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 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_files(), and wimlib_extract_image_from_pipe().  Internally,
  30  * all end up calling extract_tree() zero or more times to extract a tree of
  31  * files from the currently selected WIM image to the specified target directory
  32  * or NTFS volume.
  33  *
  34  * Although wimlib supports multiple extraction modes/backends (NTFS-3g, UNIX,
  35  * Win32), this file does not itself have code to extract files or directories
  36  * to any specific target; instead, it handles generic functionality and relies
  37  * on lower-level callback functions declared in `struct apply_operations' to do
  38  * the actual extraction.
  39  */
  40
  41 #ifdef HAVE_CONFIG_H
  42 #  include "config.h"
  43 #endif
  44
  45 #include "wimlib/apply.h"
  46 #include "wimlib/dentry.h"
  47 #include "wimlib/encoding.h"
  48 #include "wimlib/endianness.h"
  49 #include "wimlib/error.h"
  50 #include "wimlib/lookup_table.h"
  51 #include "wimlib/metadata.h"
  52 #include "wimlib/pathlist.h"
  53 #include "wimlib/paths.h"
  54 #include "wimlib/reparse.h"
  55 #include "wimlib/resource.h"
  56 #include "wimlib/security.h"
  57 #ifdef __WIN32__
  58 #  include "wimlib/win32.h" /* for realpath() equivalent */
  59 #endif
  60 #include "wimlib/xml.h"
  61 #include "wimlib/wildcard.h"
  62 #include "wimlib/wim.h"
  63
  64 #include <errno.h>
  65 #include <fcntl.h>
  66 #include <stdlib.h>
  67 #include <sys/stat.h>
  68 #include <unistd.h>
  69
  70 #define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
  71 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x40000000
  72 #define WIMLIB_EXTRACT_FLAG_FILEMODE    0x20000000
  73 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x10000000
  74
  75 /* Keep in sync with wimlib.h  */
  76 #define WIMLIB_EXTRACT_MASK_PUBLIC                              \
  77         (WIMLIB_EXTRACT_FLAG_NTFS                       |       \
  78          WIMLIB_EXTRACT_FLAG_HARDLINK                   |       \
  79          WIMLIB_EXTRACT_FLAG_SYMLINK                    |       \
  80          WIMLIB_EXTRACT_FLAG_VERBOSE                    |       \
  81          WIMLIB_EXTRACT_FLAG_SEQUENTIAL                 |       \
  82          WIMLIB_EXTRACT_FLAG_UNIX_DATA                  |       \
  83          WIMLIB_EXTRACT_FLAG_NO_ACLS                    |       \
  84          WIMLIB_EXTRACT_FLAG_STRICT_ACLS                |       \
  85          WIMLIB_EXTRACT_FLAG_RPFIX                      |       \
  86          WIMLIB_EXTRACT_FLAG_NORPFIX                    |       \
  87          WIMLIB_EXTRACT_FLAG_TO_STDOUT                  |       \
  88          WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES  |       \
  89          WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS         |       \
  90          WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS          |       \
  91          WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES         |       \
  92          WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS            |       \
  93          WIMLIB_EXTRACT_FLAG_RESUME                     |       \
  94          WIMLIB_EXTRACT_FLAG_FILE_ORDER                 |       \
  95          WIMLIB_EXTRACT_FLAG_GLOB_PATHS                 |       \
  96          WIMLIB_EXTRACT_FLAG_STRICT_GLOB                |       \
  97          WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES              |       \
  98          WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)
  99
 100 /* Given a WIM dentry in the tree to be extracted, resolve all streams in the
 101  * corresponding inode and set 'out_refcnt' in each to 0.  */
 102 static int
 103 dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_ctx)
 104 {
 105         struct apply_ctx *ctx = _ctx;
 106         struct wim_inode *inode = dentry->d_inode;
 107         struct wim_lookup_table_entry *lte;
 108         int ret;
 109         bool force = false;
 110
 111         if (dentry->extraction_skipped)
 112                 return 0;
 113
 114         /* Special case:  when extracting from a pipe, the WIM lookup table is
 115          * initially empty, so "resolving" an inode's streams is initially not
 116          * possible.  However, we still need to keep track of which streams,
 117          * identified by SHA1 message digests, need to be extracted, so we
 118          * "resolve" the inode's streams anyway by allocating new entries.  */
 119         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
 120                 force = true;
 121         ret = inode_resolve_streams(inode, ctx->wim->lookup_table, force);
 122         if (ret)
 123                 return ret;
 124         for (unsigned i = 0; i <= inode->i_num_ads; i++) {
 125                 lte = inode_stream_lte_resolved(inode, i);
 126                 if (lte)
 127                         lte->out_refcnt = 0;
 128         }
 129         return 0;
 130 }
 131
 132 static inline bool
 133 is_linked_extraction(const struct apply_ctx *ctx)
 134 {
 135         return 0 != (ctx->extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
 136                                            WIMLIB_EXTRACT_FLAG_SYMLINK));
 137 }
 138
 139 static inline bool
 140 can_extract_named_data_streams(const struct apply_ctx *ctx)
 141 {
 142         return ctx->supported_features.named_data_streams &&
 143                 !is_linked_extraction(ctx);
 144 }
 145
 146 static int
 147 ref_stream_to_extract(struct wim_lookup_table_entry *lte,
 148                       struct wim_dentry *dentry, struct apply_ctx *ctx)
 149 {
 150         if (!lte)
 151                 return 0;
 152
 153         /* Tally the size only for each extraction of the stream (not hard
 154          * links).  */
 155         if (!(dentry->d_inode->i_visited &&
 156               ctx->supported_features.hard_links) &&
 157             (!is_linked_extraction(ctx) || (lte->out_refcnt == 0 &&
 158                                             lte->extracted_file == NULL)))
 159         {
 160                 ctx->progress.extract.total_bytes += lte->size;
 161                 ctx->progress.extract.num_streams++;
 162         }
 163
 164         /* Add stream to the extraction_list only one time, even if it's going
 165          * to be extracted to multiple locations.  */
 166         if (lte->out_refcnt == 0) {
 167                 list_add_tail(&lte->extraction_list, &ctx->stream_list);
 168                 ctx->num_streams_remaining++;
 169         }
 170
 171         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
 172                 struct wim_dentry **lte_dentries;
 173
 174                 /* Append dentry to this stream's array of dentries referencing
 175                  * it.  Use inline array to avoid memory allocation until the
 176                  * number of dentries becomes too large.  */
 177                 if (lte->out_refcnt < ARRAY_LEN(lte->inline_lte_dentries)) {
 178                         lte_dentries = lte->inline_lte_dentries;
 179                 } else {
 180                         struct wim_dentry **prev_lte_dentries;
 181                         size_t alloc_lte_dentries;
 182
 183                         if (lte->out_refcnt == ARRAY_LEN(lte->inline_lte_dentries)) {
 184                                 prev_lte_dentries = NULL;
 185                                 alloc_lte_dentries = ARRAY_LEN(lte->inline_lte_dentries);
 186                         } else {
 187                                 prev_lte_dentries = lte->lte_dentries;
 188                                 alloc_lte_dentries = lte->alloc_lte_dentries;
 189                         }
 190
 191                         if (lte->out_refcnt == alloc_lte_dentries) {
 192                                 alloc_lte_dentries *= 2;
 193                                 lte_dentries = REALLOC(prev_lte_dentries,
 194                                                        alloc_lte_dentries *
 195                                                         sizeof(lte_dentries[0]));
 196                                 if (lte_dentries == NULL)
 197                                         return WIMLIB_ERR_NOMEM;
 198                                 if (prev_lte_dentries == NULL) {
 199                                         memcpy(lte_dentries,
 200                                                lte->inline_lte_dentries,
 201                                                sizeof(lte->inline_lte_dentries));
 202                                 }
 203                                 lte->lte_dentries = lte_dentries;
 204                                 lte->alloc_lte_dentries = alloc_lte_dentries;
 205                         }
 206                         lte_dentries = lte->lte_dentries;
 207                 }
 208                 lte_dentries[lte->out_refcnt] = dentry;
 209         }
 210         lte->out_refcnt++;
 211         return 0;
 212 }
 213
 214 /* Given a WIM dentry in the tree to be extracted, iterate through streams that
 215  * need to be extracted.  For each one, add it to the list of streams to be
 216  * extracted (ctx->stream_list) if not already done so, and also update the
 217  * progress information (ctx->progress) with the stream.  Furthermore, if doing
 218  * a sequential extraction, build a mapping from each the stream to the dentries
 219  * referencing it.
 220  *
 221  * This uses the i_visited member of the inodes (assumed to be 0 initially).  */
 222 static int
 223 dentry_add_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
 224 {
 225         struct apply_ctx *ctx = _ctx;
 226         struct wim_inode *inode = dentry->d_inode;
 227         int ret;
 228
 229         /* Don't process dentries marked as skipped.  */
 230         if (dentry->extraction_skipped)
 231                 return 0;
 232
 233         /* The unnamed data stream will always be extracted, except in an
 234          * unlikely case.  */
 235         if (!inode_is_encrypted_directory(inode)) {
 236                 ret = ref_stream_to_extract(inode_unnamed_lte_resolved(inode),
 237                                             dentry, ctx);
 238                 if (ret)
 239                         return ret;
 240         }
 241
 242         /* Named data streams will be extracted only if supported in the current
 243          * extraction mode and volume, and to avoid complications, if not doing
 244          * a linked extraction.  */
 245         if (can_extract_named_data_streams(ctx)) {
 246                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 247                         if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
 248                                 continue;
 249                         ret = ref_stream_to_extract(inode->i_ads_entries[i].lte,
 250                                                     dentry, ctx);
 251                         if (ret)
 252                                 return ret;
 253                 }
 254         }
 255         inode->i_visited = 1;
 256         return 0;
 257 }
 258
 259 /* Inform library user of progress of stream extraction following the successful
 260  * extraction of a copy of the stream specified by @lte.  */
 261 static void
 262 update_extract_progress(struct apply_ctx *ctx,
 263                         const struct wim_lookup_table_entry *lte)
 264 {
 265         wimlib_progress_func_t progress_func = ctx->progress_func;
 266         union wimlib_progress_info *progress = &ctx->progress;
 267
 268         progress->extract.completed_bytes += lte->size;
 269         if (progress_func &&
 270             progress->extract.completed_bytes >= ctx->next_progress)
 271         {
 272                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, progress);
 273                 if (progress->extract.completed_bytes >=
 274                     progress->extract.total_bytes)
 275                 {
 276                         ctx->next_progress = ~0ULL;
 277                 } else {
 278                         ctx->next_progress += progress->extract.total_bytes / 128;
 279                         if (ctx->next_progress > progress->extract.total_bytes)
 280                                 ctx->next_progress = progress->extract.total_bytes;
 281                 }
 282         }
 283 }
 284
 285 #ifndef __WIN32__
 286 /* Extract a symbolic link (not directly as reparse data), handling fixing up
 287  * the target of absolute symbolic links and updating the extract progress.
 288  *
 289  * @inode must specify the WIM inode for a symbolic link or junction reparse
 290  * point.
