMEM;
 987                 }
 988         } else if (inode->i_nlink > 1 && ctx->supported_features.hard_links) {
 989                 inode->i_extracted_file = TSTRDUP(path);
 990                 if (!inode->i_extracted_file)
 991                         return WIMLIB_ERR_NOMEM;
 992         }
 993         return 0;
 994
 995 symlink:
 996         ret = extract_multiimage_symlink(oldpath, path, ctx, dentry);
 997         if (ret)
 998                 return ret;
 999         dentry->was_linked = 1;
1000         return 0;
1001
1002 hardlink:
1003         ret = extract_hardlink(oldpath, path, ctx);
1004         if (ret)
1005                 return ret;
1006         dentry->was_linked = 1;
1007         return 0;
1008 }
1009
1010 /* This is a wrapper around do_dentry_extract_skeleton() that handles building
1011  * the path, doing short name reordering.  This is also idempotent; dentries
1012  * already processed have skeleton_extracted set and no action is taken.  See
1013  * apply_operations.requires_short_name_reordering for more details about short
1014  * name reordering.  */
1015 static int
1016 dentry_extract_skeleton(struct wim_dentry *dentry, struct apply_ctx *ctx)
1017 {
1018         tchar path[ctx->ops->path_max];
1019         struct wim_dentry *orig_dentry;
1020         struct wim_dentry *other_dentry;
1021         int ret;
1022
1023         if (dentry->skeleton_extracted)
1024                 return 0;
1025
1026         orig_dentry = NULL;
1027         if (ctx->supported_features.short_names
1028             && ctx->ops->requires_short_name_reordering
1029             && !dentry_has_short_name(dentry)
1030             && !dentry->d_inode->i_dos_name_extracted)
1031         {
1032                 inode_for_each_dentry(other_dentry, dentry->d_inode) {
1033                         if (dentry_has_short_name(other_dentry)
1034                             && !other_dentry->skeleton_extracted
1035                             && dentry_in_list(other_dentry))
1036                         {
1037                                 DEBUG("Creating %"TS" before %"TS" "
1038                                       "to guarantee correct DOS name extraction",
1039                                       dentry_full_path(other_dentry),
1040                                       dentry_full_path(dentry));
1041                                 orig_dentry = dentry;
1042                                 dentry = other_dentry;
1043                                 break;
1044                         }
1045                 }
1046         }
1047 again:
1048         if (!build_extraction_path(path, dentry, ctx))
1049                 return 0;
1050         ret = do_dentry_extract_skeleton(path, dentry, ctx);
1051         if (ret)
1052                 return ret;
1053
1054         dentry->skeleton_extracted = 1;
1055
1056         if (orig_dentry) {
1057                 dentry = orig_dentry;
1058                 orig_dentry = NULL;
1059                 goto again;
1060         }
1061         dentry->d_inode->i_dos_name_extracted = 1;
1062         return 0;
1063 }
1064
1065 /* Create a file or directory, then immediately extract all streams.  The WIM
1066  * may not be read sequentially by this function.  */
1067 static int
1068 dentry_extract(struct wim_dentry *dentry, struct apply_ctx *ctx)
1069 {
1070         tchar path[ctx->ops->path_max];
1071         int ret;
1072
1073         ret = dentry_extract_skeleton(dentry, ctx);
1074         if (ret)
1075                 return ret;
1076
1077         if (!build_extraction_path(path, dentry, ctx))
1078                 return 0;
1079
1080         return extract_streams(path, ctx, dentry, NULL, NULL);
1081 }
1082
1083 /* Finish extracting a file, directory, or symbolic link by setting file
1084  * security and timestamps.  */
1085 static int
1086 dentry_extract_final(struct wim_dentry *dentry, struct apply_ctx *ctx)
1087 {
1088         int ret;
1089         tchar path[ctx->ops->path_max];
1090
1091         if (!build_extraction_path(path, dentry, ctx))
1092                 return 0;
1093
1094         ret = extract_security(path, ctx, dentry);
1095         if (ret)
1096                 return ret;
1097
1098         if (ctx->ops->requires_final_set_attributes_pass) {
1099                 /* Set file attributes (if supported).  */
1100                 ret = extract_file_attributes(path, ctx, dentry, 1);
1101                 if (ret)
1102                         return ret;
1103         }
1104
1105         return extract_timestamps(path, ctx, dentry);
1106 }
1107
1108 static int
1109 extract_structure(struct list_head *dentry_list, struct apply_ctx *ctx)
1110 {
1111         struct wim_dentry *dentry;
1112         int ret;
1113
1114         list_for_each_entry(dentry, dentry_list, extraction_list) {
1115                 ret = dentry_extract_skeleton(dentry, ctx);
1116                 if (ret)
1117                         return ret;
1118         }
1119         return 0;
1120 }
1121
1122 static int
1123 extract_dir_structure(struct list_head *dentry_list, struct apply_ctx *ctx)
1124 {
1125         struct wim_dentry *dentry;
1126         int ret;
1127
1128         list_for_each_entry(dentry, dentry_list, extraction_list) {
1129                 if (dentry_is_directory(dentry)) {
1130                         ret = dentry_extract_skeleton(dentry, ctx);
1131                         if (ret)
1132                                 return ret;
1133                 }
1134         }
1135         return 0;
1136 }
1137
1138 static int
1139 extract_dentries(struct list_head *dentry_list, struct apply_ctx *ctx)
1140 {
1141         struct wim_dentry *dentry;
1142         int ret;
1143
1144         list_for_each_entry(dentry, dentry_list, extraction_list) {
1145                 ret = dentry_extract(dentry, ctx);
1146                 if (ret)
1147                         return ret;
1148         }
1149         return 0;
1150 }
1151
1152 static int
1153 extract_final_metadata(struct list_head *dentry_list, struct apply_ctx *ctx)
1154 {
1155         struct wim_dentry *dentry;
1156         int ret;
1157
1158         list_for_each_entry_reverse(dentry, dentry_list, extraction_list) {
1159                 ret = dentry_extract_final(dentry, ctx);
1160                 if (ret)
1161                         return ret;
1162         }
1163         return 0;
1164 }
1165
1166 /* Creates a temporary file opened for writing.  The open file descriptor is
1167  * returned in @fd_ret and its name is returned in @name_ret (dynamically
1168  * allocated).  */
1169 static int
1170 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
1171 {
1172         tchar *name;
1173         int raw_fd;
1174         int open_flags;
1175
1176 retry:
1177         name = ttempnam(NULL, T("wimlib"));
1178         if (name == NULL) {
1179                 ERROR_WITH_ERRNO("Failed to create temporary filename");
1180                 return WIMLIB_ERR_NOMEM;
1181         }
1182
1183         open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
1184 #ifdef __WIN32__
1185         open_flags |= _O_SHORT_LIVED;
1186 #endif
1187         raw_fd = topen(name, open_flags, 0600);
1188
1189         if (raw_fd < 0) {
1190                 if (errno == EEXIST) {
1191                         FREE(name);
1192                         goto retry;
1193                 }
1194                 ERROR_WITH_ERRNO("Failed to open temporary file \"%"TS"\"", name);
1195                 FREE(name);
1196                 return WIMLIB_ERR_OPEN;
1197         }
1198
1199         filedes_init(fd_ret, raw_fd);
1200         *name_ret = name;
1201         return 0;
1202 }
1203
1204 /* Extract all instances of the stream @lte that are being extracted in this
1205  * call of extract_tree(), but actually read the stream data from @lte_override.
1206  */
1207 static int
1208 extract_stream_instances(struct wim_lookup_table_entry *lte,
1209                          struct wim_lookup_table_entry *lte_override,
1210                          struct apply_ctx *ctx)
1211 {
1212         struct wim_dentry **lte_dentries;
1213         tchar path[ctx->ops->path_max];
1214         size_t i;
1215         int ret;
1216
1217         if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries))
1218                 lte_dentries = lte->inline_lte_dentries;
1219         else
1220                 lte_dentries = lte->lte_dentries;
1221
1222         for (i = 0; i < lte->out_refcnt; i++) {
1223                 struct wim_dentry *dentry = lte_dentries[i];
1224
1225                 if (dentry->tmp_flag)
1226                         continue;
1227                 if (!build_extraction_path(path, dentry, ctx))
1228                         continue;
1229                 ret = extract_streams(path, ctx, dentry, lte, lte_override);
1230                 if (ret)
1231                         goto out_clear_tmp_flags;
1232                 dentry->tmp_flag = 1;
1233         }
1234         ret = 0;
1235 out_clear_tmp_flags:
1236         for (i = 0; i < lte->out_refcnt; i++)
1237                 lte_dentries[i]->tmp_flag = 0;
1238         return ret;
1239 }
1240
1241 /* Determine whether the specified stream needs to be extracted to a temporary
1242  * file or not.
1243  *
1244  * @lte->out_refcnt specifies the number of instances of this stream that must
1245  * be extracted.
1246  *
1247  * @is_partial_res is %true if this stream is just one of multiple in a single
1248  * WIM resource being extracted.  */
1249 static bool
1250 need_tmpfile_to_extract(struct wim_lookup_table_entry *lte,
1251                         bool is_partial_res)
1252 {
1253         /* Temporary file is always required when reading a partial resource,
1254          * since in that case we retrieve all the contained streams in one pass.