 291  *
 292  * @lte_override overrides the resource used as the reparse data for the
 293  * symbolic link.  */
 294 static int
 295 extract_symlink(const tchar *path, struct apply_ctx *ctx,
 296                 struct wim_inode *inode,
 297                 struct wim_lookup_table_entry *lte_override)
 298 {
 299         ssize_t bufsize = ctx->ops->path_max;
 300         tchar target[bufsize];
 301         tchar *buf = target;
 302         tchar *fixed_target;
 303         ssize_t sret;
 304         int ret;
 305
 306         /* If absolute symbolic link fixups requested, reserve space in the link
 307          * target buffer for the absolute path of the target directory.  */
 308         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)
 309         {
 310                 buf += ctx->realtarget_nchars;
 311                 bufsize -= ctx->realtarget_nchars;
 312         }
 313
 314         /* Translate the WIM inode's reparse data into the link target.  */
 315         sret = wim_inode_readlink(inode, buf, bufsize - 1, lte_override);
 316         if (sret < 0) {
 317                 errno = -sret;
 318                 return WIMLIB_ERR_READLINK;
 319         }
 320         buf[sret] = '\0';
 321
 322         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
 323             buf[0] == '/')
 324         {
 325                 /* Fix absolute symbolic link target to point into the
 326                  * actual extraction destination.  */
 327                 tmemcpy(target, ctx->realtarget, ctx->realtarget_nchars);
 328                 fixed_target = target;
 329         } else {
 330                 /* Keep same link target.  */
 331                 fixed_target = buf;
 332         }
 333
 334         /* Call into the apply_operations to create the symbolic link.  */
 335         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
 336               path, fixed_target);
 337         ret = ctx->ops->create_symlink(fixed_target, path, ctx);
 338         if (ret) {
 339                 ERROR_WITH_ERRNO("Failed to create symlink "
 340                                  "\"%"TS"\" => \"%"TS"\"", path, fixed_target);
 341                 return ret;
 342         }
 343
 344         /* Account for reparse data consumed.  */
 345         update_extract_progress(ctx,
 346                                 (lte_override ? lte_override :
 347                                       inode_unnamed_lte_resolved(inode)));
 348         return 0;
 349 }
 350 #endif /* !__WIN32__ */
 351
 352 /* Create a file, directory, or symbolic link.  */
 353 static int
 354 extract_inode(const tchar *path, struct apply_ctx *ctx, struct wim_inode *inode)
 355 {
 356         int ret;
 357
 358 #ifndef __WIN32__
 359         if (ctx->supported_features.symlink_reparse_points &&
 360             !ctx->supported_features.reparse_points &&
 361             inode_is_symlink(inode))
 362         {
 363                 ret = extract_symlink(path, ctx, inode, NULL);
 364         } else
 365 #endif /* !__WIN32__ */
 366         if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
 367                 ret = ctx->ops->create_directory(path, ctx, &inode->extract_cookie);
 368                 if (ret) {
 369                         ERROR_WITH_ERRNO("Failed to create the directory "
 370                                          "\"%"TS"\"", path);
 371                 }
 372         } else if ((inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) &&
 373                     ctx->ops->extract_encrypted_stream_creates_file &&
 374                     ctx->supported_features.encrypted_files) {
 375                 ret = ctx->ops->extract_encrypted_stream(
 376                                 path, inode_unnamed_lte_resolved(inode), ctx);
 377                 if (ret) {
 378                         ERROR_WITH_ERRNO("Failed to create and extract "
 379                                          "encrypted file \"%"TS"\"", path);
 380                 }
 381         } else {
 382                 ret = ctx->ops->create_file(path, ctx, &inode->extract_cookie);
 383                 if (ret) {
 384                         ERROR_WITH_ERRNO("Failed to create the file "
 385                                          "\"%"TS"\"", path);
 386                 }
 387         }
 388         return ret;
 389 }
 390
 391 static int
 392 extract_hardlink(const tchar *oldpath, const tchar *newpath,
 393                  struct apply_ctx *ctx)
 394 {
 395         int ret;
 396
 397         DEBUG("Creating hardlink \"%"TS"\" => \"%"TS"\"", newpath, oldpath);
 398         ret = ctx->ops->create_hardlink(oldpath, newpath, ctx);
 399         if (ret) {
 400                 ERROR_WITH_ERRNO("Failed to create hardlink "
 401                                  "\"%"TS"\" => \"%"TS"\"",
 402                                  newpath, oldpath);
 403         }
 404         return ret;
 405 }
 406
 407 #ifdef __WIN32__
 408 static int
 409 try_extract_rpfix(u8 *rpbuf,
 410                   u16 *rpbuflen_p,
 411                   const wchar_t *extract_root_realpath,
 412                   unsigned extract_root_realpath_nchars)
 413 {
 414         struct reparse_data rpdata;
 415         wchar_t *target;
 416         size_t target_nchars;
 417         size_t stripped_nchars;
 418         wchar_t *stripped_target;
 419         wchar_t stripped_target_nchars;
 420         int ret;
 421
 422         utf16lechar *new_target;
 423         utf16lechar *new_print_name;
 424         size_t new_target_nchars;
 425         size_t new_print_name_nchars;
 426         utf16lechar *p;
 427
 428         ret = parse_reparse_data(rpbuf, *rpbuflen_p, &rpdata);
 429         if (ret)
 430                 return ret;
 431
 432         if (extract_root_realpath[0] == L'\0' ||
 433             extract_root_realpath[1] != L':' ||
 434             extract_root_realpath[2] != L'\\')
 435                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
 436
 437         ret = parse_substitute_name(rpdata.substitute_name,
 438                                     rpdata.substitute_name_nbytes,
 439                                     rpdata.rptag);
 440         if (ret < 0)
 441                 return 0;
 442         stripped_nchars = ret;
 443         target = rpdata.substitute_name;
 444         target_nchars = rpdata.substitute_name_nbytes / sizeof(utf16lechar);
 445         stripped_target = target + stripped_nchars;
 446         stripped_target_nchars = target_nchars - stripped_nchars;
 447
 448         new_target = alloca((6 + extract_root_realpath_nchars +
 449                              stripped_target_nchars) * sizeof(utf16lechar));
 450
 451         p = new_target;
 452         if (stripped_nchars == 6) {
 453                 /* Include \??\ prefix if it was present before */
 454                 p = wmempcpy(p, L"\\??\\", 4);
 455         }
 456
 457         /* Print name excludes the \??\ if present. */
 458         new_print_name = p;
 459         if (stripped_nchars != 0) {
 460                 /* Get drive letter from real path to extract root, if a drive
 461                  * letter was present before. */
 462                 *p++ = extract_root_realpath[0];
 463                 *p++ = extract_root_realpath[1];
 464         }
 465         /* Copy the rest of the extract root */
 466         p = wmempcpy(p, extract_root_realpath + 2, extract_root_realpath_nchars - 2);
 467
 468         /* Append the stripped target */
 469         p = wmempcpy(p, stripped_target, stripped_target_nchars);
 470         new_target_nchars = p - new_target;
 471         new_print_name_nchars = p - new_print_name;
 472
 473         if (new_target_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE ||
 474             new_print_name_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE)
 475                 return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED;
 476
 477         rpdata.substitute_name = new_target;
 478         rpdata.substitute_name_nbytes = new_target_nchars * sizeof(utf16lechar);
 479         rpdata.print_name = new_print_name;
 480         rpdata.print_name_nbytes = new_print_name_nchars * sizeof(utf16lechar);
 481         return make_reparse_buffer(&rpdata, rpbuf, rpbuflen_p);
 482 }
 483 #endif /* __WIN32__ */
 484
 485 /* Set reparse data on extracted file or directory that has
 486  * FILE_ATTRIBUTE_REPARSE_POINT set.  */
 487 static int
 488 extract_reparse_data(const tchar *path, struct apply_ctx *ctx,
 489                      struct wim_inode *inode,
 490                      struct wim_lookup_table_entry *lte_override)
 491 {
 492         int ret;
 493         u8 rpbuf[REPARSE_POINT_MAX_SIZE];
 494         u16 rpbuflen;
 495
 496         ret = wim_inode_get_reparse_data(inode, rpbuf, &rpbuflen, lte_override);
 497         if (ret)
 498                 goto error;
 499
 500 #ifdef __WIN32__
 501         /* Fix up target of absolute symbolic link or junction points so
 502          * that they point into the actual extraction target.  */
 503         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
 504             (inode->i_reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK ||
 505              inode->i_reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT) &&
 506             !inode->i_not_rpfixed)
 507         {
 508                 ret = try_extract_rpfix(rpbuf, &rpbuflen, ctx->realtarget,
 509                                         ctx->realtarget_nchars);
 510                 if (ret && !(ctx->extract_flags &
 511                              WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
 512                 {
 513                         WARNING("Reparse point fixup of \"%"TS"\" "
 514                                 "failed", path);
 515                         ret = 0;
 516                 }
 517                 if (ret)
 518                         goto error;
 519         }
 520 #endif
 521
 522         ret = ctx->ops->set_reparse_data(path, rpbuf, rpbuflen, ctx);
 523
 524         /* On Windows, the SeCreateSymbolicLink privilege is required to create
 525          * symbolic links.  To be more friendly towards non-Administrator users,
 526          * we merely warn the user if symbolic links cannot be created due to
 527          * insufficient permissions or privileges, unless
 528          * WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS was provided.  */
 529 #ifdef __WIN32__
 530         if (ret && inode_is_symlink(inode) &&
 531             (errno == EACCES || errno == EPERM) &&
 532             !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS))
 533         {
 534                 WARNING("Can't set reparse data on \"%"TS"\": "
 535                         "Access denied!\n"
 536                         "          You may be trying to "
 537                         "extract a symbolic link without the\n"
 538                         "          SeCreateSymbolicLink privilege, "
 539                         "which by default non-Administrator\n"
 540                         "          accounts do not have.",
 541                         path);
 542                 ret = 0;
 543         }
 544 #endif
 545         if (ret)
 546                 goto error;
 547
 548         /* Account for reparse data consumed.  */
 549         update_extract_progress(ctx,
 550                                 (lte_override ? lte_override :
 551                                       inode_unnamed_lte_resolved(inode)));
 552         return 0;
 553
 554 error:
 555         ERROR_WITH_ERRNO("Failed to set reparse data on \"%"TS"\"", path);
 556         return ret;
 557 }
 558
 559 /*
 560  * Extract zero or more streams to a file.
 561  *
 562  * This function operates slightly differently depending on whether @lte_spec is
 563  * NULL or not.  When @lte_spec is NULL, the behavior is to extract the default
 564  * file contents (unnamed stream), and, if named data streams are supported in
 565  * the extract mode and volume, any named data streams.  When @lte_spec is not
 566  * NULL, the behavior is to extract only all copies of the stream @lte_spec, and
 567  * in addition use @lte_spec to set the reparse data or create the symbolic link
 568  * if appropriate.
 569  *
 570  * @path
 571  *      Path to file to extract (as can be passed to apply_operations
 572  *      functions).
 573  * @ctx
 574  *      Apply context.
 575  * @dentry
 576  *      WIM dentry that corresponds to the file being extracted.
 577  * @lte_spec
 578  *      If non-NULL, specifies the lookup table entry for a stream to extract,
 579  *      and only that stream will be extracted (although there may be more than
 580  *      one instance of it).
 581  * @lte_override
 582  *      Used only if @lte_spec != NULL; it is passed to the extraction functions
 583  *      rather than @lte_spec, allowing the location of the stream to be
 584  *      overridden.  (This is used when the WIM is being read from a nonseekable
 585  *      file, such as a pipe, when streams need to be used more than once; each
 586  *      such stream is extracted to a temporary file.)