1255          * */
1256         if (is_partial_res)
1257                 return true;
1258
1259         /* Otherwise we don't need a temporary file if only a single instance of
1260          * the stream is needed.  */
1261         if (lte->out_refcnt == 1)
1262                 return false;
1263
1264         wimlib_assert(lte->out_refcnt >= 2);
1265
1266         /* We also don't need a temporary file if random access to the stream is
1267          * allowed.  */
1268         if (lte->resource_location != RESOURCE_IN_WIM ||
1269             filedes_is_seekable(&lte->rspec->wim->in_fd))
1270                 return false;
1271
1272         return true;
1273 }
1274
1275 static int
1276 begin_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1277                                 bool is_partial_res,
1278                                 void *_ctx)
1279 {
1280         struct apply_ctx *ctx = _ctx;
1281         int ret;
1282
1283         if (!need_tmpfile_to_extract(lte, is_partial_res)) {
1284                 DEBUG("Temporary file not needed "
1285                       "for stream (size=%"PRIu64")", lte->size);
1286                 ret = extract_stream_instances(lte, lte, ctx);
1287                 if (ret)
1288                         return ret;
1289
1290                 return BEGIN_STREAM_STATUS_SKIP_STREAM;
1291         }
1292
1293         DEBUG("Temporary file needed for stream (size=%"PRIu64")", lte->size);
1294         return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
1295 }
1296
1297 static int
1298 end_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1299                               int status, void *_ctx)
1300 {
1301         struct apply_ctx *ctx = _ctx;
1302         struct wim_lookup_table_entry lte_override;
1303         int ret;
1304         int errno_save = errno;
1305
1306         ret = filedes_close(&ctx->tmpfile_fd);
1307
1308         if (status) {
1309                 ret = status;
1310                 errno = errno_save;
1311                 goto out_delete_tmpfile;
1312         }
1313
1314         if (ret) {
1315                 ERROR_WITH_ERRNO("Error writing temporary file %"TS, ctx->tmpfile_name);
1316                 ret = WIMLIB_ERR_WRITE;
1317                 goto out_delete_tmpfile;
1318         }
1319
1320         /* Now that a full stream has been extracted to a temporary file,
1321          * extract all instances of it to the actual target.  */
1322
1323         memcpy(&lte_override, lte, sizeof(struct wim_lookup_table_entry));
1324         lte_override.resource_location = RESOURCE_IN_FILE_ON_DISK;
1325         lte_override.file_on_disk = ctx->tmpfile_name;
1326
1327         ret = extract_stream_instances(lte, &lte_override, ctx);
1328
1329 out_delete_tmpfile:
1330         errno_save = errno;
1331         tunlink(ctx->tmpfile_name);
1332         FREE(ctx->tmpfile_name);
1333         errno = errno_save;
1334         return ret;
1335 }
1336
1337 /* Extracts a list of streams (ctx.stream_list), assuming that the directory
1338  * structure and empty files were already created.  This relies on the
1339  * per-`struct wim_lookup_table_entry' list of dentries that reference each
1340  * stream that was constructed earlier.  */
1341 static int
1342 extract_stream_list(struct apply_ctx *ctx)
1343 {
1344         struct read_stream_list_callbacks cbs = {
1345                 .begin_stream           = begin_extract_stream_to_tmpfile,
1346                 .begin_stream_ctx       = ctx,
1347                 .consume_chunk          = extract_chunk_to_fd,
1348                 .consume_chunk_ctx      = &ctx->tmpfile_fd,
1349                 .end_stream             = end_extract_stream_to_tmpfile,
1350                 .end_stream_ctx         = ctx,
1351         };
1352         return read_stream_list(&ctx->stream_list,
1353                                 offsetof(struct wim_lookup_table_entry, extraction_list),
1354                                 &cbs, VERIFY_STREAM_HASHES);
1355 }
1356
1357 #define PWM_ALLOW_WIM_HDR 0x00001
1358 #define PWM_SILENT_EOF    0x00002
1359
1360 /* Read the header from a stream in a pipable WIM.  */
1361 static int
1362 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
1363                        struct wim_resource_spec *rspec,
1364                        int flags, struct wim_header_disk *hdr_ret)
1365 {
1366         union {
1367                 struct pwm_stream_hdr stream_hdr;
1368                 struct wim_header_disk pwm_hdr;
1369         } buf;
1370         struct wim_reshdr reshdr;
1371         int ret;
1372
1373         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
1374         if (ret)
1375                 goto read_error;
1376
1377         if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
1378                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
1379                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
1380                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
1381
1382                 if (ret)
1383                         goto read_error;
1384                 lte->resource_location = RESOURCE_NONEXISTENT;
1385                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
1386                 return 0;
1387         }
1388
1389         if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
1390                 ERROR("Data read on pipe is invalid (expected stream header).");
1391                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1392         }
1393
1394         copy_hash(lte->hash, buf.stream_hdr.hash);
1395
1396         reshdr.size_in_wim = 0;
1397         reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
1398         reshdr.offset_in_wim = pwm->in_fd.offset;
1399         reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
1400         wim_res_hdr_to_spec(&reshdr, pwm, rspec);
1401         lte_bind_wim_resource_spec(lte, rspec);
1402         lte->flags = rspec->flags;
1403         lte->size = rspec->uncompressed_size;
1404         lte->offset_in_res = 0;
1405         return 0;
1406
1407 read_error:
1408         if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
1409                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
1410         return ret;
1411 }
1412
1413 static int
1414 extract_streams_from_pipe(struct apply_ctx *ctx)
1415 {
1416         struct wim_lookup_table_entry *found_lte;
1417         struct wim_resource_spec *rspec;
1418         struct wim_lookup_table_entry *needed_lte;
1419         struct wim_lookup_table *lookup_table;
1420         struct wim_header_disk pwm_hdr;
1421         int ret;
1422         int pwm_flags;
1423
1424         ret = WIMLIB_ERR_NOMEM;
1425         found_lte = new_lookup_table_entry();
1426         if (found_lte == NULL)
1427                 goto out;
1428
1429         rspec = MALLOC(sizeof(struct wim_resource_spec));
1430         if (rspec == NULL)
1431                 goto out_free_found_lte;
1432
1433         lookup_table = ctx->wim->lookup_table;
1434         pwm_flags = PWM_ALLOW_WIM_HDR;
1435         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1436                 pwm_flags |= PWM_SILENT_EOF;
1437         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN);
1438         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
1439         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
1440         if (ctx->progress_func)
1441                 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1442                                    &ctx->progress);
1443         while (ctx->num_streams_remaining) {
1444                 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
1445                         lte_unbind_wim_resource_spec(found_lte);
1446                 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
1447                                              pwm_flags, &pwm_hdr);
1448                 if (ret) {
1449                         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE &&
1450                             (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1451                         {
1452                                 goto resume_done;
1453                         }
1454                         goto out_free_found_lte;
1455                 }
1456
1457                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
1458                     && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
1459                     && (needed_lte = lookup_stream(lookup_table, found_lte->hash))
1460                     && (needed_lte->out_refcnt))
1461                 {
1462                         tchar *tmpfile_name = NULL;
1463                         struct wim_lookup_table_entry *lte_override;
1464                         struct wim_lookup_table_entry tmpfile_lte;
1465
1466                         needed_lte->offset_in_res = found_lte->offset_in_res;
1467                         needed_lte->flags = found_lte->flags;
1468                         needed_lte->size = found_lte->size;
1469
1470                         lte_unbind_wim_resource_spec(found_lte);
1471                         lte_bind_wim_resource_spec(needed_lte, rspec);
1472
1473                         if (needed_lte->out_refcnt > 1) {
1474
1475                                 struct filedes tmpfile_fd;
1476
1477                                 /* Extract stream to temporary file.  */
1478                                 ret = create_temporary_file(&tmpfile_fd, &tmpfile_name);
1479                                 if (ret) {
1480                                         lte_unbind_wim_resource_spec(needed_lte);
1481                                         goto out_free_found_lte;
1482                                 }
1483
1484                                 ret = extract_full_stream_to_fd(needed_lte,
1485                                                                 &tmpfile_fd);
1486                                 if (ret) {
1487                                         filedes_close(&tmpfile_fd);
1488                                         goto delete_tmpfile;
1489                                 }
1490
1491                                 if (filedes_close(&tmpfile_fd)) {
1492                                         ERROR_WITH_ERRNO("Error writing to temporary "
1493                                                          "file \"%"TS"\"", tmpfile_name);
1494                                         ret = WIMLIB_ERR_WRITE;
1495                                         goto delete_tmpfile;
1496                                 }
1497                                 memcpy(&tmpfile_lte, needed_lte,
1498                                        sizeof(struct wim_lookup_table_entry));
1499                                 tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK;
1500                                 tmpfile_lte.file_on_disk = tmpfile_name;
1501                                 lte_override = &tmpfile_lte;
1502                         } else {
1503                                 lte_override = needed_lte;
1504                         }
1505
1506                         ret = extract_stream_instances(needed_lte, lte_override, ctx);
1507                 delete_tmpfile:
1508                         lte_unbind_wim_resource_spec(needed_lte);
1509                         if (tmpfile_name) {
1510                                 tunlink(tmpfile_name);
1511                                 FREE(tmpfile_name);
1512                         }
1513                         if (ret)
1514                                 goto out_free_found_lte;
1515                         ctx->num_streams_remaining--;
1516                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
1517                         ret = skip_wim_stream(found_lte);
1518                         if (ret)
1519                                 goto out_free_found_lte;
1520                 } else {
1521                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
1522                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
1523
1524                         if (part_number != ctx->progress.extract.part_number ||
1525                             total_parts != ctx->progress.extract.total_parts ||
1526                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
1527                                    WIM_GID_LEN))
1528                         {
1529                                 ctx->progress.extract.part_number = part_number;
1530                                 ctx->progress.extract.total_parts = total_parts;
1531                                 memcpy(ctx->progress.extract.guid,
1532                                        pwm_hdr.guid, WIM_GID_LEN);
1533                                 if (ctx->progress_func) {
1534                                         ctx->progress_func(
1535                                                 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1536                                                            &ctx->progress);
1537                                 }
1538
1539                         }
1540                 }
1541         }
1542         ret = 0;
1543 out_free_found_lte:
1544         if (found_lte->resource_location != RESOURCE_IN_WIM)
1545                 FREE(rspec);
1546         free_lookup_table_entry(found_lte);
1547 out:
1548         return ret;
1549
1550 resume_done:
1551         /* TODO */
1552         return 0;
1553 }
1554
1555 /* Extract a WIM dentry to standard output.
1556  *
1557  * This obviously doesn't make sense in all cases.  We return an error if the
1558  * dentry does not correspond to a regular file.  Otherwise we extract the
1559  * unnamed data stream only.  */
1560 static int
1561 extract_dentry_to_stdout(struct wim_dentry *dentry,
1562                          const struct wim_lookup_table *lookup_table)
1563 {
1564         struct wim_inode *inode = dentry->d_inode;
1565         struct wim_lookup_table_entry *lte;
1566         struct filedes _stdout;
1567
1568         if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
1569                                    FILE_ATTRIBUTE_DIRECTORY))
1570         {
1571                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
1572                       "extracted to standard output", dentry_full_path(dentry));
1573                 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
1574         }
1575
1576         lte = inode_unnamed_lte(inode, lookup_table);
1577         if (!lte) {
1578                 const u8 *hash = inode_unnamed_stream_hash(inode);
1579                 if (!is_zero_hash(hash))
1580                         return stream_not_found_error(inode, hash);
1581                 return 0;
1582         }
1583
1584         filedes_init(&_stdout, STDOUT_FILENO);
1585         return extract_full_stream_to_fd(lte, &_stdout);
1586 }
1587
1588 static int
1589 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
1590                            const struct wim_lookup_table *lookup_table)
1591 {
1592         for (size_t i = 0; i < num_dentries; i++) {
1593                 int ret = extract_dentry_to_stdout(dentries[i], lookup_table);
1594                 if (ret)
1595                         return ret;
1596         }
1597         return 0;
1598 }
1599
1600 /**********************************************************************/
1601
1602 /*
1603  * Removes duplicate dentries from the array.
1604  *
1605  * Returns the new number of dentries, packed at the front of the array.
1606  */
1607 static size_t
1608 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
1609 {
1610         size_t i, j = 0;
1611         for (i = 0; i < num_trees; i++) {
1612                 if (!trees[i]->tmp_flag) {
1613                         /* Found distinct dentry.  */
1614                         trees[i]->tmp_flag = 1;
1615                         trees[j++] = trees[i];
1616                 }
1617         }
1618         for (i = 0; i < j; i++)
1619                 trees[i]->tmp_flag = 0;
1620         return j;
1621 }
1622
1623 /*
1624  * Remove dentries that are descendants of other dentries in the array.
1625  *
1626  * Returns the new number of dentries, packed at the front of the array.
1627  */
1628 static size_t
1629 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
1630 {
1631         size_t i, j = 0;
1632         for (i = 0; i < num_trees; i++)
1633                 trees[i]->tmp_flag = 1;
1634         for (i = 0; i < num_trees; i++) {
1635                 struct wim_dentry *d = trees[i];
1636                 while (!dentry_is_root(d)) {
1637                         d = d->parent;
1638                         if (d->tmp_flag)
1639                                 goto tree_contained;
1640                 }
1641                 trees[j++] = trees[i];
1642                 continue;
1643
1644         tree_contained:
1645                 trees[i]->tmp_flag = 0;
1646         }
1647
1648         for (i = 0; i < j; i++)
1649                 trees[i]->tmp_flag = 0;
1650         return j;
1651 }
1652
1653 static int
1654 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
1655 {
1656         struct list_head *dentry_list = _dentry_list;
1657         list_add_tail(&dentry->extraction_list, dentry_list);
1658         return 0;
1659 }
1660
1661 static void
1662 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
1663 {
1664         dentry->extraction_list = (struct list_head){NULL, NULL};
1665 }
1666
1667 static int
1668 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
1669 {
1670         list_del(&dentry->extraction_list);
1671         dentry_reset_extraction_list_node(dentry);
1672         return 0;
1673 }
1674
1675 /*
1676  * Build the preliminary list of dentries to be extracted.
1677  *
1678  * The list maintains the invariant that if d1 and d2 are in the list and d1 is
1679  * an ancestor of d2, then d1 appears before d2 in the list.
1680  */
1681 static void
1682 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
1683                   size_t num_trees, bool add_ancestors)
1684 {
1685         INIT_LIST_HEAD(dentry_list);
1686
1687         /* Add the trees recursively.  */
1688         for (size_t i = 0; i < num_trees; i++)
1689                 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
1690
1691         /* If requested, add ancestors of the trees.  */
1692         if (add_ancestors) {
1693                 for (size_t i = 0; i < num_trees; i++) {
1694                         struct wim_dentry *dentry = trees[i];
1695                         struct wim_dentry *ancestor;
1696                         struct list_head *place_after;
1697
1698                         if (dentry_is_root(dentry))
1699                                 continue;
1700
1701                         place_after = dentry_list;
1702                         ancestor = dentry;
1703                         do {
1704                                 ancestor = ancestor->parent;
1705                                 if (dentry_in_list(ancestor)) {
1706                                         place_after = &ancestor->extraction_list;
1707                                         break;
1708                                 }
1709                         } while (!dentry_is_root(ancestor));
1710
1711                         ancestor = dentry;
1712                         do {
1713                                 ancestor = ancestor->parent;
1714                                 if (dentry_in_list(ancestor))
1715                                         break;
1716                                 list_add(&ancestor->extraction_list, place_after);
1717                         } while (!dentry_is_root(ancestor));
1718                 }
1719         }
1720 }
1721
1722 static const struct apply_operations *
1723 select_apply_operations(int extract_flags)
1724 {
1725 #ifdef WITH_NTFS_3G
1726         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1727                 return &ntfs_3g_apply_ops;
1728 #endif
1729 #ifdef __WIN32__
1730         return &win32_apply_ops;
1731 #else
1732         return &unix_apply_ops;
1733 #endif
1734 }
1735
1736 #ifdef __WIN32__
1737 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
1738 #else
1739 static const utf16lechar replacement_char = cpu_to_le16('?');
1740 #endif
1741
1742 static bool
1743 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
1744 {
1745         size_t i;
1746
1747         if (num_chars == 0)
1748                 return true;
1749         for (i = 0; i < num_chars; i++) {
1750                 switch (name[i]) {
1751         #ifdef __WIN32__
1752                 case cpu_to_le16('\\'):
1753                 case cpu_to_le16(':'):
1754                 case cpu_to_le16('*'):
1755                 case cpu_to_le16('?'):
1756                 case cpu_to_le16('"'):
1757                 case cpu_to_le16('<'):
1758                 case cpu_to_le16('>'):
1759                 case cpu_to_le16('|'):
1760         #endif
1761                 case cpu_to_le16('/'):
1762                 case cpu_to_le16('\0'):
1763                         if (fix)
1764                                 name[i] = replacement_char;
1765                         else
1766                                 return false;
1767                 }
1768         }
1769
1770 #ifdef __WIN32__
1771         if (name[num_chars - 1] == cpu_to_le16(' ') ||
1772             name[num_chars - 1] == cpu_to_le16('.'))
1773         {
1774                 if (fix)
1775                         name[num_chars - 1] = replacement_char;
1776                 else
1777                         return false;
1778         }
1779 #endif
1780         return true;
1781 }
1782
1783 static int
1784 dentry_calculate_extraction_name(struct wim_dentry *dentry,
1785                                  struct apply_ctx *ctx)
1786 {
1787         int ret;
1788
1789         if (dentry == ctx->target_dentry)
1790                 return 0;
1791
1792         if (!dentry_is_supported(dentry, &ctx->supported_features))
1793                 goto skip_dentry;
1794
1795         if (!ctx->ops->supports_case_sensitive_filenames)
1796         {
1797                 struct wim_dentry *other;
1798                 list_for_each_entry(other, &dentry->d_ci_conflict_list,
1799                                     d_ci_conflict_list)
1800                 {
1801                         if (dentry_in_list(other)) {
1802                                 if (ctx->extract_flags &
1803                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
1804                                         WARNING("\"%"TS"\" has the same "
1805                                                 "case-insensitive name as "
1806                                                 "\"%"TS"\"; extracting "
1807                                                 "dummy name instead",
1808                                                 dentry_full_path(dentry),
1809                                                 dentry_full_path(other));
1810                                         goto out_replace;
1811                                 } else {
1812                                         WARNING("Not extracting \"%"TS"\": "
1813                                                 "has same case-insensitive "
1814                                                 "name as \"%"TS"\"",
1815                                                 dentry_full_path(dentry),
1816                                                 dentry_full_path(other));
1817                                         goto skip_dentry;
1818                                 }
1819                         }
1820                 }
1821         }
1822
1823         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
1824 #if TCHAR_IS_UTF16LE
1825                 dentry->extraction_name = dentry->file_name;
1826                 dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
1827                 return 0;
1828 #else
1829                 return utf16le_to_tstr(dentry->file_name,
1830                                        dentry->file_name_nbytes,
1831                                        &dentry->extraction_name,
1832                                        &dentry->extraction_name_nchars);
1833 #endif
1834         } else {
1835                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
1836                 {
1837                         WARNING("\"%"TS"\" has an invalid filename "
1838                                 "that is not supported on this platform; "
1839                                 "extracting dummy name instead",
1840                                 dentry_full_path(dentry));
1841                         goto out_replace;
1842                 } else {
1843                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
1844                                 "that is not supported on this platform",
1845                                 dentry_full_path(dentry));
1846                         goto skip_dentry;
1847                 }
1848         }
1849
1850 out_replace:
1851         {
1852                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
1853
1854                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
1855                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
1856
1857                 tchar *tchar_name;
1858                 size_t tchar_nchars;
1859         #if TCHAR_IS_UTF16LE
1860                 tchar_name = utf16_name_copy;
1861                 tchar_nchars = dentry->file_name_nbytes / 2;
1862         #else
1863                 ret = utf16le_to_tstr(utf16_name_copy,
1864                                       dentry->file_name_nbytes,
1865                                       &tchar_name, &tchar_nchars);
1866                 if (ret)
1867                         return ret;
1868         #endif
1869                 size_t fixed_name_num_chars = tchar_nchars;
1870                 tchar fixed_name[tchar_nchars + 50];
1871
1872                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
1873                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
1874                                                  T(" (invalid filename #%lu)"),
1875                                                  ++ctx->invalid_sequence);
1876         #if !TCHAR_IS_UTF16LE
1877                 FREE(tchar_name);
1878         #endif
1879                 dentry->extraction_name = memdup(fixed_name,
1880                                                  2 * fixed_name_num_chars + 2);
1881                 if (!dentry->extraction_name)
1882                         return WIMLIB_ERR_NOMEM;
1883                 dentry->extraction_name_nchars = fixed_name_num_chars;
1884         }
1885         return 0;
1886
1887 skip_dentry:
1888         for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
1889         return 0;
1890 }
1891
1892 /*
1893  * Calculate the actual filename component at which each WIM dentry will be
1894  * extracted, with special handling for dentries that are unsupported by the
1895  * extraction backend or have invalid names.