 587  */
 588 static int
 589 extract_streams(const tchar *path, struct apply_ctx *ctx,
 590                 struct wim_dentry *dentry,
 591                 struct wim_lookup_table_entry *lte_spec,
 592                 struct wim_lookup_table_entry *lte_override)
 593 {
 594         struct wim_inode *inode = dentry->d_inode;
 595         struct wim_lookup_table_entry *lte;
 596         file_spec_t file_spec;
 597         int ret;
 598
 599         if (dentry->was_hardlinked)
 600                 return 0;
 601
 602 #ifdef ENABLE_DEBUG
 603         if (lte_spec) {
 604                 char sha1_str[100];
 605                 char *p = sha1_str;
 606                 for (unsigned i = 0; i < SHA1_HASH_SIZE; i++)
 607                         p += sprintf(p, "%02x", lte_override->hash[i]);
 608                 DEBUG("Extracting stream SHA1=%s to \"%"TS"\"",
 609                       sha1_str, path, inode->i_ino);
 610         } else {
 611                 DEBUG("Extracting streams to \"%"TS"\"", path, inode->i_ino);
 612         }
 613 #endif
 614
 615         if (ctx->ops->uses_cookies)
 616                 file_spec.cookie = inode->extract_cookie;
 617         else
 618                 file_spec.path = path;
 619
 620         /* Unnamed data stream.  */
 621         lte = inode_unnamed_lte_resolved(inode);
 622         if (lte && (!lte_spec || lte == lte_spec)) {
 623                 if (lte_spec)
 624                         lte = lte_override;
 625                 if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
 626                                              FILE_ATTRIBUTE_REPARSE_POINT)))
 627                 {
 628                         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED &&
 629                             ctx->supported_features.encrypted_files) {
 630                                 if (!ctx->ops->extract_encrypted_stream_creates_file) {
 631                                         ret = ctx->ops->extract_encrypted_stream(
 632                                                                 path, lte, ctx);
 633                                         if (ret)
 634                                                 goto error;
 635                                 }
 636                         } else {
 637                                 ret = ctx->ops->extract_unnamed_stream(
 638                                                         file_spec, lte, ctx);
 639                                 if (ret)
 640                                         goto error;
 641                         }
 642                         update_extract_progress(ctx, lte);
 643                 }
 644                 else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)
 645                 {
 646                         ret = 0;
 647                         if (ctx->supported_features.reparse_points)
 648                                 ret = extract_reparse_data(path, ctx, inode, lte);
 649                 #ifndef __WIN32__
 650                         else if ((inode_is_symlink(inode) &&
 651                                   ctx->supported_features.symlink_reparse_points))
 652                                 ret = extract_symlink(path, ctx, inode, lte);
 653                 #endif
 654                         if (ret)
 655                                 return ret;
 656                 }
 657         }
 658
 659         /* Named data streams.  */
 660         if (can_extract_named_data_streams(ctx)) {
 661                 for (u16 i = 0; i < inode->i_num_ads; i++) {
 662                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
 663
 664                         if (!ads_entry_is_named_stream(entry))
 665                                 continue;
 666                         lte = entry->lte;
 667                         if (!lte)
 668                                 continue;
 669                         if (lte_spec && lte_spec != lte)
 670                                 continue;
 671                         if (lte_spec)
 672                                 lte = lte_override;
 673                         ret = ctx->ops->extract_named_stream(file_spec, entry->stream_name,
 674                                                              entry->stream_name_nbytes / 2,
 675                                                              lte, ctx);
 676                         if (ret)
 677                                 goto error;
 678                         update_extract_progress(ctx, lte);
 679                 }
 680         }
 681         return 0;
 682
 683 error:
 684         ERROR_WITH_ERRNO("Failed to extract data of \"%"TS"\"", path);
 685         return ret;
 686 }
 687
 688 /* Set attributes on an extracted file or directory if supported by the
 689  * extraction mode.  */
 690 static int
 691 extract_file_attributes(const tchar *path, struct apply_ctx *ctx,
 692                         struct wim_dentry *dentry, unsigned pass)
 693 {
 694         int ret;
 695
 696         if (ctx->ops->set_file_attributes &&
 697             !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES) &&
 698             !(dentry == ctx->extract_root && ctx->root_dentry_is_special)) {
 699                 u32 attributes = dentry->d_inode->i_attributes;
 700
 701                 /* Clear unsupported attributes.  */
 702                 attributes &= ctx->supported_attributes_mask;
 703
 704                 if ((attributes & FILE_ATTRIBUTE_DIRECTORY &&
 705                      !ctx->supported_features.encrypted_directories) ||
 706                     (!(attributes & FILE_ATTRIBUTE_DIRECTORY) &&
 707                      !ctx->supported_features.encrypted_files))
 708                 {
 709                         attributes &= ~FILE_ATTRIBUTE_ENCRYPTED;
 710                 }
 711
 712                 if (attributes == 0)
 713                         attributes = FILE_ATTRIBUTE_NORMAL;
 714
 715                 ret = ctx->ops->set_file_attributes(path, attributes, ctx, pass);
 716                 if (ret) {
 717                         ERROR_WITH_ERRNO("Failed to set attributes on "
 718                                          "\"%"TS"\"", path);
 719                         return ret;
 720                 }
 721         }
 722         return 0;
 723 }
 724
 725
 726 /* Set or remove the short (DOS) name on an extracted file or directory if
 727  * supported by the extraction mode.  Since DOS names are unimportant and it's
 728  * easy to run into problems setting them on Windows (SetFileShortName()
 729  * requires SE_RESTORE privilege, which only the Administrator can request, and
 730  * also requires DELETE access to the file), failure is ignored unless
 731  * WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES is set.  */
 732 static int
 733 extract_short_name(const tchar *path, struct apply_ctx *ctx,
 734                    struct wim_dentry *dentry)
 735 {
 736         int ret;
 737
 738         /* The root of the dentry tree being extracted may not be extracted to
 739          * its original name, so its short name should be ignored.  */
 740         if (dentry == ctx->extract_root)
 741                 return 0;
 742
 743         if (ctx->supported_features.short_names) {
 744                 ret = ctx->ops->set_short_name(path,
 745                                                dentry->short_name,
 746                                                dentry->short_name_nbytes / 2,
 747                                                ctx);
 748                 if (ret && (ctx->extract_flags &
 749                             WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES))
 750                 {
 751                         ERROR_WITH_ERRNO("Failed to set short name of "
 752                                          "\"%"TS"\"", path);
 753                         return ret;
 754                 }
 755         }
 756         return 0;
 757 }
 758
 759 /* Set security descriptor, UNIX data, or neither on an extracted file, taking
 760  * into account the current extraction mode and flags.  */
 761 static int
 762 extract_security(const tchar *path, struct apply_ctx *ctx,
 763                  struct wim_dentry *dentry)
 764 {
 765         int ret;
 766         struct wim_inode *inode = dentry->d_inode;
 767
 768         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS)
 769                 return 0;
 770
 771         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
 772                 return 0;
 773
 774 #ifndef __WIN32__
 775         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 776                 struct wimlib_unix_data data;
 777
 778                 ret = inode_get_unix_data(inode, &data, NULL);
 779                 if (ret < 0)
 780                         ret = 0;
 781                 else if (ret == 0)
 782                         ret = ctx->ops->set_unix_data(path, &data, ctx);
 783                 if (ret) {
 784                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
 785                                 ERROR_WITH_ERRNO("Failed to set UNIX owner, "
 786                                                  "group, and/or mode on "
 787                                                  "\"%"TS"\"", path);
 788                                 return ret;
 789                         } else {
 790                                 WARNING_WITH_ERRNO("Failed to set UNIX owner, "
 791                                                    "group, and/or/mode on "
 792                                                    "\"%"TS"\"", path);
 793                         }
 794                 }
 795         }
 796         else
 797 #endif /* __WIN32__ */
 798         if (ctx->supported_features.security_descriptors &&
 799             inode->i_security_id != -1)
 800         {
 801                 const struct wim_security_data *sd;
 802                 const u8 *desc;
 803                 size_t desc_size;
 804
 805                 sd = wim_const_security_data(ctx->wim);
 806                 desc = sd->descriptors[inode->i_security_id];
 807                 desc_size = sd->sizes[inode->i_security_id];
 808
 809                 ret = ctx->ops->set_security_descriptor(path, desc,
 810                                                         desc_size, ctx);
 811                 if (ret) {
 812                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
 813                                 ERROR_WITH_ERRNO("Failed to set security "
 814                                                  "descriptor on \"%"TS"\"", path);
 815                                 return ret;
 816                         } else {
 817                         #if 0
 818                                 if (errno != EACCES) {
 819                                         WARNING_WITH_ERRNO("Failed to set "
 820                                                            "security descriptor "
 821                                                            "on \"%"TS"\"", path);
 822                                 }
 823                         #endif
 824                                 ctx->no_security_descriptors++;
 825                         }
 826                 }
 827         }
 828         return 0;
 829 }
 830
 831 /* Set timestamps on an extracted file.  Failure is warning-only unless
 832  * WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS is set.  */
 833 static int
 834 extract_timestamps(const tchar *path, struct apply_ctx *ctx,
 835                    struct wim_dentry *dentry)
 836 {
 837         struct wim_inode *inode = dentry->d_inode;
 838         int ret;
 839
 840         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
 841                 return 0;
 842
 843         if (ctx->ops->set_timestamps) {
 844                 ret = ctx->ops->set_timestamps(path,
 845                                                inode->i_creation_time,
 846                                                inode->i_last_write_time,
 847                                                inode->i_last_access_time,
 848                                                ctx);
 849                 if (ret) {
 850                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) {
 851                                 ERROR_WITH_ERRNO("Failed to set timestamps "
 852                                                  "on \"%"TS"\"", path);
 853                                 return ret;
 854                         } else {
 855                                 WARNING_WITH_ERRNO("Failed to set timestamps "
 856                                                    "on \"%"TS"\"", path);
 857                         }
 858                 }
 859         }
 860         return 0;
 861 }
 862
 863 /* Check whether the extraction of a dentry should be skipped completely.  */
 864 static bool
 865 dentry_is_supported(struct wim_dentry *dentry,
 866                     const struct wim_features *supported_features)
 867 {
 868         struct wim_inode *inode = dentry->d_inode;
 869
 870         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 871                 return supported_features->reparse_points ||
 872                         (inode_is_symlink(inode) &&
 873                          supported_features->symlink_reparse_points);
 874         }
 875         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
 876                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
 877                         return supported_features->encrypted_directories != 0;
 878                 else
 879                         return supported_features->encrypted_files != 0;
 880         }
 881         return true;
 882 }
 883
 884 /* Given a WIM dentry to extract, build the path to which to extract it, in the
 885  * format understood by the callbacks in the apply_operations being used.
 886  *
 887  * Write the resulting path into @path, which must have room for at least
 888  * ctx->ops->max_path characters including the null-terminator.
 889  *
 890  * Return %true if successful; %false if this WIM dentry doesn't actually need
 891  * to be extracted or if the calculated path exceeds ctx->ops->max_path
 892  * characters.
 893  *
 894  * This function clobbers the tmp_list member of @dentry and its ancestors up
 895  * until the extraction root.  */
 896 static bool
 897 build_extraction_path(tchar path[], struct wim_dentry *dentry,
 898                       struct apply_ctx *ctx)
 899 {
 900         size_t path_nchars;
 901         LIST_HEAD(ancestor_list);
 902         tchar *p = path;
 903         const tchar *target_prefix;
 904         size_t target_prefix_nchars;
 905         struct wim_dentry *d;
 906
 907         if (dentry->extraction_skipped)
 908                 return false;
 909
 910         path_nchars = ctx->ops->path_prefix_nchars;
 911
 912         if (ctx->ops->requires_realtarget_in_paths) {
 913                 target_prefix        = ctx->realtarget;
 914                 target_prefix_nchars = ctx->realtarget_nchars;
 915         } else if (ctx->ops->requires_target_in_paths) {
 916                 target_prefix        = ctx->target;
 917                 target_prefix_nchars = ctx->target_nchars;
 918         } else {
 919                 target_prefix        = NULL;
 920                 target_prefix_nchars = 0;
 921         }
 922         path_nchars += target_prefix_nchars;
 923
 924         for (d = dentry; d != ctx->extract_root; d = d->parent) {
 925                 if (!d->in_extraction_tree || d->extraction_skipped)
 926                         break;
 927
 928                 path_nchars += d->extraction_name_nchars + 1;
 929                 list_add(&d->tmp_list, &ancestor_list);
 930         }
 931
 932         path_nchars++; /* null terminator */
 933
 934         if (path_nchars > ctx->ops->path_max) {
 935                 WARNING("\"%"TS"\": Path too long to extract",
 936                         dentry_full_path(dentry));
 937                 return false;
 938         }
 939
 940         p = tmempcpy(p, ctx->ops->path_prefix, ctx->ops->path_prefix_nchars);
 941         p = tmempcpy(p, target_prefix, target_prefix_nchars);
 942         list_for_each_entry(d, &ancestor_list, tmp_list) {
 943                 *p++ = ctx->ops->path_separator;
 944                 p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars);
 945         }
 946         *p++ = T('\0');
 947         wimlib_assert(p - path == path_nchars);
 948         return true;
 949 }
 950
 951 static unsigned
 952 get_num_path_components(const tchar *path, tchar path_separator)
 953 {
 954         unsigned num_components = 0;
 955 #ifdef __WIN32__
 956         /* Ignore drive letter.  */
 957         if (path[0] != L'\0' && path[1] == L':')
 958                 path += 2;
 959 #endif
 960
 961         while (*path) {
 962                 while (*path == path_separator)
 963                         path++;
 964                 if (*path)
 965                         num_components++;
 966                 while (*path && *path != path_separator)
 967                         path++;
 968         }
 969         return num_components;
 970 }
 971
 972 static int
 973 extract_multiimage_symlink(const tchar *oldpath, const tchar *newpath,
 974                            struct apply_ctx *ctx, struct wim_dentry *dentry)
 975 {
 976         size_t num_raw_path_components;
 977         const struct wim_dentry *d;
 978         size_t num_target_path_components;
 979         tchar *p;
 980         const tchar *p_old;
 981         int ret;
 982
 983         num_raw_path_components = 0;
 984         for (d = dentry; d != ctx->extract_root; d = d->parent)
 985                 num_raw_path_components++;
 986
 987         if (ctx->ops->requires_realtarget_in_paths)
 988                 num_target_path_components = get_num_path_components(ctx->realtarget,
 989                                                                      ctx->ops->path_separator);
 990         else if (ctx->ops->requires_target_in_paths)
 991                 num_target_path_components = get_num_path_components(ctx->target,
 992                                                                      ctx->ops->path_separator);
 993         else
 994                 num_target_path_components = 0;
 995
 996         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
 997                 wimlib_assert(num_target_path_components > 0);
 998                 num_raw_path_components++;
 999                 num_target_path_components--;
1000         }
1001
1002         p_old = oldpath + ctx->ops->path_prefix_nchars;
1003 #ifdef __WIN32__
1004         if (p_old[0] != L'\0' && p_old[1] == ':')
1005                 p_old += 2;
1006 #endif
1007         while (*p_old == ctx->ops->path_separator)
1008                 p_old++;
1009         while (--num_target_path_components) {
1010                 while (*p_old != ctx->ops->path_separator)
1011                         p_old++;
1012                 while (*p_old == ctx->ops->path_separator)
1013                         p_old++;
1014         }
1015
1016         tchar symlink_target[tstrlen(p_old) + 3 * num_raw_path_components + 1];
1017
1018         p = &symlink_target[0];
1019         while (num_raw_path_components--) {
1020                 *p++ = '.';
1021                 *p++ = '.';
1022                 *p++ = ctx->ops->path_separator;
1023         }
1024         tstrcpy(p, p_old);
1025         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
1026               newpath, symlink_target);
1027         ret = ctx->ops->create_symlink(symlink_target, newpath, ctx);
1028         if (ret) {
1029                 ERROR_WITH_ERRNO("Failed to create symlink "
1030                                  "\"%"TS"\" => \"%"TS"\"",
1031                                  newpath, symlink_target);
1032         }
1033         return ret;
1034 }
1035
1036 /* Create the "skeleton" of an extracted file or directory.  Don't yet extract
1037  * data streams, reparse data (including symbolic links), timestamps, and
1038  * security descriptors.  Basically, everything that doesn't require reading
1039  * non-metadata resources from the WIM file and isn't delayed until the final
1040  * pass.  */
1041 static int
1042 do_dentry_extract_skeleton(tchar path[], struct wim_dentry *dentry,
1043                            struct apply_ctx *ctx)
1044 {
1045         struct wim_inode *inode = dentry->d_inode;
1046         int ret;
1047         const tchar *oldpath;
1048
1049         if (unlikely(is_linked_extraction(ctx))) {
1050                 struct wim_lookup_table_entry *unnamed_lte;
1051
1052                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1053                 if (unnamed_lte && unnamed_lte->extracted_file) {
1054                         oldpath = unnamed_lte->extracted_file;
1055                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK)
1056                                 goto hardlink;
1057                         else
1058                                 goto symlink;
1059                 }
1060         }
1061
1062         /* Create hard link if this dentry corresponds to an already-extracted
1063          * inode.  */
1064         if (inode->i_extracted_file) {
1065                 oldpath = inode->i_extracted_file;
1066                 goto hardlink;
1067         }
1068
1069         /* Skip symlinks unless they can be extracted as reparse points rather
1070          * than created directly.  */
1071         if (inode_is_symlink(inode) && !ctx->supported_features.reparse_points)
1072                 return 0;
1073
1074         /* Create this file or directory unless it's the extraction root, which
1075          * was already created if necessary.  */
1076         if (dentry != ctx->extract_root) {
1077                 ret = extract_inode(path, ctx, inode);
1078                 if (ret)
1079                         return ret;
1080         }
1081
1082         /* Create empty named data streams.  */
1083         if (can_extract_named_data_streams(ctx)) {
1084                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1085                         file_spec_t file_spec;
1086                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
1087
1088                         if (!ads_entry_is_named_stream(entry))
1089                                 continue;
1090                         if (entry->lte)
1091                                 continue;
1092                         if (ctx->ops->uses_cookies)
1093                                 file_spec.cookie = inode->extract_cookie;
1094                         else
1095                                 file_spec.path = path;
1096                         ret = ctx->ops->extract_named_stream(file_spec,
1097                                                              entry->stream_name,
1098                                                              entry->stream_name_nbytes / 2,
1099                                                              entry->lte, ctx);
1100                         if (ret) {
1101                                 ERROR_WITH_ERRNO("\"%"TS"\": failed to create "
1102                                                  "empty named data stream",
1103                                                  path);
1104                                 return ret;
1105                         }
1106                 }
1107         }
1108
1109         /* Set file attributes (if supported).  */
1110         ret = extract_file_attributes(path, ctx, dentry, 0);
1111         if (ret)
1112                 return ret;
1113
1114         /* Set or remove file short name (if supported).  */
1115         ret = extract_short_name(path, ctx, dentry);
1116         if (ret)
1117                 return ret;
1118
1119         /* If inode has multiple links and hard links are supported in this
1120          * extraction mode and volume, save the path to the extracted file in
1121          * case it's needed to create a hard link.  */
1122         if (unlikely(is_linked_extraction(ctx))) {
1123                 struct wim_lookup_table_entry *unnamed_lte;
1124
1125                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1126                 if (unnamed_lte) {
1127                         unnamed_lte->extracted_file = TSTRDUP(path);
1128                         if (!unnamed_lte->extracted_file)
1129                                 return WIMLIB_ERR_NOMEM;
1130                 }
1131         } else if (inode->i_nlink > 1 && ctx->supported_features.hard_links) {
1132                 inode->i_extracted_file = TSTRDUP(path);
1133                 if (!inode->i_extracted_file)
1134                         return WIMLIB_ERR_NOMEM;
1135         }
1136         return 0;
1137
1138 symlink:
1139         ret = extract_multiimage_symlink(oldpath, path, ctx, dentry);
1140         if (ret)
1141                 return ret;
1142         dentry->was_hardlinked = 1;
1143         return 0;
1144
1145 hardlink:
1146         ret = extract_hardlink(oldpath, path, ctx);
1147         if (ret)
1148                 return ret;
1149         dentry->was_hardlinked = 1;
1150         return 0;
1151 }
1152
1153 /* This is a wrapper around do_dentry_extract_skeleton() that handles building
1154  * the path, doing short name reordering.  This is also idempotent; dentries
1155  * already processed have skeleton_extracted set and no action is taken.  See
1156  * apply_operations.requires_short_name_reordering for more details about short
1157  * name reordering.  */
1158 static int
1159 dentry_extract_skeleton(struct wim_dentry *dentry, void *_ctx)
1160 {
1161         struct apply_ctx *ctx = _ctx;
1162         tchar path[ctx->ops->path_max];
1163         struct wim_dentry *orig_dentry;
1164         struct wim_dentry *other_dentry;
1165         int ret;
1166
1167         if (dentry->skeleton_extracted)
1168                 return 0;
1169
1170         orig_dentry = NULL;
1171         if (ctx->supported_features.short_names
1172             && ctx->ops->requires_short_name_reordering
1173             && !dentry_has_short_name(dentry)
1174             && !dentry->d_inode->i_dos_name_extracted)
1175         {
1176                 inode_for_each_dentry(other_dentry, dentry->d_inode) {
1177                         if (dentry_has_short_name(other_dentry)
1178                             && !other_dentry->skeleton_extracted
1179                             && other_dentry->in_extraction_tree
1180                             && !other_dentry->extraction_skipped)
1181                         {
1182                                 DEBUG("Creating %"TS" before %"TS" "
1183                                       "to guarantee correct DOS name extraction",
1184                                       dentry_full_path(other_dentry),
1185                                       dentry_full_path(dentry));
1186                                 orig_dentry = dentry;
1187                                 dentry = other_dentry;
1188                                 break;
1189                         }
1190                 }
1191         }
1192 again:
1193         if (!build_extraction_path(path, dentry, ctx))
1194                 return 0;
1195         ret = do_dentry_extract_skeleton(path, dentry, ctx);
1196         if (ret)
1197                 return ret;
1198
1199         dentry->skeleton_extracted = 1;
1200
1201         if (orig_dentry) {
1202                 dentry = orig_dentry;
1203                 orig_dentry = NULL;
1204                 goto again;
1205         }
1206         dentry->d_inode->i_dos_name_extracted = 1;
1207         return 0;
1208 }
1209
1210 static int
1211 dentry_extract_dir_skeleton(struct wim_dentry *dentry, void *_ctx)
1212 {
1213         if (dentry->d_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1214                 return dentry_extract_skeleton(dentry, _ctx);
1215         return 0;
1216 }
1217
1218 /* Create a file or directory, then immediately extract all streams.  The WIM
1219  * may not be read sequentially by this function.  */
1220 static int
1221 dentry_extract(struct wim_dentry *dentry, void *_ctx)
1222 {
1223         struct apply_ctx *ctx = _ctx;
1224         tchar path[ctx->ops->path_max];
1225         int ret;
1226
1227         ret = dentry_extract_skeleton(dentry, ctx);
1228         if (ret)
1229                 return ret;
1230
1231         if (!build_extraction_path(path, dentry, ctx))
1232                 return 0;
1233
1234         return extract_streams(path, ctx, dentry, NULL, NULL);
1235 }
1236
1237 /* Creates a temporary file opened for writing.  The open file descriptor is
1238  * returned in @fd_ret and its name is returned in @name_ret (dynamically
1239  * allocated).  */
1240 static int
1241 create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
1242 {
1243         tchar *name;
1244         int raw_fd;
1245
1246 retry:
1247         name = ttempnam(NULL, T("wimlib"));
1248         if (name == NULL) {
1249                 ERROR_WITH_ERRNO("Failed to create temporary filename");
1250                 return WIMLIB_ERR_NOMEM;
1251         }
1252
1253         raw_fd = topen(name, O_WRONLY | O_CREAT | O_EXCL | O_BINARY, 0600);
1254
1255         if (raw_fd < 0) {
1256                 if (errno == EEXIST) {
1257                         FREE(name);
1258                         goto retry;
1259                 }
1260                 ERROR_WITH_ERRNO("Failed to open temporary file \"%"TS"\"", name);
1261                 FREE(name);
1262                 return WIMLIB_ERR_OPEN;
1263         }
1264
1265         filedes_init(fd_ret, raw_fd);
1266         *name_ret = name;
1267         return 0;
1268 }
1269
1270 /* Extract all instances of the stream @lte that are being extracted in this
1271  * call of extract_tree(), but actually read the stream data from @lte_override.
1272  */
1273 static int
1274 extract_stream_instances(struct wim_lookup_table_entry *lte,
1275                          struct wim_lookup_table_entry *lte_override,
1276                          struct apply_ctx *ctx)
1277 {
1278         struct wim_dentry **lte_dentries;
1279         tchar path[ctx->ops->path_max];
1280         size_t i;
1281         int ret;
1282
1283         if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries))
1284                 lte_dentries = lte->inline_lte_dentries;
1285         else
1286                 lte_dentries = lte->lte_dentries;
1287
1288         for (i = 0; i < lte->out_refcnt; i++) {
1289                 struct wim_dentry *dentry = lte_dentries[i];
1290
1291                 if (dentry->tmp_flag)
1292                         continue;
1293                 if (!build_extraction_path(path, dentry, ctx))
1294                         continue;
1295                 ret = extract_streams(path, ctx, dentry, lte, lte_override);
1296                 if (ret)
1297                         goto out_clear_tmp_flags;
1298                 dentry->tmp_flag = 1;
1299         }
1300         ret = 0;
1301 out_clear_tmp_flags:
1302         for (i = 0; i < lte->out_refcnt; i++)
1303                 lte_dentries[i]->tmp_flag = 0;
1304         return ret;
1305 }
1306
1307 /* Determine whether the specified stream needs to be extracted to a temporary
1308  * file or not.
1309  *
1310  * @lte->out_refcnt specifies the number of instances of this stream that must
1311  * be extracted.
1312  *
1313  * @is_partial_res is %true if this stream is just one of multiple in a single
1314  * WIM resource being extracted.  */
1315 static bool
1316 need_tmpfile_to_extract(struct wim_lookup_table_entry *lte,
1317                         bool is_partial_res)
1318 {
1319         /* Temporary file is always required when reading a partial resource,
1320          * since in that case we retrieve all the contained streams in one pass.