1896  *
1897  * Note: this has a dependency on start_extract() being called because
1898  * ctx.supported_features must be filled in in order to determine whether each
1899  * dentry is supported.
1900  *
1901  * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
1902  */
1903 static int
1904 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
1905                                        struct apply_ctx *ctx)
1906 {
1907         struct list_head *prev, *cur;
1908
1909         /* Can't use list_for_each_entry() because a call to
1910          * dentry_calculate_extraction_name() may delete the current dentry and
1911          * its children from the list.  */
1912
1913         prev = dentry_list;
1914         for (;;) {
1915                 struct wim_dentry *dentry;
1916                 int ret;
1917
1918                 cur = prev->next;
1919                 if (cur == dentry_list)
1920                         break;
1921
1922                 dentry = list_entry(cur, struct wim_dentry, extraction_list);
1923
1924                 ret = dentry_calculate_extraction_name(dentry, ctx);
1925                 if (ret)
1926                         return ret;
1927
1928                 if (prev->next == cur)
1929                         prev = cur;
1930                 else
1931                         ; /* Current dentry and its children (which follow in
1932                              the list) were deleted.  prev stays the same.  */
1933         }
1934         return 0;
1935 }
1936
1937 static int
1938 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
1939                        struct wim_lookup_table *lookup_table)
1940 {
1941         struct wim_inode *inode = dentry->d_inode;
1942         struct wim_lookup_table_entry *lte;
1943         int ret;
1944         bool force = false;
1945
1946         /* Special case:  when extracting from a pipe, the WIM lookup table is
1947          * initially empty, so "resolving" an inode's streams is initially not
1948          * possible.  However, we still need to keep track of which streams,
1949          * identified by SHA1 message digests, need to be extracted, so we
1950          * "resolve" the inode's streams anyway by allocating new entries.  */
1951         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
1952                 force = true;
1953         ret = inode_resolve_streams(inode, lookup_table, force);
1954         if (ret)
1955                 return ret;
1956         for (u32 i = 0; i <= inode->i_num_ads; i++) {
1957                 lte = inode_stream_lte_resolved(inode, i);
1958                 if (lte)
1959                         lte->out_refcnt = 0;
1960         }
1961         return 0;
1962 }
1963
1964 /*
1965  * For each dentry to be extracted, resolve all streams in the corresponding
1966  * inode and set 'out_refcnt' in each to 0.
1967  *
1968  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
1969  */
1970 static int
1971 dentry_list_resolve_streams(struct list_head *dentry_list,
1972                             struct apply_ctx *ctx)
1973 {
1974         struct wim_dentry *dentry;
1975         int ret;
1976
1977         list_for_each_entry(dentry, dentry_list, extraction_list) {
1978                 ret = dentry_resolve_streams(dentry,
1979                                              ctx->extract_flags,
1980                                              ctx->wim->lookup_table);
1981                 if (ret)
1982                         return ret;
1983         }
1984         return 0;
1985 }
1986
1987 static int
1988 ref_stream(struct wim_lookup_table_entry *lte,
1989            struct wim_dentry *dentry, struct apply_ctx *ctx)
1990 {
1991         if (!lte)
1992                 return 0;
1993
1994         /* Tally the size only for each extraction of the stream (not hard
1995          * links).  */
1996         if (!(dentry->d_inode->i_visited &&
1997               ctx->supported_features.hard_links) &&
1998             (!is_linked_extraction(ctx) || (lte->out_refcnt == 0 &&
1999                                             lte->extracted_file == NULL)))
2000         {
2001                 ctx->progress.extract.total_bytes += lte->size;
2002                 ctx->progress.extract.num_streams++;
2003         }
2004
2005         /* Add stream to the dentry_list only one time, even if it's going
2006          * to be extracted to multiple locations.  */
2007         if (lte->out_refcnt == 0) {
2008                 list_add_tail(&lte->extraction_list, &ctx->stream_list);
2009                 ctx->num_streams_remaining++;
2010         }
2011
2012         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
2013                 struct wim_dentry **lte_dentries;
2014
2015                 /* Append dentry to this stream's array of dentries referencing
2016                  * it.  Use inline array to avoid memory allocation until the
2017                  * number of dentries becomes too large.  */
2018                 if (lte->out_refcnt < ARRAY_LEN(lte->inline_lte_dentries)) {
2019                         lte_dentries = lte->inline_lte_dentries;
2020                 } else {
2021                         struct wim_dentry **prev_lte_dentries;
2022                         size_t alloc_lte_dentries;
2023
2024                         if (lte->out_refcnt == ARRAY_LEN(lte->inline_lte_dentries)) {
2025                                 prev_lte_dentries = NULL;
2026                                 alloc_lte_dentries = ARRAY_LEN(lte->inline_lte_dentries);
2027                         } else {
2028                                 prev_lte_dentries = lte->lte_dentries;
2029                                 alloc_lte_dentries = lte->alloc_lte_dentries;
2030                         }
2031
2032                         if (lte->out_refcnt == alloc_lte_dentries) {
2033                                 alloc_lte_dentries *= 2;
2034                                 lte_dentries = REALLOC(prev_lte_dentries,
2035                                                        alloc_lte_dentries *
2036                                                         sizeof(lte_dentries[0]));
2037                                 if (lte_dentries == NULL)
2038                                         return WIMLIB_ERR_NOMEM;
2039                                 if (prev_lte_dentries == NULL) {
2040                                         memcpy(lte_dentries,
2041                                                lte->inline_lte_dentries,
2042                                                sizeof(lte->inline_lte_dentries));
2043                                 }
2044                                 lte->lte_dentries = lte_dentries;
2045                                 lte->alloc_lte_dentries = alloc_lte_dentries;
2046                         }
2047                         lte_dentries = lte->lte_dentries;
2048                 }
2049                 lte_dentries[lte->out_refcnt] = dentry;
2050         }
2051         lte->out_refcnt++;
2052         return 0;
2053 }
2054
2055 static int
2056 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
2057 {
2058         struct wim_inode *inode = dentry->d_inode;
2059         int ret;
2060
2061         /* The unnamed data stream will always be extracted, except in an
2062          * unlikely case.  */
2063         if (!inode_is_encrypted_directory(inode)) {
2064                 ret = ref_stream(inode_unnamed_lte_resolved(inode),
2065                                  dentry, ctx);
2066                 if (ret)
2067                         return ret;
2068         }
2069
2070         /* Named data streams will be extracted only if supported in the current
2071          * extraction mode and volume, and to avoid complications, if not doing
2072          * a linked extraction.  */
2073         if (can_extract_named_data_streams(ctx)) {
2074                 for (u16 i = 0; i < inode->i_num_ads; i++) {
2075                         if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
2076                                 continue;
2077                         ret = ref_stream(inode->i_ads_entries[i].lte,
2078                                          dentry, ctx);
2079                         if (ret)
2080                                 return ret;
2081                 }
2082         }
2083         inode->i_visited = 1;
2084         return 0;
2085 }
2086
2087 /*
2088  * For each dentry to be extracted, iterate through the data streams of the
2089  * corresponding inode.  For each such stream that is not to be ignored due to
2090  * the supported features or extraction flags, add it to the list of streams to
2091  * be extracted (ctx->stream_list) if not already done so.
2092  *
2093  * Also, if doing a sequential extraction, build a mapping from each stream to
2094  * the dentries referencing it.
2095  *
2096  * This also initializes the extract progress info with byte and stream
2097  * information.
2098  *
2099  * Note: This has a dependency on start_extract being called because
2100  * ctx.supported_features must be filled in in order to determine whether named
2101  * data streams are supported.
2102  *
2103  * Note: this uses the i_visited member of the inodes (assumed to be 0
2104  * initially), but does not reset it.
2105  *
2106  * Possible error codes: WIMLIB_ERR_NOMEM.
2107  */
2108 static int
2109 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
2110 {
2111         struct wim_dentry *dentry;
2112         int ret;
2113
2114         list_for_each_entry(dentry, dentry_list, extraction_list) {
2115                 ret = dentry_ref_streams(dentry, ctx);
2116                 if (ret)
2117                         return ret;
2118         }
2119         return 0;
2120 }
2121
2122 /* Tally features necessary to extract a dentry and the corresponding inode.  */
2123 static void
2124 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
2125 {
2126         struct wim_inode *inode = dentry->d_inode;
2127
2128         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
2129                 features->archive_files++;
2130         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
2131                 features->hidden_files++;
2132         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
2133                 features->system_files++;
2134         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
2135                 features->compressed_files++;
2136         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
2137                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
2138                         features->encrypted_directories++;
2139                 else
2140                         features->encrypted_files++;
2141         }
2142         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
2143                 features->not_context_indexed_files++;
2144         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
2145                 features->sparse_files++;
2146         if (inode_has_named_stream(inode))
2147                 features->named_data_streams++;
2148         if (inode->i_visited)
2149                 features->hard_links++;
2150         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
2151                 features->reparse_points++;
2152                 if (inode_is_symlink(inode))
2153                         features->symlink_reparse_points++;
2154                 else
2155                         features->other_reparse_points++;
2156         }
2157         if (inode->i_security_id != -1)
2158                 features->security_descriptors++;
2159         if (dentry->short_name_nbytes)
2160                 features->short_names++;
2161         if (inode_has_unix_data(inode))
2162                 features->unix_data++;
2163         inode->i_visited = 1;
2164 }
2165
2166 /* Tally the features necessary to extract the specified dentries.  */
2167 static void
2168 dentry_list_get_features(struct list_head *dentry_list,
2169                          struct wim_features *features)
2170 {
2171         struct wim_dentry *dentry;
2172
2173         memset(features, 0, sizeof(struct wim_features));
2174
2175         list_for_each_entry(dentry, dentry_list, extraction_list)
2176                 dentry_tally_features(dentry, features);
2177
2178         list_for_each_entry(dentry, dentry_list, extraction_list)
2179                 dentry->d_inode->i_visited = 0;
2180 }
2181
2182 static u32
2183 compute_supported_attributes_mask(const struct wim_features *supported_features)
2184 {
2185         u32 mask = (u32)~0UL;
2186
2187         if (!supported_features->archive_files)
2188                 mask &= ~FILE_ATTRIBUTE_ARCHIVE;
2189
2190         if (!supported_features->hidden_files)
2191                 mask &= ~FILE_ATTRIBUTE_HIDDEN;
2192
2193         if (!supported_features->system_files)
2194                 mask &= ~FILE_ATTRIBUTE_SYSTEM;
2195
2196         if (!supported_features->not_context_indexed_files)
2197                 mask &= ~FILE_ATTRIBUTE_NOT_CONTENT_INDEXED;
2198
2199         if (!supported_features->compressed_files)
2200                 mask &= ~FILE_ATTRIBUTE_COMPRESSED;
2201
2202         if (!supported_features->sparse_files)
2203                 mask &= ~FILE_ATTRIBUTE_SPARSE_FILE;
2204
2205         if (!supported_features->reparse_points)
2206                 mask &= ~FILE_ATTRIBUTE_REPARSE_POINT;
2207
2208         return mask;
2209 }
2210
2211 static int
2212 do_feature_check(const struct wim_features *required_features,
2213                  const struct wim_features *supported_features,
2214                  int extract_flags, const struct apply_operations *ops)
2215 {
2216         /* File attributes.  */
2217         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
2218                 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
2219                  * We're an archive program, so theoretically we can do what we
2220                  * want with it.  */
2221
2222                 if (required_features->hidden_files &&
2223                     !supported_features->hidden_files)
2224                         WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
2225                                 required_features->hidden_files);
2226
2227                 if (required_features->system_files &&
2228                     !supported_features->system_files)
2229                         WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
2230                                 required_features->system_files);
2231
2232                 if (required_features->compressed_files &&
2233                     !supported_features->compressed_files)
2234                         WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
2235                                 required_features->compressed_files);
2236
2237                 if (required_features->not_context_indexed_files &&
2238                     !supported_features->not_context_indexed_files)
2239                         WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
2240                                 required_features->not_context_indexed_files);
2241
2242                 if (required_features->sparse_files &&
2243                     !supported_features->sparse_files)
2244                         WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
2245                                 required_features->sparse_files);
2246
2247                 if (required_features->encrypted_directories &&
2248                     !supported_features->encrypted_directories)
2249                         WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
2250                                 required_features->encrypted_directories);
2251         }
2252
2253         /* Encrypted files.  */
2254         if (required_features->encrypted_files &&
2255             !supported_features->encrypted_files)
2256                 WARNING("Ignoring %lu encrypted files",
2257                         required_features->encrypted_files);
2258
2259         /* Named data streams.  */
2260         if (required_features->named_data_streams &&
2261             (!supported_features->named_data_streams ||
2262              (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2263                                WIMLIB_EXTRACT_FLAG_HARDLINK))))
2264                 WARNING("Ignoring named data streams of %lu files",
2265                         required_features->named_data_streams);
2266
2267         /* Hard links.  */
2268         if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) &&
2269             !supported_features->hard_links)
2270         {
2271                 ERROR("Extraction backend does not support hard links!");
2272                 return WIMLIB_ERR_UNSUPPORTED;
2273         }
2274         if (required_features->hard_links && !supported_features->hard_links)
2275                 WARNING("Extracting %lu hard links as independent files",
2276                         required_features->hard_links);
2277
2278         /* Symbolic links and reparse points.  */
2279         if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) &&
2280             !supported_features->symlink_reparse_points)
2281         {
2282                 ERROR("Extraction backend does not support symbolic links!");
2283                 return WIMLIB_ERR_UNSUPPORTED;
2284         }
2285         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
2286             required_features->symlink_reparse_points &&
2287             !supported_features->symlink_reparse_points &&
2288             !supported_features->reparse_points)
2289         {
2290                 ERROR("Extraction backend does not support symbolic links!");
2291                 return WIMLIB_ERR_UNSUPPORTED;
2292         }
2293         if (required_features->reparse_points &&
2294             !supported_features->reparse_points)
2295         {
2296                 if (supported_features->symlink_reparse_points) {
2297                         if (required_features->other_reparse_points) {
2298                                 WARNING("Ignoring %lu non-symlink/junction "
2299                                         "reparse point files",
2300                                         required_features->other_reparse_points);
2301                         }
2302                 } else {
2303                         WARNING("Ignoring %lu reparse point files",
2304                                 required_features->reparse_points);
2305                 }
2306         }
2307
2308         /* Security descriptors.  */
2309         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
2310                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
2311              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
2312             required_features->security_descriptors &&
2313             !supported_features->security_descriptors)
2314         {
2315                 ERROR("Extraction backend does not support security descriptors!");
2316                 return WIMLIB_ERR_UNSUPPORTED;
2317         }
2318         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
2319             required_features->security_descriptors &&
2320             !supported_features->security_descriptors)
2321                 WARNING("Ignoring Windows NT security descriptors of %lu files",
2322                         required_features->security_descriptors);
2323
2324         /* UNIX data.  */
2325         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
2326             required_features->unix_data && !supported_features->unix_data)
2327         {
2328                 ERROR("Extraction backend does not support UNIX data!");
2329                 return WIMLIB_ERR_UNSUPPORTED;
2330         }
2331
2332         /* DOS Names.  */
2333         if (required_features->short_names &&
2334             !supported_features->short_names)
2335         {
2336                 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
2337                         ERROR("Extraction backend does not support DOS names!");
2338                         return WIMLIB_ERR_UNSUPPORTED;
2339                 }
2340                 WARNING("Ignoring DOS names of %lu files",
2341                         required_features->short_names);
2342         }
2343
2344         /* Timestamps.  */
2345         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
2346             !ops->set_timestamps)
2347         {
2348                 ERROR("Extraction backend does not support timestamps!");
2349                 return WIMLIB_ERR_UNSUPPORTED;
2350         }
2351
2352         return 0;
2353 }
2354
2355 static void
2356 do_extract_warnings(struct apply_ctx *ctx)
2357 {
2358         if (ctx->partial_security_descriptors == 0 &&
2359             ctx->no_security_descriptors == 0)
2360                 return;
2361
2362         WARNING("Extraction to \"%"TS"\" complete, but with one or more warnings:",
2363                 ctx->target);
2364         if (ctx->partial_security_descriptors != 0) {
2365                 WARNING("- Could only partially set the security descriptor\n"
2366                         "            on %lu files or directories.",
2367                         ctx->partial_security_descriptors);
2368         }
2369         if (ctx->no_security_descriptors != 0) {
2370                 WARNING("- Could not set security descriptor at all\n"
2371                         "            on %lu files or directories.",
2372                         ctx->no_security_descriptors);
2373         }
2374 #ifdef __WIN32__
2375         WARNING("To fully restore all security descriptors, run the program\n"
2376                 "          with Administrator rights.");
2377 #endif
2378 }
2379
2380 static void
2381 destroy_dentry_list(struct list_head *dentry_list)
2382 {
2383         struct wim_dentry *dentry, *tmp;
2384         struct wim_inode *inode;
2385
2386         list_for_each_entry_safe(dentry, tmp, dentry_list, extraction_list) {
2387                 inode = dentry->d_inode;
2388                 dentry_reset_extraction_list_node(dentry);
2389                 dentry->was_linked = 0;
2390                 dentry->skeleton_extracted = 0;
2391                 inode->i_visited = 0;
2392                 FREE(inode->i_extracted_file);
2393                 inode->i_extracted_file = NULL;
2394                 inode->i_dos_name_extracted = 0;
2395                 if ((void*)dentry->extraction_name != (void*)dentry->file_name)
2396                         FREE(dentry->extraction_name);
2397                 dentry->extraction_name = NULL;
2398         }
2399 }
2400
2401 static void
2402 destroy_stream_list(struct list_head *stream_list)
2403 {
2404         struct wim_lookup_table_entry *lte;
2405
2406         list_for_each_entry(lte, stream_list, extraction_list)
2407                 if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries))
2408                         FREE(lte->lte_dentries);
2409 }
2410
2411 static int
2412 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
2413               const tchar *target, int extract_flags,
2414               wimlib_progress_func_t progress_func)
2415 {
2416         int ret;
2417         struct apply_ctx ctx;
2418         struct list_head dentry_list;
2419         struct wim_features required_features;
2420
2421         /* Handle stdout extraction as a separate case.  */
2422         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT)
2423                 return extract_dentries_to_stdout(trees, num_trees,
2424                                                   wim->lookup_table);
2425
2426         /* Start initializing the apply_ctx.  */
2427         memset(&ctx, 0, sizeof(struct apply_ctx));
2428         ctx.wim = wim;
2429         ctx.extract_flags = extract_flags;
2430         ctx.target = target;
2431         ctx.target_nchars = tstrlen(target);
2432         ctx.progress_func = progress_func;
2433         if (progress_func) {
2434                 ctx.progress.extract.wimfile_name = wim->filename;
2435                 ctx.progress.extract.image = wim->current_image;
2436                 ctx.progress.extract.extract_flags = (extract_flags &
2437                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
2438                 ctx.progress.extract.image_name = wimlib_get_image_name(wim,
2439                                                                         wim->current_image);
2440                 ctx.progress.extract.target = target;
2441         }
2442
2443         ctx.target_dentry = wim_root_dentry(wim);
2444         /* Note: ctx.target_dentry represents the dentry that gets extracted to
2445          * @target.  There may be none, in which case it gets set to the image
2446          * root and never matches any of the dentries actually being extracted.