1321          * */
1322         if (is_partial_res)
1323                 return true;
1324
1325         /* Otherwise we don't need a temporary file if only a single instance of
1326          * the stream is needed.  */
1327         if (lte->out_refcnt == 1)
1328                 return false;
1329
1330         wimlib_assert(lte->out_refcnt >= 2);
1331
1332         /* We also don't need a temporary file if random access to the stream is
1333          * allowed.  */
1334         if (lte->resource_location != RESOURCE_IN_WIM ||
1335             filedes_is_seekable(&lte->rspec->wim->in_fd))
1336                 return false;
1337
1338         return true;
1339 }
1340
1341 static int
1342 begin_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1343                                 bool is_partial_res,
1344                                 void *_ctx)
1345 {
1346         struct apply_ctx *ctx = _ctx;
1347         int ret;
1348
1349         if (!need_tmpfile_to_extract(lte, is_partial_res)) {
1350                 DEBUG("Temporary file not needed "
1351                       "for stream (size=%"PRIu64")", lte->size);
1352                 ret = extract_stream_instances(lte, lte, ctx);
1353                 if (ret)
1354                         return ret;
1355
1356                 return BEGIN_STREAM_STATUS_SKIP_STREAM;
1357         }
1358
1359         DEBUG("Temporary file needed for stream (size=%"PRIu64")", lte->size);
1360         return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
1361 }
1362
1363 static int
1364 end_extract_stream_to_tmpfile(struct wim_lookup_table_entry *lte,
1365                               int status, void *_ctx)
1366 {
1367         struct apply_ctx *ctx = _ctx;
1368         struct wim_lookup_table_entry lte_override;
1369         int ret;
1370         int errno_save = errno;
1371
1372         ret = filedes_close(&ctx->tmpfile_fd);
1373
1374         if (status) {
1375                 ret = status;
1376                 errno = errno_save;
1377                 goto out_delete_tmpfile;
1378         }
1379
1380         if (ret) {
1381                 ERROR_WITH_ERRNO("Error writing temporary file %"TS, ctx->tmpfile_name);
1382                 ret = WIMLIB_ERR_WRITE;
1383                 goto out_delete_tmpfile;
1384         }
1385
1386         /* Now that a full stream has been extracted to a temporary file,
1387          * extract all instances of it to the actual target.  */
1388
1389         memcpy(&lte_override, lte, sizeof(struct wim_lookup_table_entry));
1390         lte_override.resource_location = RESOURCE_IN_FILE_ON_DISK;
1391         lte_override.file_on_disk = ctx->tmpfile_name;
1392
1393         ret = extract_stream_instances(lte, &lte_override, ctx);
1394
1395 out_delete_tmpfile:
1396         errno_save = errno;
1397         tunlink(ctx->tmpfile_name);
1398         FREE(ctx->tmpfile_name);
1399         errno = errno_save;
1400         return ret;
1401 }
1402
1403 /* Extracts a list of streams (ctx.stream_list), assuming that the directory
1404  * structure and empty files were already created.  This relies on the
1405  * per-`struct wim_lookup_table_entry' list of dentries that reference each
1406  * stream that was constructed earlier.  */
1407 static int
1408 extract_stream_list(struct apply_ctx *ctx)
1409 {
1410         if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
1411                 /* Sequential extraction: read the streams in the order in which
1412                  * they appear in the WIM file.  */
1413                 struct read_stream_list_callbacks cbs = {
1414                         .begin_stream           = begin_extract_stream_to_tmpfile,
1415                         .begin_stream_ctx       = ctx,
1416                         .consume_chunk          = extract_chunk_to_fd,
1417                         .consume_chunk_ctx      = &ctx->tmpfile_fd,
1418                         .end_stream             = end_extract_stream_to_tmpfile,
1419                         .end_stream_ctx         = ctx,
1420                 };
1421                 return read_stream_list(&ctx->stream_list,
1422                                         offsetof(struct wim_lookup_table_entry, extraction_list),
1423                                         &cbs, VERIFY_STREAM_HASHES);
1424         } else {
1425                 /* Extract the streams in unsorted order.  */
1426                 struct wim_lookup_table_entry *lte;
1427                 int ret;
1428
1429                 list_for_each_entry(lte, &ctx->stream_list, extraction_list) {
1430                         ret = extract_stream_instances(lte, lte, ctx);
1431                         if (ret)
1432                                 return ret;
1433                 }
1434                 return 0;
1435         }
1436 }
1437
1438 #define PWM_ALLOW_WIM_HDR 0x00001
1439 #define PWM_SILENT_EOF    0x00002
1440
1441 /* Read the header from a stream in a pipable WIM.  */
1442 static int
1443 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
1444                        struct wim_resource_spec *rspec,
1445                        int flags, struct wim_header_disk *hdr_ret)
1446 {
1447         union {
1448                 struct pwm_stream_hdr stream_hdr;
1449                 struct wim_header_disk pwm_hdr;
1450         } buf;
1451         struct wim_reshdr reshdr;
1452         int ret;
1453
1454         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
1455         if (ret)
1456                 goto read_error;
1457
1458         if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
1459                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
1460                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
1461                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
1462
1463                 if (ret)
1464                         goto read_error;
1465                 lte->resource_location = RESOURCE_NONEXISTENT;
1466                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
1467                 return 0;
1468         }
1469
1470         if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
1471                 ERROR("Data read on pipe is invalid (expected stream header).");
1472                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1473         }
1474
1475         copy_hash(lte->hash, buf.stream_hdr.hash);
1476
1477         reshdr.size_in_wim = 0;
1478         reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
1479         reshdr.offset_in_wim = pwm->in_fd.offset;
1480         reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
1481         wim_res_hdr_to_spec(&reshdr, pwm, rspec);
1482         lte_bind_wim_resource_spec(lte, rspec);
1483         lte->flags = rspec->flags;
1484         lte->size = rspec->uncompressed_size;
1485         lte->offset_in_res = 0;
1486         return 0;
1487
1488 read_error:
1489         if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
1490                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
1491         return ret;
1492 }
1493
1494 static int
1495 extract_streams_from_pipe(struct apply_ctx *ctx)
1496 {
1497         struct wim_lookup_table_entry *found_lte;
1498         struct wim_resource_spec *rspec;
1499         struct wim_lookup_table_entry *needed_lte;
1500         struct wim_lookup_table *lookup_table;
1501         struct wim_header_disk pwm_hdr;
1502         int ret;
1503         int pwm_flags;
1504
1505         ret = WIMLIB_ERR_NOMEM;
1506         found_lte = new_lookup_table_entry();
1507         if (found_lte == NULL)
1508                 goto out;
1509
1510         rspec = MALLOC(sizeof(struct wim_resource_spec));
1511         if (rspec == NULL)
1512                 goto out_free_found_lte;
1513
1514         lookup_table = ctx->wim->lookup_table;
1515         pwm_flags = PWM_ALLOW_WIM_HDR;
1516         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1517                 pwm_flags |= PWM_SILENT_EOF;
1518         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN);
1519         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
1520         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
1521         if (ctx->progress_func)
1522                 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1523                                    &ctx->progress);
1524         while (ctx->num_streams_remaining) {
1525                 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
1526                         lte_unbind_wim_resource_spec(found_lte);
1527                 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
1528                                              pwm_flags, &pwm_hdr);
1529                 if (ret) {
1530                         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE &&
1531                             (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1532                         {
1533                                 goto resume_done;
1534                         }
1535                         goto out_free_found_lte;
1536                 }
1537
1538                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
1539                     && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
1540                     && (needed_lte = lookup_stream(lookup_table, found_lte->hash))
1541                     && (needed_lte->out_refcnt))
1542                 {
1543                         tchar *tmpfile_name = NULL;
1544                         struct wim_lookup_table_entry *lte_override;
1545                         struct wim_lookup_table_entry tmpfile_lte;
1546
1547                         needed_lte->offset_in_res = found_lte->offset_in_res;
1548                         needed_lte->flags = found_lte->flags;
1549                         needed_lte->size = found_lte->size;
1550
1551                         lte_unbind_wim_resource_spec(found_lte);
1552                         lte_bind_wim_resource_spec(needed_lte, rspec);
1553
1554                         if (needed_lte->out_refcnt > 1) {
1555
1556                                 struct filedes tmpfile_fd;
1557
1558                                 /* Extract stream to temporary file.  */
1559                                 ret = create_temporary_file(&tmpfile_fd, &tmpfile_name);
1560                                 if (ret) {
1561                                         lte_unbind_wim_resource_spec(needed_lte);
1562                                         goto out_free_found_lte;
1563                                 }
1564
1565                                 ret = extract_full_stream_to_fd(needed_lte,
1566                                                                 &tmpfile_fd);
1567                                 if (ret) {
1568                                         filedes_close(&tmpfile_fd);
1569                                         goto delete_tmpfile;
1570                                 }
1571
1572                                 if (filedes_close(&tmpfile_fd)) {
1573                                         ERROR_WITH_ERRNO("Error writing to temporary "
1574                                                          "file \"%"TS"\"", tmpfile_name);
1575                                         ret = WIMLIB_ERR_WRITE;
1576                                         goto delete_tmpfile;
1577                                 }
1578                                 memcpy(&tmpfile_lte, needed_lte,
1579                                        sizeof(struct wim_lookup_table_entry));
1580                                 tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK;
1581                                 tmpfile_lte.file_on_disk = tmpfile_name;
1582                                 lte_override = &tmpfile_lte;
1583                         } else {
1584                                 lte_override = needed_lte;
1585                         }
1586
1587                         ret = extract_stream_instances(needed_lte, lte_override, ctx);
1588                 delete_tmpfile:
1589                         lte_unbind_wim_resource_spec(needed_lte);
1590                         if (tmpfile_name) {
1591                                 tunlink(tmpfile_name);
1592                                 FREE(tmpfile_name);
1593                         }
1594                         if (ret)
1595                                 goto out_free_found_lte;
1596                         ctx->num_streams_remaining--;
1597                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
1598                         ret = skip_wim_stream(found_lte);
1599                         if (ret)
1600                                 goto out_free_found_lte;
1601                 } else {
1602                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
1603                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
1604
1605                         if (part_number != ctx->progress.extract.part_number ||
1606                             total_parts != ctx->progress.extract.total_parts ||
1607                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
1608                                    WIM_GID_LEN))
1609                         {
1610                                 ctx->progress.extract.part_number = part_number;
1611                                 ctx->progress.extract.total_parts = total_parts;
1612                                 memcpy(ctx->progress.extract.guid,
1613                                        pwm_hdr.guid, WIM_GID_LEN);
1614                                 if (ctx->progress_func) {
1615                                         ctx->progress_func(
1616                                                 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1617                                                            &ctx->progress);
1618                                 }
1619
1620                         }
1621                 }
1622         }
1623         ret = 0;
1624 out_free_found_lte:
1625         if (found_lte->resource_location != RESOURCE_IN_WIM)
1626                 FREE(rspec);
1627         free_lookup_table_entry(found_lte);
1628 out:
1629         return ret;
1630
1631 resume_done:
1632         /* TODO */
1633         return 0;
1634 }
1635
1636 /* Finish extracting a file, directory, or symbolic link by setting file
1637  * security and timestamps.  */
1638 static int
1639 dentry_extract_final(struct wim_dentry *dentry, void *_ctx)
1640 {
1641         struct apply_ctx *ctx = _ctx;
1642         int ret;
1643         tchar path[ctx->ops->path_max];
1644
1645         if (!build_extraction_path(path, dentry, ctx))
1646                 return 0;
1647
1648         ret = extract_security(path, ctx, dentry);
1649         if (ret)
1650                 return ret;
1651
1652         if (ctx->ops->requires_final_set_attributes_pass) {
1653                 /* Set file attributes (if supported).  */
1654                 ret = extract_file_attributes(path, ctx, dentry, 1);
1655                 if (ret)
1656                         return ret;
1657         }
1658
1659         return extract_timestamps(path, ctx, dentry);
1660 }
1661
1662 /*
1663  * Extract a WIM dentry to standard output.
1664  *
1665  * This obviously doesn't make sense in all cases.  We return an error if the
1666  * dentry does not correspond to a regular file.  Otherwise we extract the
1667  * unnamed data stream only.
1668  */
1669 static int
1670 extract_dentry_to_stdout(struct wim_dentry *dentry)
1671 {
1672         int ret = 0;
1673         if (dentry->d_inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
1674                                              FILE_ATTRIBUTE_DIRECTORY))
1675         {
1676                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
1677                       "extracted to standard output", dentry_full_path(dentry));
1678                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
1679         } else {
1680                 struct wim_lookup_table_entry *lte;
1681
1682                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
1683                 if (lte) {
1684                         struct filedes _stdout;
1685                         filedes_init(&_stdout, STDOUT_FILENO);
1686                         ret = extract_full_stream_to_fd(lte, &_stdout);
1687                 }
1688         }
1689         return ret;
1690 }
1691
1692 #ifdef __WIN32__
1693 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
1694 #else
1695 static const utf16lechar replacement_char = cpu_to_le16('?');
1696 #endif
1697
1698 static bool
1699 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
1700 {
1701         size_t i;
1702
1703         if (num_chars == 0)
1704                 return true;
1705         for (i = 0; i < num_chars; i++) {
1706                 switch (name[i]) {
1707         #ifdef __WIN32__
1708                 case cpu_to_le16('\\'):
1709                 case cpu_to_le16(':'):
1710                 case cpu_to_le16('*'):
1711                 case cpu_to_le16('?'):
1712                 case cpu_to_le16('"'):
1713                 case cpu_to_le16('<'):
1714                 case cpu_to_le16('>'):
1715                 case cpu_to_le16('|'):
1716         #endif
1717                 case cpu_to_le16('/'):
1718                 case cpu_to_le16('\0'):
1719                         if (fix)
1720                                 name[i] = replacement_char;
1721                         else
1722                                 return false;
1723                 }
1724         }
1725
1726 #ifdef __WIN32__
1727         if (name[num_chars - 1] == cpu_to_le16(' ') ||
1728             name[num_chars - 1] == cpu_to_le16('.'))
1729         {
1730                 if (fix)
1731                         name[num_chars - 1] = replacement_char;
1732                 else
1733                         return false;
1734         }
1735 #endif
1736         return true;
1737 }
1738
1739 static int
1740 dentry_mark_skipped(struct wim_dentry *dentry, void *_ignore)
1741 {
1742         dentry->extraction_skipped = 1;
1743         return 0;
1744 }
1745
1746 /*
1747  * dentry_calculate_extraction_path-
1748  *
1749  * Calculate the actual filename component at which a WIM dentry will be
1750  * extracted, handling invalid filenames "properly".
1751  *
1752  * dentry->extraction_name usually will be set the same as dentry->file_name (on
1753  * UNIX, converted into the platform's multibyte encoding).  However, if the
1754  * file name contains characters that are not valid on the current platform or
1755  * has some other format that is not valid, leave dentry->extraction_name as
1756  * NULL and set dentry->extraction_skipped to indicate that this dentry should
1757  * not be extracted, unless the appropriate flag
1758  * WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES is set in the extract flags, in
1759  * which case a substitute filename will be created and set instead.
1760  *
1761  * Conflicts with case-insensitive names on Windows are handled similarly; see
1762  * below.