2447          */
2448
2449         num_trees = remove_duplicate_trees(trees, num_trees);
2450
2451         /* All trees are now distinct.  */
2452
2453         num_trees = remove_contained_trees(trees, num_trees);
2454
2455         /* All trees are now distinct and non-overlapping.  */
2456
2457         /* Build list of dentries to be extracted.  */
2458         build_dentry_list(&dentry_list, trees, num_trees,
2459                           !(extract_flags & WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
2460
2461         /* Select the appropriate apply_operations based on the platform and
2462          * extract_flags.  */
2463         ctx.ops = select_apply_operations(extract_flags);
2464
2465         /* Figure out whether the root dentry is being extracted to the root of
2466          * a volume and therefore needs to be treated "specially", for example
2467          * not being explicitly created and not having attributes set.  */
2468         if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special)
2469                 ctx.root_dentry_is_special = ctx.ops->target_is_root(target);
2470
2471         /* Call the start_extract() callback.  This gives the apply_operations
2472          * implementation a chance to do any setup needed to access the volume.
2473          * Furthermore, start_extract() is expected to set the supported
2474          * features of this extraction mode (ctx.supported_features), which are
2475          * determined at runtime as they may vary depending on the actual
2476          * volume.  */
2477         ret = ctx.ops->start_extract(target, &ctx);
2478         if (ret)
2479                 goto out_destroy_dentry_list;
2480
2481         /* Get and check the features required to extract the dentries.  */
2482         dentry_list_get_features(&dentry_list, &required_features);
2483         ret = do_feature_check(&required_features, &ctx.supported_features,
2484                                extract_flags, ctx.ops);
2485         if (ret)
2486                 goto out_finish_or_abort_extract;
2487
2488         ctx.supported_attributes_mask =
2489                 compute_supported_attributes_mask(&ctx.supported_features);
2490
2491         /* Calculate extraction name for each dentry and remove subtrees that
2492          * can't be extracted due to naming problems.  */
2493         ret = dentry_list_calculate_extraction_names(&dentry_list, &ctx);
2494         if (ret)
2495                 goto out_finish_or_abort_extract;
2496
2497         /* Build list of streams to extract.  */
2498         ret = dentry_list_resolve_streams(&dentry_list, &ctx);
2499         if (ret)
2500                 goto out_finish_or_abort_extract;
2501         INIT_LIST_HEAD(&ctx.stream_list);
2502         ret = dentry_list_ref_streams(&dentry_list, &ctx);
2503         if (ret)
2504                 goto out_destroy_stream_list;
2505
2506         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2507                 /* When extracting from a pipe, the number of bytes of data to
2508                  * extract can't be determined in the normal way (examining the
2509                  * lookup table), since at this point all we have is a set of
2510                  * SHA1 message digests of streams that need to be extracted.
2511                  * However, we can get a reasonably accurate estimate by taking
2512                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
2513                  * data.  This does assume that a full image is being extracted,
2514                  * but currently there is no API for doing otherwise.  (Also,
2515                  * subtract <HARDLINKBYTES> from this if hard links are
2516                  * supported by the extraction mode.)  */
2517                 ctx.progress.extract.total_bytes =
2518                         wim_info_get_image_total_bytes(wim->wim_info,
2519                                                        wim->current_image);
2520                 if (ctx.supported_features.hard_links) {
2521                         ctx.progress.extract.total_bytes -=
2522                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
2523                                                                    wim->current_image);
2524                 }
2525         }
2526
2527         if (ctx.ops->realpath_works_on_nonexisting_files &&
2528             ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) ||
2529              ctx.ops->requires_realtarget_in_paths))
2530         {
2531                 ctx.realtarget = realpath(target, NULL);
2532                 if (!ctx.realtarget) {
2533                         ret = WIMLIB_ERR_NOMEM;
2534                         goto out_destroy_stream_list;
2535                 }
2536                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2537         #ifdef __WIN32__
2538                /* Strip trailing slashes.  If we don't do this, we may create a
2539                 * path with multiple consecutive backslashes, which for some
2540                 * reason causes Windows to report that the file cannot be found.
2541                 */
2542                 while (ctx.realtarget_nchars >= 2
2543                        && ctx.realtarget[ctx.realtarget_nchars - 1] == L'\\'
2544                        && ctx.realtarget[ctx.realtarget_nchars - 2] != L':')
2545                 {
2546                         ctx.realtarget[--ctx.realtarget_nchars] = L'\0';
2547                 }
2548         #endif
2549         }
2550
2551         if (progress_func) {
2552                 int msg;
2553                 if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE)
2554                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
2555                 else
2556                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN;
2557                 progress_func(msg, &ctx.progress);
2558         }
2559
2560         if (!ctx.root_dentry_is_special) {
2561                 tchar path[ctx.ops->path_max];
2562                 if (build_extraction_path(path, ctx.target_dentry, &ctx)) {
2563                         ret = extract_inode(path, &ctx, ctx.target_dentry->d_inode);
2564                         if (ret)
2565                                 goto out_free_realtarget;
2566                 }
2567         }
2568
2569         /* If we need to fix up the targets of absolute symbolic links
2570          * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to
2571          * be absolute, use realpath() (or its replacement on Windows) to get
2572          * the absolute path to the extraction target.  Note that this requires
2573          * the target directory to exist, unless
2574          * realpath_works_on_nonexisting_files is set in the apply_operations.
2575          * */
2576         if (!ctx.realtarget &&
2577             (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
2578               required_features.symlink_reparse_points) ||
2579              ctx.ops->requires_realtarget_in_paths))
2580         {
2581                 ctx.realtarget = realpath(target, NULL);
2582                 if (!ctx.realtarget) {
2583                         ret = WIMLIB_ERR_NOMEM;
2584                         goto out_free_realtarget;
2585                 }
2586                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2587         }
2588
2589         if (ctx.ops->requires_short_name_reordering) {
2590                 if (progress_func)
2591                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2592                                       &ctx.progress);
2593                 ret = extract_dir_structure(&dentry_list, &ctx);
2594                 if (ret)
2595                         goto out_free_realtarget;
2596         }
2597
2598         /* Finally, the important part: extract the tree of files.  */
2599         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
2600                 /* Sequential extraction requested, so two passes are needed
2601                  * (one for file structure, one for streams.)  */
2602                 if (progress_func && !ctx.ops->requires_short_name_reordering)
2603                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2604                                       &ctx.progress);
2605
2606                 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) {
2607                         ret = extract_structure(&dentry_list, &ctx);
2608                         if (ret)
2609                                 goto out_free_realtarget;
2610                 }
2611                 if (progress_func)
2612                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2613                                       &ctx.progress);
2614                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
2615                         ret = extract_streams_from_pipe(&ctx);
2616                 else
2617                         ret = extract_stream_list(&ctx);
2618                 if (ret)
2619                         goto out_free_realtarget;
2620         } else {
2621                 /* Sequential extraction was not requested, so we can make do
2622                  * with one pass where we both create the files and extract
2623                  * streams.   */
2624                 if (progress_func && !ctx.ops->requires_short_name_reordering)
2625                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2626                                       &ctx.progress);
2627                 ret = extract_dentries(&dentry_list, &ctx);
2628                 if (ret)
2629                         goto out_free_realtarget;
2630                 if (progress_func)
2631                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2632                                       &ctx.progress);
2633         }
2634
2635         /* If the total number of bytes to extract was miscalculated, just jump
2636          * to the calculated number in order to avoid confusing the progress
2637          * function.  This should only occur when extracting from a pipe.  */
2638         if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes)
2639         {
2640                 DEBUG("Calculated %"PRIu64" bytes to extract, but actually "
2641                       "extracted %"PRIu64,
2642                       ctx.progress.extract.total_bytes,
2643                       ctx.progress.extract.completed_bytes);
2644         }
2645         if (progress_func &&
2646             ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes)
2647         {
2648                 ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes;
2649                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress);
2650         }
2651
2652         /* Apply security descriptors and timestamps.  This is done at the end,
2653          * and in a depth-first manner, to prevent timestamps from getting
2654          * changed by subsequent extract operations and to minimize the chance
2655          * of the restored security descriptors getting in our way.  */
2656         if (progress_func)
2657                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
2658                               &ctx.progress);
2659         ret = extract_final_metadata(&dentry_list, &ctx);
2660         if (ret)
2661                 goto out_free_realtarget;
2662
2663         if (progress_func) {
2664                 int msg;
2665                 if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE)
2666                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END;
2667                 else
2668                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END;
2669                 progress_func(msg, &ctx.progress);
2670         }
2671         do_extract_warnings(&ctx);
2672         ret = 0;
2673 out_free_realtarget:
2674         FREE(ctx.realtarget);
2675 out_destroy_stream_list:
2676         if (!(ctx.extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2677                 destroy_stream_list(&ctx.stream_list);
2678 out_finish_or_abort_extract:
2679         if (ret) {
2680                 if (ctx.ops->abort_extract)
2681                         ctx.ops->abort_extract(&ctx);
2682         } else {
2683                 if (ctx.ops->finish_extract)
2684                         ret = ctx.ops->finish_extract(&ctx);
2685         }
2686 out_destroy_dentry_list:
2687         destroy_dentry_list(&dentry_list);
2688         return ret;
2689 }
2690
2691 static int
2692 mkdir_if_needed(const tchar *target)
2693 {
2694         struct stat stbuf;
2695         if (tstat(target, &stbuf)) {
2696                 if (errno == ENOENT) {
2697                         if (tmkdir(target, 0755)) {
2698                                 ERROR_WITH_ERRNO("Failed to create directory "
2699                                                  "\"%"TS"\"", target);
2700                                 return WIMLIB_ERR_MKDIR;
2701                         }
2702                 } else {
2703                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
2704                         return WIMLIB_ERR_STAT;
2705                 }
2706         } else if (!S_ISDIR(stbuf.st_mode)) {
2707                 ERROR("\"%"TS"\" is not a directory", target);
2708                 return WIMLIB_ERR_NOTDIR;
2709         }
2710         return 0;
2711 }
2712
2713 /* Make sure the extraction flags make sense, and update them if needed.  */
2714 static int
2715 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
2716 {
2717         int extract_flags = *extract_flags_p;
2718
2719         /* Check for invalid flag combinations  */
2720         if ((extract_flags &
2721              (WIMLIB_EXTRACT_FLAG_SYMLINK |
2722               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
2723                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
2724                 return WIMLIB_ERR_INVALID_PARAM;
2725
2726         if ((extract_flags &
2727              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2728               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2729                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
2730                 return WIMLIB_ERR_INVALID_PARAM;
2731
2732         if ((extract_flags &
2733              (WIMLIB_EXTRACT_FLAG_RPFIX |
2734               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
2735                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
2736                 return WIMLIB_ERR_INVALID_PARAM;
2737
2738         if ((extract_flags &
2739              (WIMLIB_EXTRACT_FLAG_RESUME |
2740               WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME)
2741                 return WIMLIB_ERR_INVALID_PARAM;
2742
2743 #ifndef WITH_NTFS_3G
2744         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2745                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
2746                       "        it cannot apply a WIM image directly to a NTFS volume.");
2747                 return WIMLIB_ERR_UNSUPPORTED;
2748         }
2749 #endif
2750
2751 #ifndef __WIN32__
2752         if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
2753                 ERROR("WIMBoot extraction is only supported on Windows!");
2754                 return WIMLIB_ERR_UNSUPPORTED;
2755         }
2756 #endif
2757
2758         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
2759                               WIMLIB_EXTRACT_FLAG_NORPFIX |
2760                               WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
2761                                         WIMLIB_EXTRACT_FLAG_IMAGEMODE)
2762         {
2763                 /* For full-image extraction, do reparse point fixups by default
2764                  * if the WIM header says they are enabled.  */
2765                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
2766                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
2767         }
2768
2769         /* TODO: Since UNIX data entries are stored in the file resources, in a
2770          * completely sequential extraction they may come up before the
2771          * corresponding file or symbolic link data.  This needs to be handled
2772          * better.  */
2773         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2774                               WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2775                                     == WIMLIB_EXTRACT_FLAG_UNIX_DATA)
2776         {
2777                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2778                         WARNING("Setting UNIX file/owner group may "
2779                                 "be impossible on some\n"
2780                                 "          symbolic links "
2781                                 "when applying from a pipe.");
2782                 } else {
2783                         extract_flags |= WIMLIB_EXTRACT_FLAG_FILE_ORDER;
2784                         WARNING("Disabling sequential extraction for "
2785                                 "UNIX data mode");
2786                 }
2787         }
2788
2789         *extract_flags_p = extract_flags;
2790         return 0;
2791 }
2792
2793 static u32
2794 get_wildcard_flags(int extract_flags)
2795 {
2796         u32 wildcard_flags = 0;
2797
2798         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
2799                 wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
2800         else
2801                 wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
2802
2803         if (default_ignore_case)
2804                 wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
2805
2806         return wildcard_flags;
2807 }
2808
2809 struct append_dentry_ctx {
2810         struct wim_dentry **dentries;
2811         size_t num_dentries;
2812         size_t num_alloc_dentries;
2813 };
2814
2815 static int
2816 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
2817 {
2818         struct append_dentry_ctx *ctx = _ctx;
2819
2820         if (ctx->num_dentries == ctx->num_alloc_dentries) {
2821                 struct wim_dentry **new_dentries;
2822                 size_t new_length;
2823
2824                 new_length = max(ctx->num_alloc_dentries + 8,
2825                                  ctx->num_alloc_dentries * 3 / 2);
2826                 new_dentries = REALLOC(ctx->dentries,
2827                                        new_length * sizeof(ctx->dentries[0]));
2828                 if (new_dentries == NULL)
2829                         return WIMLIB_ERR_NOMEM;
2830                 ctx->dentries = new_dentries;
2831                 ctx->num_alloc_dentries = new_length;
2832         }
2833         ctx->dentries[ctx->num_dentries++] = dentry;
2834         return 0;
2835 }
2836
2837 static int
2838 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
2839                         const tchar * const *paths, size_t num_paths,
2840                         int extract_flags, wimlib_progress_func_t progress_func)
2841 {
2842         int ret;
2843         struct wim_dentry **trees;
2844         size_t num_trees;
2845
2846         if (wim == NULL || target == NULL || target[0] == T('\0') ||
2847             (num_paths != 0 && paths == NULL))
2848                 return WIMLIB_ERR_INVALID_PARAM;
2849
2850         ret = check_extract_flags(wim, &extract_flags);
2851         if (ret)
2852                 return ret;
2853
2854         ret = select_wim_image(wim, image);
2855         if (ret)
2856                 return ret;
2857
2858         ret = wim_checksum_unhashed_streams(wim);
2859         if (ret)
2860                 return ret;
2861
2862         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
2863                               WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
2864             (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
2865         {
2866                 ret = mkdir_if_needed(target);
2867                 if (ret)
2868                         return ret;
2869         }
2870
2871         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
2872
2873                 struct append_dentry_ctx append_dentry_ctx = {
2874                         .dentries = NULL,
2875                         .num_dentries = 0,
2876                         .num_alloc_dentries = 0,
2877                 };
2878
2879                 u32 wildcard_flags = get_wildcard_flags(extract_flags);
2880
2881                 for (size_t i = 0; i < num_paths; i++) {
2882                         tchar *path = canonicalize_wim_path(paths[i]);
2883                         if (path == NULL) {
2884                                 ret = WIMLIB_ERR_NOMEM;
2885                                 trees = append_dentry_ctx.dentries;
2886                                 goto out_free_trees;
2887                         }
2888                         ret = expand_wildcard(wim, path,
2889                                               append_dentry_cb,
2890                                               &append_dentry_ctx,
2891                                               wildcard_flags);
2892                         FREE(path);
2893                         if (ret) {
2894                                 trees = append_dentry_ctx.dentries;
2895                                 goto out_free_trees;
2896                         }
2897                 }
2898                 trees = append_dentry_ctx.dentries;
2899                 num_trees = append_dentry_ctx.num_dentries;
2900         } else {
2901                 trees = MALLOC(num_paths * sizeof(trees[0]));
2902                 if (trees == NULL)
2903                         return WIMLIB_ERR_NOMEM;
2904
2905                 for (size_t i = 0; i < num_paths; i++) {
2906
2907                         tchar *path = canonicalize_wim_path(paths[i]);
2908                         if (path == NULL) {
2909                                 ret = WIMLIB_ERR_NOMEM;
2910                                 goto out_free_trees;
2911                         }
2912
2913                         trees[i] = get_dentry(wim, path,
2914                                               WIMLIB_CASE_PLATFORM_DEFAULT);
2915                         FREE(path);
2916                         if (trees[i] == NULL) {
2917                                   ERROR("Path \"%"TS"\" does not exist "
2918                                         "in WIM image %d",
2919                                         paths[i], wim->current_image);
2920                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
2921                                   goto out_free_trees;
2922                         }
2923                 }
2924                 num_trees = num_paths;
2925         }
2926
2927         if (num_trees == 0) {
2928                 ret = 0;
2929                 goto out_free_trees;
2930         }
2931
2932         ret = extract_trees(wim, trees, num_trees,
2933                             target, extract_flags, progress_func);
2934 out_free_trees:
2935         FREE(trees);
2936         return ret;
2937 }
2938
2939 static int
2940 extract_single_image(WIMStruct *wim, int image,
2941                      const tchar *target, int extract_flags,
2942                      wimlib_progress_func_t progress_func)
2943 {
2944         const tchar *path = WIMLIB_WIM_ROOT_PATH;
2945         extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
2946         return do_wimlib_extract_paths(wim, image, target, &path, 1,
2947                                        extract_flags, progress_func);
2948 }
2949
2950 static const tchar * const filename_forbidden_chars =
2951 T(
2952 #ifdef __WIN32__
2953 "<>:\"/\\|?*"
2954 #else
2955 "/"
2956 #endif
2957 );
2958
2959 /* This function checks if it is okay to use a WIM image's name as a directory
2960  * name.  */
2961 static bool
2962 image_name_ok_as_dir(const tchar *image_name)
2963 {
2964         return image_name && *image_name &&
2965                 !tstrpbrk(image_name, filename_forbidden_chars) &&
2966                 tstrcmp(image_name, T(".")) &&
2967                 tstrcmp(image_name, T(".."));
2968 }
2969
2970 /* Extracts all images from the WIM to the directory @target, with the images
2971  * placed in subdirectories named by their image names. */
2972 static int
2973 extract_all_images(WIMStruct *wim,
2974                    const tchar *target,
2975                    int extract_flags,
2976                    wimlib_progress_func_t progress_func)
2977 {
2978         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
2979         size_t output_path_len = tstrlen(target);
2980         tchar buf[output_path_len + 1 + image_name_max_len + 1];
2981         int ret;
2982         int image;
2983         const tchar *image_name;
2984
2985         extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
2986
2987         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2988                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
2989                 return WIMLIB_ERR_INVALID_PARAM;
2990         }
2991
2992         ret = mkdir_if_needed(target);
2993         if (ret)
2994                 return ret;
2995         tmemcpy(buf, target, output_path_len);
2996         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
2997         for (image = 1; image <= wim->hdr.image_count; image++) {
2998                 image_name = wimlib_get_image_name(wim, image);
2999                 if (image_name_ok_as_dir(image_name)) {
3000                         tstrcpy(buf + output_path_len + 1, image_name);
3001                 } else {
3002                         /* Image name is empty or contains forbidden characters.