1763  */
1764 static int
1765 dentry_calculate_extraction_path(struct wim_dentry *dentry, void *_args)
1766 {
1767         struct apply_ctx *ctx = _args;
1768         int ret;
1769
1770         if (dentry == ctx->extract_root || dentry->extraction_skipped)
1771                 return 0;
1772
1773         if (!dentry_is_supported(dentry, &ctx->supported_features))
1774                 goto skip_dentry;
1775
1776         if (!ctx->ops->supports_case_sensitive_filenames)
1777         {
1778                 struct wim_dentry *other;
1779                 list_for_each_entry(other, &dentry->case_insensitive_conflict_list,
1780                                     case_insensitive_conflict_list)
1781                 {
1782                         if (!other->extraction_skipped) {
1783                                 if (ctx->extract_flags &
1784                                     WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
1785                                         WARNING("\"%"TS"\" has the same "
1786                                                 "case-insensitive name as "
1787                                                 "\"%"TS"\"; extracting "
1788                                                 "dummy name instead",
1789                                                 dentry_full_path(dentry),
1790                                                 dentry_full_path(other));
1791                                         goto out_replace;
1792                                 } else {
1793                                         WARNING("Not extracting \"%"TS"\": "
1794                                                 "has same case-insensitive "
1795                                                 "name as \"%"TS"\"",
1796                                                 dentry_full_path(dentry),
1797                                                 dentry_full_path(other));
1798                                         goto skip_dentry;
1799                                 }
1800                         }
1801                 }
1802         }
1803
1804         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
1805 #if TCHAR_IS_UTF16LE
1806                 dentry->extraction_name = dentry->file_name;
1807                 dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
1808                 return 0;
1809 #else
1810                 return utf16le_to_tstr(dentry->file_name,
1811                                        dentry->file_name_nbytes,
1812                                        &dentry->extraction_name,
1813                                        &dentry->extraction_name_nchars);
1814 #endif
1815         } else {
1816                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
1817                 {
1818                         WARNING("\"%"TS"\" has an invalid filename "
1819                                 "that is not supported on this platform; "
1820                                 "extracting dummy name instead",
1821                                 dentry_full_path(dentry));
1822                         goto out_replace;
1823                 } else {
1824                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
1825                                 "that is not supported on this platform",
1826                                 dentry_full_path(dentry));
1827                         goto skip_dentry;
1828                 }
1829         }
1830
1831 out_replace:
1832         {
1833                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
1834
1835                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
1836                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
1837
1838                 tchar *tchar_name;
1839                 size_t tchar_nchars;
1840         #if TCHAR_IS_UTF16LE
1841                 tchar_name = utf16_name_copy;
1842                 tchar_nchars = dentry->file_name_nbytes / 2;
1843         #else
1844                 ret = utf16le_to_tstr(utf16_name_copy,
1845                                       dentry->file_name_nbytes,
1846                                       &tchar_name, &tchar_nchars);
1847                 if (ret)
1848                         return ret;
1849         #endif
1850                 size_t fixed_name_num_chars = tchar_nchars;
1851                 tchar fixed_name[tchar_nchars + 50];
1852
1853                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
1854                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
1855                                                  T(" (invalid filename #%lu)"),
1856                                                  ++ctx->invalid_sequence);
1857         #if !TCHAR_IS_UTF16LE
1858                 FREE(tchar_name);
1859         #endif
1860                 dentry->extraction_name = memdup(fixed_name,
1861                                                  2 * fixed_name_num_chars + 2);
1862                 if (!dentry->extraction_name)
1863                         return WIMLIB_ERR_NOMEM;
1864                 dentry->extraction_name_nchars = fixed_name_num_chars;
1865         }
1866         return 0;
1867
1868 skip_dentry:
1869         for_dentry_in_tree(dentry, dentry_mark_skipped, NULL);
1870         return 0;
1871 }
1872
1873 /* Clean up dentry and inode structure after extraction.  */
1874 static int
1875 dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
1876 {
1877         struct wim_inode *inode = dentry->d_inode;
1878
1879         dentry->in_extraction_tree = 0;
1880         dentry->extraction_skipped = 0;
1881         dentry->was_hardlinked = 0;
1882         dentry->skeleton_extracted = 0;
1883         inode->i_visited = 0;
1884         FREE(inode->i_extracted_file);
1885         inode->i_extracted_file = NULL;
1886         inode->i_dos_name_extracted = 0;
1887         if ((void*)dentry->extraction_name != (void*)dentry->file_name)
1888                 FREE(dentry->extraction_name);
1889         dentry->extraction_name = NULL;
1890         return 0;
1891 }
1892
1893 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1894 static int
1895 dentry_tally_features(struct wim_dentry *dentry, void *_features)
1896 {
1897         struct wim_features *features = _features;
1898         struct wim_inode *inode = dentry->d_inode;
1899
1900         if (dentry->extraction_skipped)
1901                 return 0;
1902
1903         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1904                 features->archive_files++;
1905         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1906                 features->hidden_files++;
1907         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1908                 features->system_files++;
1909         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1910                 features->compressed_files++;
1911         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
1912                 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
1913                         features->encrypted_directories++;
1914                 else
1915                         features->encrypted_files++;
1916         }
1917         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1918                 features->not_context_indexed_files++;
1919         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1920                 features->sparse_files++;
1921         if (inode_has_named_stream(inode))
1922                 features->named_data_streams++;
1923         if (inode->i_visited)
1924                 features->hard_links++;
1925         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1926                 features->reparse_points++;
1927                 if (inode_is_symlink(inode))
1928                         features->symlink_reparse_points++;
1929                 else
1930                         features->other_reparse_points++;
1931         }
1932         if (inode->i_security_id != -1)
1933                 features->security_descriptors++;
1934         if (dentry->short_name_nbytes)
1935                 features->short_names++;
1936         if (inode_has_unix_data(inode))
1937                 features->unix_data++;
1938         inode->i_visited = 1;
1939         return 0;
1940 }
1941
1942 /* Tally the features necessary to extract a dentry tree.  */
1943 static void
1944 dentry_tree_get_features(struct wim_dentry *root, struct wim_features *features)
1945 {
1946         memset(features, 0, sizeof(struct wim_features));
1947         for_dentry_in_tree(root, dentry_tally_features, features);
1948         dentry_tree_clear_inode_visited(root);
1949 }
1950
1951 static u32
1952 compute_supported_attributes_mask(const struct wim_features *supported_features)
1953 {
1954         u32 mask = ~(u32)0;
1955
1956         if (!supported_features->archive_files)
1957                 mask &= ~FILE_ATTRIBUTE_ARCHIVE;
1958
1959         if (!supported_features->hidden_files)
1960                 mask &= ~FILE_ATTRIBUTE_HIDDEN;
1961
1962         if (!supported_features->system_files)
1963                 mask &= ~FILE_ATTRIBUTE_SYSTEM;
1964
1965         if (!supported_features->not_context_indexed_files)
1966                 mask &= ~FILE_ATTRIBUTE_NOT_CONTENT_INDEXED;
1967
1968         if (!supported_features->compressed_files)
1969                 mask &= ~FILE_ATTRIBUTE_COMPRESSED;
1970
1971         if (!supported_features->sparse_files)
1972                 mask &= ~FILE_ATTRIBUTE_SPARSE_FILE;
1973
1974         if (!supported_features->reparse_points)
1975                 mask &= ~FILE_ATTRIBUTE_REPARSE_POINT;
1976
1977         return mask;
1978 }
1979
1980 static int
1981 do_feature_check(const struct wim_features *required_features,
1982                  const struct wim_features *supported_features,
1983                  int extract_flags,
1984                  const struct apply_operations *ops,
1985                  bool warn)
1986 {
1987         const tchar *loc = T("the extraction operation");;
1988         const tchar *mode = T("this extraction mode");
1989
1990         if (warn) {
1991                 /* We're an archive program, so theoretically we can do what we want
1992                  * with FILE_ATTRIBUTE_ARCHIVE (which is a dumb flag anyway).  Don't
1993                  * bother the user about it.  */
1994 #if 0
1995                 if (required_features->archive_files && !supported_features->archive_files)
1996                 {
1997                         WARNING(
1998                   "%lu files in %"TS" are marked as archived, but this attribute\n"
1999         "          is not supported in %"TS".",
2000                                 required_features->archive_files, loc, mode);
2001                 }
2002 #endif
2003
2004                 if (required_features->hidden_files && !supported_features->hidden_files)
2005                 {
2006                         WARNING(
2007                   "%lu files in %"TS" are marked as hidden, but this\n"
2008         "          attribute is not supported in %"TS".",
2009                                 required_features->hidden_files, loc, mode);
2010                 }
2011
2012                 if (required_features->system_files && !supported_features->system_files)
2013                 {
2014                         WARNING(
2015                   "%lu files in %"TS" are marked as system files,\n"
2016         "          but this attribute is not supported in %"TS".",
2017                                 required_features->system_files, loc, mode);
2018                 }
2019
2020                 if (required_features->compressed_files && !supported_features->compressed_files)
2021                 {
2022                         WARNING(
2023                   "%lu files in %"TS" are marked as being transparently\n"
2024         "          compressed, but transparent compression is not supported in\n"
2025         "          %"TS".  These files will be extracted as uncompressed.",
2026                                 required_features->compressed_files, loc, mode);
2027                 }
2028
2029                 if (required_features->encrypted_files && !supported_features->encrypted_files)
2030                 {
2031                         WARNING(
2032                   "%lu files in %"TS" are marked as being encrypted,\n"
2033         "           but encryption is not supported in %"TS".  These files\n"
2034         "           will not be extracted.",
2035                                 required_features->encrypted_files, loc, mode);
2036                 }
2037
2038                 if (required_features->encrypted_directories &&
2039                     !supported_features->encrypted_directories)
2040                 {
2041                         WARNING(
2042                   "%lu directories in %"TS" are marked as being encrypted,\n"
2043         "           but encryption is not supported in %"TS".\n"
2044         "           These directories will be extracted as unencrypted.",
2045                                 required_features->encrypted_directories, loc, mode);
2046                 }
2047
2048                 if (required_features->not_context_indexed_files &&
2049                     !supported_features->not_context_indexed_files)
2050                 {
2051                         WARNING(
2052                   "%lu files in %"TS" are marked as not content indexed,\n"
2053         "          but this attribute is not supported in %"TS".",
2054                                 required_features->not_context_indexed_files, loc, mode);
2055                 }
2056
2057                 if (required_features->sparse_files && !supported_features->sparse_files)
2058                 {
2059                         WARNING(
2060                   "%lu files in %"TS" are marked as sparse, but creating\n"
2061         "           sparse files is not supported in %"TS".  These files\n"
2062         "           will be extracted as non-sparse.",
2063                                 required_features->sparse_files, loc, mode);
2064                 }
2065
2066                 if (required_features->named_data_streams &&
2067                     !supported_features->named_data_streams)
2068                 {
2069                         WARNING(
2070                   "%lu files in %"TS" contain one or more alternate (named)\n"
2071         "          data streams, which are not supported in %"TS".\n"
2072         "          Alternate data streams will NOT be extracted.",
2073                                 required_features->named_data_streams, loc, mode);
2074                 }
2075
2076                 if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
2077                                               WIMLIB_EXTRACT_FLAG_SYMLINK)) &&
2078                     required_features->named_data_streams &&
2079                     supported_features->named_data_streams)
2080                 {
2081                         WARNING(
2082                   "%lu files in %"TS" contain one or more alternate (named)\n"
2083         "          data streams, which are not supported in linked extraction mode.\n"
2084         "          Alternate data streams will NOT be extracted.",
2085                                 required_features->named_data_streams, loc);
2086                 }
2087
2088                 if (required_features->hard_links && !supported_features->hard_links)
2089                 {
2090                         WARNING(
2091                   "%lu files in %"TS" are hard links, but hard links are\n"
2092         "          not supported in %"TS".  Hard links will be extracted as\n"
2093         "          duplicate copies of the linked files.",
2094                                 required_features->hard_links, loc, mode);
2095                 }
2096
2097                 if (required_features->reparse_points && !supported_features->reparse_points)
2098                 {
2099                         if (supported_features->symlink_reparse_points) {
2100                                 if (required_features->other_reparse_points) {
2101                                         WARNING(
2102                   "%lu files in %"TS" are reparse points that are neither\n"
2103         "          symbolic links nor junction points and are not supported in\n"
2104         "          %"TS".  These reparse points will not be extracted.",
2105                                                 required_features->other_reparse_points, loc,
2106                                                 mode);
2107                                 }
2108                         } else {
2109                                 WARNING(
2110                   "%lu files in %"TS" are reparse points, which are\n"
2111         "          not supported in %"TS" and will not be extracted.",
2112                                         required_features->reparse_points, loc, mode);
2113                         }
2114                 }
2115
2116                 if (required_features->security_descriptors &&
2117                     !supported_features->security_descriptors)
2118                 {
2119                         WARNING(
2120                   "%lu files in %"TS" have Windows NT security descriptors,\n"
2121         "          but extracting security descriptors is not supported in\n"
2122         "          %"TS".  No security descriptors will be extracted.",
2123                                 required_features->security_descriptors, loc, mode);
2124                 }
2125
2126                 if (required_features->short_names && !supported_features->short_names)
2127                 {
2128                         WARNING(
2129                   "%lu files in %"TS" have short (DOS) names, but\n"
2130         "          extracting short names is not supported in %"TS".\n"
2131         "          Short names will not be extracted.\n",
2132                                 required_features->short_names, loc, mode);
2133                 }
2134         }
2135
2136         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
2137             required_features->unix_data && !supported_features->unix_data)
2138         {
2139                 ERROR("Extracting UNIX data is not supported in %"TS, mode);
2140                 return WIMLIB_ERR_UNSUPPORTED;
2141         }
2142         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) &&
2143             required_features->short_names && !supported_features->short_names)
2144         {
2145                 ERROR("Extracting short names is not supported in %"TS"", mode);
2146                 return WIMLIB_ERR_UNSUPPORTED;
2147         }
2148         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
2149             !ops->set_timestamps)
2150         {
2151                 ERROR("Extracting timestamps is not supported in %"TS"", mode);
2152                 return WIMLIB_ERR_UNSUPPORTED;
2153         }
2154         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
2155                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
2156              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
2157             required_features->security_descriptors &&
2158             !supported_features->security_descriptors)
2159         {
2160                 ERROR("Extracting security descriptors is not supported in %"TS, mode);
2161                 return WIMLIB_ERR_UNSUPPORTED;
2162         }
2163
2164         if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) &&
2165             !supported_features->hard_links)
2166         {
2167                 ERROR("Hard link extraction mode requested, but "
2168                       "%"TS" does not support hard links!", mode);
2169                 return WIMLIB_ERR_UNSUPPORTED;
2170         }
2171
2172         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
2173             required_features->symlink_reparse_points &&
2174             !(supported_features->symlink_reparse_points ||
2175               supported_features->reparse_points))
2176         {
2177                 ERROR("Extracting symbolic links is not supported in %"TS, mode);
2178                 return WIMLIB_ERR_UNSUPPORTED;
2179         }
2180
2181         if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) &&
2182             !supported_features->symlink_reparse_points)
2183         {
2184                 ERROR("Symbolic link extraction mode requested, but "
2185                       "%"TS" does not support symbolic "
2186                       "links!", mode);
2187                 return WIMLIB_ERR_UNSUPPORTED;
2188         }
2189         return 0;
2190 }
2191
2192 static void
2193 do_extract_warnings(struct apply_ctx *ctx)
2194 {
2195         if (ctx->partial_security_descriptors == 0 &&
2196             ctx->no_security_descriptors == 0)
2197                 return;
2198
2199         WARNING("Extraction to \"%"TS"\" complete, but with one or more warnings:",
2200                 ctx->target);
2201         if (ctx->partial_security_descriptors != 0) {
2202                 WARNING("- Could only partially set the security descriptor\n"
2203                         "            on %lu files or directories.",
2204                         ctx->partial_security_descriptors);
2205         }
2206         if (ctx->no_security_descriptors != 0) {
2207                 WARNING("- Could not set security descriptor at all\n"
2208                         "            on %lu files or directories.",
2209                         ctx->no_security_descriptors);
2210         }
2211 #ifdef __WIN32__
2212         WARNING("To fully restore all security descriptors, run the program\n"
2213                 "          with Administrator rights.");
2214 #endif
2215 }
2216
2217 static int
2218 dentry_set_skipped(struct wim_dentry *dentry, void *_ignore)
2219 {
2220         dentry->in_extraction_tree = 1;
2221         dentry->extraction_skipped = 1;
2222         return 0;
2223 }
2224
2225 static int
2226 dentry_set_not_skipped(struct wim_dentry *dentry, void *_ignore)
2227 {
2228         dentry->in_extraction_tree = 1;
2229         dentry->extraction_skipped = 0;
2230         return 0;
2231 }
2232
2233 static int
2234 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
2235               const tchar *target, int extract_flags,
2236               wimlib_progress_func_t progress_func)
2237 {
2238         struct wim_features required_features;
2239         struct apply_ctx ctx;
2240         int ret;
2241         struct wim_lookup_table_entry *lte;
2242
2243         /* Start initializing the apply_ctx.  */
2244         memset(&ctx, 0, sizeof(struct apply_ctx));
2245         ctx.wim = wim;
2246         ctx.extract_flags = extract_flags;
2247         ctx.target = target;
2248         ctx.target_nchars = tstrlen(target);
2249         ctx.progress_func = progress_func;
2250         if (progress_func) {
2251                 ctx.progress.extract.wimfile_name = wim->filename;
2252                 ctx.progress.extract.image = wim->current_image;
2253                 ctx.progress.extract.extract_flags = (extract_flags &
2254                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
2255                 ctx.progress.extract.image_name = wimlib_get_image_name(wim,
2256                                                                         wim->current_image);
2257                 ctx.progress.extract.target = target;
2258         }
2259         INIT_LIST_HEAD(&ctx.stream_list);
2260
2261         if (extract_flags & WIMLIB_EXTRACT_FLAG_FILEMODE) {
2262                 /* File mode --- target is explicit.  */
2263                 wimlib_assert(num_trees == 1);
2264                 ret = calculate_dentry_full_path(trees[0]);
2265                 if (ret)
2266                         return ret;
2267                 ctx.progress.extract.extract_root_wim_source_path = trees[0]->_full_path;
2268                 ctx.extract_root = trees[0];
2269                 for_dentry_in_tree(ctx.extract_root, dentry_set_not_skipped, NULL);
2270         } else {
2271                 /* Targets are to be set relative to the root of the image
2272                  * (preserving original directory structure).  */
2273
2274                 ctx.progress.extract.extract_root_wim_source_path = T("");
2275                 ctx.extract_root = wim_root_dentry(wim);
2276                 for_dentry_in_tree(ctx.extract_root, dentry_set_skipped, NULL);
2277
2278                 for (size_t i = 0; i < num_trees; i++) {
2279                         struct wim_dentry *d;
2280
2281                         for_dentry_in_tree(trees[i], dentry_set_not_skipped, NULL);
2282                         d = trees[i];
2283
2284                         /* Extract directories up to image root if preserving
2285                          * directory structure.  */
2286                         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) {
2287                                 while (d != ctx.extract_root) {
2288                                         d = d->parent;
2289                                         dentry_set_not_skipped(d, NULL);
2290                                 }
2291                         }
2292                 }
2293         }
2294
2295         /* Select the appropriate apply_operations based on the
2296          * platform and extract_flags.  */
2297 #ifdef __WIN32__
2298         ctx.ops = &win32_apply_ops;
2299 #else
2300         ctx.ops = &unix_apply_ops;
2301 #endif
2302
2303 #ifdef WITH_NTFS_3G
2304         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
2305                 ctx.ops = &ntfs_3g_apply_ops;
2306 #endif
2307
2308         /* Call the start_extract() callback.  This gives the apply_operations
2309          * implementation a chance to do any setup needed to access the volume.
2310          * Furthermore, it's expected to set the supported features of this
2311          * extraction mode (ctx.supported_features), which are determined at
2312          * runtime as they may vary depending on the actual volume.  These
2313          * features are then compared with the actual features extracting this
2314          * dentry tree requires.  Some mismatches will merely produce warnings
2315          * and the unsupported data will be ignored; others will produce errors.
2316          */
2317         ret = ctx.ops->start_extract(target, &ctx);
2318         if (ret)
2319                 goto out_dentry_reset_needs_extraction;
2320
2321         /* Get and check the features required to extract the dentry tree.  */
2322         dentry_tree_get_features(ctx.extract_root, &required_features);
2323         ret = do_feature_check(&required_features, &ctx.supported_features,
2324                                extract_flags, ctx.ops, true);
2325         if (ret)
2326                 goto out_finish_or_abort_extract;
2327
2328         ctx.supported_attributes_mask =
2329                 compute_supported_attributes_mask(&ctx.supported_features);
2330
2331         /* Figure out whether the root dentry is being extracted to the root of
2332          * a volume and therefore needs to be treated "specially", for example
2333          * not being explicitly created and not having attributes set.  */
2334         if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special)
2335                 ctx.root_dentry_is_special = ctx.ops->target_is_root(target);
2336
2337         /* Calculate the actual filename component of each extracted dentry.  In
2338          * the process, set the dentry->extraction_skipped flag on dentries that
2339          * are being skipped because of filename or supported features problems.  */
2340         ret = for_dentry_in_tree(ctx.extract_root,
2341                                  dentry_calculate_extraction_path, &ctx);
2342         if (ret)
2343                 goto out_dentry_reset_needs_extraction;
2344
2345         /* Build the list of the streams that need to be extracted and
2346          * initialize ctx.progress.extract with stream information.  */
2347         ret = for_dentry_in_tree(ctx.extract_root,
2348                                  dentry_resolve_and_zero_lte_refcnt, &ctx);
2349         if (ret)
2350                 goto out_dentry_reset_needs_extraction;
2351
2352         ret = for_dentry_in_tree(ctx.extract_root,
2353                                  dentry_add_streams_to_extract, &ctx);
2354         if (ret)
2355                 goto out_teardown_stream_list;
2356
2357         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2358                 /* When extracting from a pipe, the number of bytes of data to
2359                  * extract can't be determined in the normal way (examining the
2360                  * lookup table), since at this point all we have is a set of
2361                  * SHA1 message digests of streams that need to be extracted.
2362                  * However, we can get a reasonably accurate estimate by taking
2363                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
2364                  * data.  This does assume that a full image is being extracted,
2365                  * but currently there is no API for doing otherwise.  (Also,
2366                  * subtract <HARDLINKBYTES> from this if hard links are
2367                  * supported by the extraction mode.)  */
2368                 ctx.progress.extract.total_bytes =
2369                         wim_info_get_image_total_bytes(wim->wim_info,
2370                                                        wim->current_image);
2371                 if (ctx.supported_features.hard_links) {
2372                         ctx.progress.extract.total_bytes -=
2373                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
2374                                                                    wim->current_image);
2375                 }
2376         }
2377
2378         /* Handle the special case of extracting a file to standard
2379          * output.  In that case, "root" should be a single file, not a
2380          * directory tree.  (If not, extract_dentry_to_stdout() will
2381          * return an error.)  */
2382         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
2383                 ret = 0;
2384                 for (size_t i = 0; i < num_trees; i++) {
2385                         ret = extract_dentry_to_stdout(trees[i]);
2386                         if (ret)
2387                                 break;
2388                 }
2389                 goto out_teardown_stream_list;
2390         }
2391
2392         if (ctx.ops->realpath_works_on_nonexisting_files &&
2393             ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) ||
2394              ctx.ops->requires_realtarget_in_paths))
2395         {
2396                 ctx.realtarget = realpath(target, NULL);
2397                 if (!ctx.realtarget) {
2398                         ret = WIMLIB_ERR_NOMEM;
2399                         goto out_teardown_stream_list;
2400                 }
2401                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2402         }
2403
2404         if (progress_func) {
2405                 int msg;
2406                 if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE)
2407                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
2408                 else
2409                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN;
2410                 progress_func(msg, &ctx.progress);
2411         }
2412
2413         if (!ctx.root_dentry_is_special)
2414         {
2415                 tchar path[ctx.ops->path_max];
2416                 if (build_extraction_path(path, ctx.extract_root, &ctx))
2417                 {
2418                         ret = extract_inode(path, &ctx, ctx.extract_root->d_inode);
2419                         if (ret)
2420                                 goto out_free_realtarget;
2421                 }
2422         }
2423
2424         /* If we need to fix up the targets of absolute symbolic links
2425          * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to
2426          * be absolute, use realpath() (or its replacement on Windows) to get
2427          * the absolute path to the extraction target.  Note that this requires
2428          * the target directory to exist, unless
2429          * realpath_works_on_nonexisting_files is set in the apply_operations.