3003                          * Use image number instead. */
3004                         tsprintf(buf + output_path_len + 1, T("%d"), image);
3005                 }
3006                 ret = extract_single_image(wim, image, buf, extract_flags,
3007                                            progress_func);
3008                 if (ret)
3009                         return ret;
3010         }
3011         return 0;
3012 }
3013
3014 static void
3015 clear_lte_extracted_file(WIMStruct *wim, int extract_flags)
3016 {
3017         if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
3018                                       WIMLIB_EXTRACT_FLAG_HARDLINK)))
3019                 for_lookup_table_entry(wim->lookup_table,
3020                                        lte_free_extracted_file, NULL);
3021 }
3022
3023 static int
3024 do_wimlib_extract_image(WIMStruct *wim,
3025                         int image,
3026                         const tchar *target,
3027                         int extract_flags,
3028                         wimlib_progress_func_t progress_func)
3029 {
3030         int ret;
3031
3032         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
3033                              WIMLIB_EXTRACT_FLAG_TO_STDOUT |
3034                              WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
3035                 return WIMLIB_ERR_INVALID_PARAM;
3036
3037         if (image == WIMLIB_ALL_IMAGES)
3038                 ret = extract_all_images(wim, target, extract_flags,
3039                                          progress_func);
3040         else
3041                 ret = extract_single_image(wim, image, target, extract_flags,
3042                                            progress_func);
3043
3044         clear_lte_extracted_file(wim, extract_flags);
3045         return ret;
3046 }
3047
3048
3049 /****************************************************************************
3050  *                          Extraction API                                  *
3051  ****************************************************************************/
3052
3053 WIMLIBAPI int
3054 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
3055                      const tchar * const *paths, size_t num_paths,
3056                      int extract_flags, wimlib_progress_func_t progress_func)
3057 {
3058         int ret;
3059
3060         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
3061                 return WIMLIB_ERR_INVALID_PARAM;
3062
3063         ret = do_wimlib_extract_paths(wim, image, target, paths, num_paths,
3064                                       extract_flags, progress_func);
3065         clear_lte_extracted_file(wim, extract_flags);
3066         return ret;
3067 }
3068
3069 WIMLIBAPI int
3070 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
3071                         const tchar *path_list_file, int extract_flags,
3072                         wimlib_progress_func_t progress_func)
3073 {
3074         int ret;
3075         tchar **paths;
3076         size_t num_paths;
3077         void *mem;
3078
3079         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
3080         if (ret) {
3081                 ERROR("Failed to read path list file \"%"TS"\"",
3082                       path_list_file);
3083                 return ret;
3084         }
3085
3086         ret = wimlib_extract_paths(wim, image, target,
3087                                    (const tchar * const *)paths, num_paths,
3088                                    extract_flags, progress_func);
3089         FREE(paths);
3090         FREE(mem);
3091         return ret;
3092 }
3093
3094 WIMLIBAPI int
3095 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
3096                                const tchar *target, int extract_flags,
3097                                wimlib_progress_func_t progress_func)
3098 {
3099         int ret;
3100         WIMStruct *pwm;
3101         struct filedes *in_fd;
3102         int image;
3103         unsigned i;
3104
3105         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
3106                 return WIMLIB_ERR_INVALID_PARAM;
3107
3108         if (extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)
3109                 return WIMLIB_ERR_INVALID_PARAM;
3110
3111         /* Read the WIM header from the pipe and get a WIMStruct to represent
3112          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
3113          * wimlib_open_wim(), getting a WIMStruct in this way will result in
3114          * an empty lookup table, no XML data read, and no filename set.  */
3115         ret = open_wim_as_WIMStruct(&pipe_fd,
3116                                     WIMLIB_OPEN_FLAG_FROM_PIPE,
3117                                     &pwm, progress_func);
3118         if (ret)
3119                 return ret;
3120
3121         /* Sanity check to make sure this is a pipable WIM.  */
3122         if (pwm->hdr.magic != PWM_MAGIC) {
3123                 ERROR("The WIM being read from file descriptor %d "
3124                       "is not pipable!", pipe_fd);
3125                 ret = WIMLIB_ERR_NOT_PIPABLE;
3126                 goto out_wimlib_free;
3127         }
3128
3129         /* Sanity check to make sure the first part of a pipable split WIM is
3130          * sent over the pipe first.  */
3131         if (pwm->hdr.part_number != 1) {
3132                 ERROR("The first part of the split WIM must be "
3133                       "sent over the pipe first.");
3134                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3135                 goto out_wimlib_free;
3136         }
3137
3138         in_fd = &pwm->in_fd;
3139         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
3140
3141         /* As mentioned, the WIMStruct we created from the pipe does not have
3142          * XML data yet.  Fix this by reading the extra copy of the XML data
3143          * that directly follows the header in pipable WIMs.  (Note: see
3144          * write_pipable_wim() for more details about the format of pipable
3145          * WIMs.)  */
3146         {
3147                 struct wim_lookup_table_entry xml_lte;
3148                 struct wim_resource_spec xml_rspec;
3149                 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
3150                 if (ret)
3151                         goto out_wimlib_free;
3152
3153                 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
3154                 {
3155                         ERROR("Expected XML data, but found non-metadata "
3156                               "stream.");
3157                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3158                         goto out_wimlib_free;
3159                 }
3160
3161                 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
3162
3163                 ret = read_wim_xml_data(pwm);
3164                 if (ret)
3165                         goto out_wimlib_free;
3166
3167                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
3168                         ERROR("Image count in XML data is not the same as in WIM header.");
3169                         ret = WIMLIB_ERR_IMAGE_COUNT;
3170                         goto out_wimlib_free;
3171                 }
3172         }
3173
3174         /* Get image index (this may use the XML data that was just read to
3175          * resolve an image name).  */
3176         if (image_num_or_name) {
3177                 image = wimlib_resolve_image(pwm, image_num_or_name);
3178                 if (image == WIMLIB_NO_IMAGE) {
3179                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
3180                               image_num_or_name);
3181                         ret = WIMLIB_ERR_INVALID_IMAGE;
3182                         goto out_wimlib_free;
3183                 } else if (image == WIMLIB_ALL_IMAGES) {
3184                         ERROR("Applying all images from a pipe is not supported!");
3185                         ret = WIMLIB_ERR_INVALID_IMAGE;
3186                         goto out_wimlib_free;
3187                 }
3188         } else {
3189                 if (pwm->hdr.image_count != 1) {
3190                         ERROR("No image was specified, but the pipable WIM "
3191                               "did not contain exactly 1 image");
3192                         ret = WIMLIB_ERR_INVALID_IMAGE;
3193                         goto out_wimlib_free;
3194                 }
3195                 image = 1;
3196         }
3197
3198         /* Load the needed metadata resource.  */
3199         for (i = 1; i <= pwm->hdr.image_count; i++) {
3200                 struct wim_lookup_table_entry *metadata_lte;
3201                 struct wim_image_metadata *imd;
3202                 struct wim_resource_spec *metadata_rspec;
3203
3204                 metadata_lte = new_lookup_table_entry();
3205                 if (metadata_lte == NULL) {
3206                         ret = WIMLIB_ERR_NOMEM;
3207                         goto out_wimlib_free;
3208                 }
3209                 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
3210                 if (metadata_rspec == NULL) {
3211                         ret = WIMLIB_ERR_NOMEM;
3212                         free_lookup_table_entry(metadata_lte);
3213                         goto out_wimlib_free;
3214                 }
3215
3216                 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
3217                 imd = pwm->image_metadata[i - 1];
3218                 imd->metadata_lte = metadata_lte;
3219                 if (ret) {
3220                         FREE(metadata_rspec);
3221                         goto out_wimlib_free;
3222                 }
3223
3224                 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
3225                         ERROR("Expected metadata resource, but found "
3226                               "non-metadata stream.");
3227                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3228                         goto out_wimlib_free;
3229                 }
3230
3231                 if (i == image) {
3232                         /* Metadata resource is for the image being extracted.
3233                          * Parse it and save the metadata in memory.  */
3234                         ret = read_metadata_resource(pwm, imd);
3235                         if (ret)
3236                                 goto out_wimlib_free;
3237                         imd->modified = 1;
3238                 } else {
3239                         /* Metadata resource is not for the image being
3240                          * extracted.  Skip over it.  */
3241                         ret = skip_wim_stream(metadata_lte);
3242                         if (ret)
3243                                 goto out_wimlib_free;
3244                 }
3245         }
3246         /* Extract the image.  */
3247         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
3248         ret = do_wimlib_extract_image(pwm, image, target,
3249                                       extract_flags, progress_func);
3250         /* Clean up and return.  */
3251 out_wimlib_free:
3252         wimlib_free(pwm);
3253         return ret;
3254 }
3255
3256 WIMLIBAPI int
3257 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
3258                      int extract_flags, wimlib_progress_func_t progress_func)
3259 {
3260         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
3261                 return WIMLIB_ERR_INVALID_PARAM;
3262         return do_wimlib_extract_image(wim, image, target, extract_flags,
3263                                        progress_func);
3264 }