2430          * */
2431         if (!ctx.realtarget &&
2432             (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
2433               required_features.symlink_reparse_points) ||
2434              ctx.ops->requires_realtarget_in_paths))
2435         {
2436                 ctx.realtarget = realpath(target, NULL);
2437                 if (!ctx.realtarget) {
2438                         ret = WIMLIB_ERR_NOMEM;
2439                         goto out_free_realtarget;
2440                 }
2441                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2442         }
2443
2444         if (ctx.ops->requires_short_name_reordering) {
2445                 ret = for_dentry_in_tree(ctx.extract_root, dentry_extract_dir_skeleton,
2446                                          &ctx);
2447                 if (ret)
2448                         goto out_free_realtarget;
2449         }
2450
2451         /* Finally, the important part: extract the tree of files.  */
2452         if (!(extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) {
2453                 /* Sequential extraction requested, so two passes are needed
2454                  * (one for directory structure, one for streams.)  */
2455                 if (progress_func)
2456                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2457                                       &ctx.progress);
2458
2459                 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) {
2460                         ret = for_dentry_in_tree(ctx.extract_root, dentry_extract_skeleton, &ctx);
2461                         if (ret)
2462                                 goto out_free_realtarget;
2463                 }
2464                 if (progress_func)
2465                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2466                                       &ctx.progress);
2467                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
2468                         ret = extract_streams_from_pipe(&ctx);
2469                 else
2470                         ret = extract_stream_list(&ctx);
2471                 if (ret)
2472                         goto out_free_realtarget;
2473         } else {
2474                 /* Sequential extraction was not requested, so we can make do
2475                  * with one pass where we both create the files and extract
2476                  * streams.   */
2477                 if (progress_func)
2478                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2479                                       &ctx.progress);
2480                 ret = for_dentry_in_tree(ctx.extract_root, dentry_extract, &ctx);
2481                 if (ret)
2482                         goto out_free_realtarget;
2483                 if (progress_func)
2484                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2485                                       &ctx.progress);
2486         }
2487
2488         /* If the total number of bytes to extract was miscalculated, just jump
2489          * to the calculated number in order to avoid confusing the progress
2490          * function.  This should only occur when extracting from a pipe.  */
2491         if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes)
2492         {
2493                 DEBUG("Calculated %"PRIu64" bytes to extract, but actually "
2494                       "extracted %"PRIu64,
2495                       ctx.progress.extract.total_bytes,
2496                       ctx.progress.extract.completed_bytes);
2497         }
2498         if (progress_func &&
2499             ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes)
2500         {
2501                 ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes;
2502                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress);
2503         }
2504
2505         /* Apply security descriptors and timestamps.  This is done at the end,
2506          * and in a depth-first manner, to prevent timestamps from getting
2507          * changed by subsequent extract operations and to minimize the chance
2508          * of the restored security descriptors getting in our way.  */
2509         if (progress_func)
2510                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
2511                               &ctx.progress);
2512         ret = for_dentry_in_tree_depth(ctx.extract_root, dentry_extract_final, &ctx);
2513         if (ret)
2514                 goto out_free_realtarget;
2515
2516         if (progress_func) {
2517                 int msg;
2518                 if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE)
2519                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END;
2520                 else
2521                         msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END;
2522                 progress_func(msg, &ctx.progress);
2523         }
2524
2525         do_extract_warnings(&ctx);
2526
2527         ret = 0;
2528 out_free_realtarget:
2529         FREE(ctx.realtarget);
2530 out_teardown_stream_list:
2531         /* Free memory allocated as part of the mapping from each
2532          * wim_lookup_table_entry to the dentries that reference it.  */
2533         if (!(ctx.extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2534                 list_for_each_entry(lte, &ctx.stream_list, extraction_list)
2535                         if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries))
2536                                 FREE(lte->lte_dentries);
2537 out_finish_or_abort_extract:
2538         if (ret) {
2539                 if (ctx.ops->abort_extract)
2540                         ctx.ops->abort_extract(&ctx);
2541         } else {
2542                 if (ctx.ops->finish_extract)
2543                         ret = ctx.ops->finish_extract(&ctx);
2544         }
2545 out_dentry_reset_needs_extraction:
2546         for_dentry_in_tree(ctx.extract_root, dentry_reset_needs_extraction, NULL);
2547         return ret;
2548 }
2549
2550 /* Make sure the extraction flags make sense, and update them if needed.  */
2551 static int
2552 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
2553 {
2554         int extract_flags = *extract_flags_p;
2555
2556         /* Check for invalid flag combinations  */
2557         if ((extract_flags &
2558              (WIMLIB_EXTRACT_FLAG_SYMLINK |
2559               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
2560                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
2561                 return WIMLIB_ERR_INVALID_PARAM;
2562
2563         if ((extract_flags &
2564              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2565               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2566                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
2567                 return WIMLIB_ERR_INVALID_PARAM;
2568
2569         if ((extract_flags &
2570              (WIMLIB_EXTRACT_FLAG_RPFIX |
2571               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
2572                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
2573                 return WIMLIB_ERR_INVALID_PARAM;
2574
2575         if ((extract_flags &
2576              (WIMLIB_EXTRACT_FLAG_RESUME |
2577               WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME)
2578                 return WIMLIB_ERR_INVALID_PARAM;
2579
2580 #ifndef WITH_NTFS_3G
2581         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2582                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
2583                       "        we cannot apply a WIM image directly to a NTFS volume.");
2584                 return WIMLIB_ERR_UNSUPPORTED;
2585         }
2586 #endif
2587
2588         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
2589                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
2590         {
2591                 /* Do reparse point fixups by default if the WIM header says
2592                  * they are enabled.  */
2593                 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
2594                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
2595         }
2596
2597         /* TODO: Since UNIX data entries are stored in the file resources, in a
2598          * completely sequential extraction they may come up before the
2599          * corresponding file or symbolic link data.  This needs to be handled
2600          * better.  */
2601         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2602                               WIMLIB_EXTRACT_FLAG_FILE_ORDER))
2603                                     == WIMLIB_EXTRACT_FLAG_UNIX_DATA)
2604         {
2605                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2606                         WARNING("Setting UNIX file/owner group may "
2607                                 "be impossible on some\n"
2608                                 "          symbolic links "
2609                                 "when applying from a pipe.");
2610                 } else {
2611                         extract_flags |= WIMLIB_EXTRACT_FLAG_FILE_ORDER;
2612                         WARNING("Disabling sequential extraction for "
2613                                 "UNIX data mode");
2614                 }
2615         }
2616
2617         *extract_flags_p = extract_flags;
2618         return 0;
2619 }
2620
2621 static u32
2622 get_wildcard_flags(int extract_flags)
2623 {
2624         u32 wildcard_flags = 0;
2625
2626         if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
2627                 wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
2628         else
2629                 wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
2630
2631         if (default_ignore_case)
2632                 wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
2633
2634         return wildcard_flags;
2635 }
2636
2637 struct append_dentry_ctx {
2638         struct wim_dentry **dentries;
2639         size_t num_dentries;
2640         size_t num_alloc_dentries;
2641 };
2642
2643 static int
2644 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
2645 {
2646         struct append_dentry_ctx *ctx = _ctx;
2647
2648         if (ctx->num_dentries == ctx->num_alloc_dentries) {
2649                 struct wim_dentry **new_dentries;
2650                 size_t new_length;
2651
2652                 new_length = max(ctx->num_alloc_dentries + 8,
2653                                  ctx->num_alloc_dentries * 3 / 2);
2654                 new_dentries = REALLOC(ctx->dentries,
2655                                        new_length * sizeof(ctx->dentries[0]));
2656                 if (new_dentries == NULL)
2657                         return WIMLIB_ERR_NOMEM;
2658                 ctx->dentries = new_dentries;
2659                 ctx->num_alloc_dentries = new_length;
2660         }
2661         ctx->dentries[ctx->num_dentries++] = dentry;
2662         return 0;
2663 }
2664
2665 static int
2666 do_wimlib_extract_paths(WIMStruct *wim,
2667                         int image,
2668                         const tchar *target,
2669                         const tchar * const *paths,
2670                         size_t num_paths,
2671                         int extract_flags,
2672                         wimlib_progress_func_t progress_func)
2673 {
2674         int ret;
2675         struct wim_dentry **trees;
2676         size_t num_trees;
2677
2678         if (wim == NULL || target == NULL || target[0] == T('\0') ||
2679             (num_paths != 0 && paths == NULL))
2680                 return WIMLIB_ERR_INVALID_PARAM;
2681
2682         ret = check_extract_flags(wim, &extract_flags);
2683         if (ret)
2684                 return ret;
2685
2686         ret = select_wim_image(wim, image);
2687         if (ret)
2688                 return ret;
2689
2690         ret = wim_checksum_unhashed_streams(wim);
2691         if (ret)
2692                 return ret;
2693
2694         if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
2695
2696                 struct append_dentry_ctx append_dentry_ctx = {
2697                         .dentries = NULL,
2698                         .num_dentries = 0,
2699                         .num_alloc_dentries = 0,
2700                 };
2701
2702                 u32 wildcard_flags = get_wildcard_flags(extract_flags);
2703
2704                 for (size_t i = 0; i < num_paths; i++) {
2705                         tchar *path = canonicalize_wim_path(paths[i]);
2706                         if (path == NULL) {
2707                                 ret = WIMLIB_ERR_NOMEM;
2708                                 trees = append_dentry_ctx.dentries;
2709                                 goto out_free_trees;
2710                         }
2711                         ret = expand_wildcard(wim, path,
2712                                               append_dentry_cb,
2713                                               &append_dentry_ctx,
2714                                               wildcard_flags);
2715                         FREE(path);
2716                         if (ret) {
2717                                 trees = append_dentry_ctx.dentries;
2718                                 goto out_free_trees;
2719                         }
2720                 }
2721                 trees = append_dentry_ctx.dentries;
2722                 num_trees = append_dentry_ctx.num_dentries;
2723         } else {
2724                 trees = MALLOC(num_paths * sizeof(trees[0]));
2725                 if (trees == NULL)
2726                         return WIMLIB_ERR_NOMEM;
2727
2728                 for (size_t i = 0; i < num_paths; i++) {
2729
2730                         tchar *path = canonicalize_wim_path(paths[i]);
2731                         if (path == NULL) {
2732                                 ret = WIMLIB_ERR_NOMEM;
2733                                 goto out_free_trees;
2734                         }
2735
2736                         trees[i] = get_dentry(wim, path,
2737                                               WIMLIB_CASE_PLATFORM_DEFAULT);
2738                         FREE(path);
2739                         if (trees[i] == NULL) {
2740                                   ERROR("Path \"%"TS"\" does not exist "
2741                                         "in WIM image %d",
2742                                         paths[i], wim->current_image);
2743                                   ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
2744                                   goto out_free_trees;
2745                         }
2746                 }
2747                 num_trees = num_paths;
2748         }
2749
2750         if (num_trees == 0) {
2751                 ret = 0;
2752                 goto out_free_trees;
2753         }
2754
2755         ret = extract_trees(wim, trees, num_trees,
2756                             target, extract_flags, progress_func);
2757 out_free_trees:
2758         FREE(trees);
2759         return ret;
2760 }
2761
2762 static int
2763 extract_single_image(WIMStruct *wim, int image,
2764                      const tchar *target, int extract_flags,
2765                      wimlib_progress_func_t progress_func)
2766 {
2767         const tchar *path = T("");
2768         return do_wimlib_extract_paths(wim, image, target, &path, 1,
2769                                        extract_flags | WIMLIB_EXTRACT_FLAG_IMAGEMODE,
2770                                        progress_func);
2771 }
2772
2773 static const tchar * const filename_forbidden_chars =
2774 T(
2775 #ifdef __WIN32__
2776 "<>:\"/\\|?*"
2777 #else
2778 "/"
2779 #endif
2780 );
2781
2782 /* This function checks if it is okay to use a WIM image's name as a directory
2783  * name.  */
2784 static bool
2785 image_name_ok_as_dir(const tchar *image_name)
2786 {
2787         return image_name && *image_name &&
2788                 !tstrpbrk(image_name, filename_forbidden_chars) &&
2789                 tstrcmp(image_name, T(".")) &&
2790                 tstrcmp(image_name, T(".."));
2791 }
2792
2793 /* Extracts all images from the WIM to the directory @target, with the images
2794  * placed in subdirectories named by their image names. */
2795 static int
2796 extract_all_images(WIMStruct *wim,
2797                    const tchar *target,
2798                    int extract_flags,
2799                    wimlib_progress_func_t progress_func)
2800 {
2801         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
2802         size_t output_path_len = tstrlen(target);
2803         tchar buf[output_path_len + 1 + image_name_max_len + 1];
2804         int ret;
2805         int image;
2806         const tchar *image_name;
2807         struct stat stbuf;
2808
2809         extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
2810
2811         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2812                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
2813                 return WIMLIB_ERR_INVALID_PARAM;
2814         }
2815
2816         if (tstat(target, &stbuf)) {
2817                 if (errno == ENOENT) {
2818                         if (tmkdir(target, 0755)) {
2819                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
2820                                 return WIMLIB_ERR_MKDIR;
2821                         }
2822                 } else {
2823                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
2824                         return WIMLIB_ERR_STAT;
2825                 }
2826         } else if (!S_ISDIR(stbuf.st_mode)) {
2827                 ERROR("\"%"TS"\" is not a directory", target);
2828                 return WIMLIB_ERR_NOTDIR;
2829         }
2830
2831         tmemcpy(buf, target, output_path_len);
2832         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
2833         for (image = 1; image <= wim->hdr.image_count; image++) {
2834                 image_name = wimlib_get_image_name(wim, image);
2835                 if (image_name_ok_as_dir(image_name)) {
2836                         tstrcpy(buf + output_path_len + 1, image_name);
2837                 } else {
2838                         /* Image name is empty or contains forbidden characters.
2839                          * Use image number instead. */
2840                         tsprintf(buf + output_path_len + 1, T("%d"), image);
2841                 }
2842                 ret = extract_single_image(wim, image, buf, extract_flags,
2843                                            progress_func);
2844                 if (ret)
2845                         return ret;
2846         }
2847         return 0;
2848 }
2849
2850 static void
2851 clear_lte_extracted_file(WIMStruct *wim, int extract_flags)
2852 {
2853         if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2854                                       WIMLIB_EXTRACT_FLAG_HARDLINK)))
2855                 for_lookup_table_entry(wim->lookup_table,
2856                                        lte_free_extracted_file, NULL);
2857 }
2858
2859 static int
2860 do_wimlib_extract_image(WIMStruct *wim,
2861                         int image,
2862                         const tchar *target,
2863                         int extract_flags,
2864                         wimlib_progress_func_t progress_func)
2865 {
2866         int ret;
2867
2868         if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
2869                              WIMLIB_EXTRACT_FLAG_TO_STDOUT))
2870                 return WIMLIB_ERR_INVALID_PARAM;
2871
2872         if (image == WIMLIB_ALL_IMAGES)
2873                 ret = extract_all_images(wim, target, extract_flags,
2874                                          progress_func);
2875         else
2876                 ret = extract_single_image(wim, image, target, extract_flags,
2877                                            progress_func);
2878
2879         clear_lte_extracted_file(wim, extract_flags);
2880         return ret;
2881 }
2882
2883
2884 /****************************************************************************
2885  *                          Extraction API                                  *
2886  ****************************************************************************/
2887
2888 /* Note: new code should use wimlib_extract_paths() instead of
2889  * wimlib_extract_files() if possible.  */
2890 WIMLIBAPI int
2891 wimlib_extract_files(WIMStruct *wim,
2892                      int image,
2893                      const struct wimlib_extract_command *cmds,
2894                      size_t num_cmds,
2895                      int default_extract_flags,
2896                      wimlib_progress_func_t progress_func)
2897 {
2898         int all_flags = 0;
2899         int link_flags;
2900         int ret;
2901
2902         if (num_cmds == 0)
2903                 return 0;
2904
2905         default_extract_flags |= WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE;
2906
2907         for (size_t i = 0; i < num_cmds; i++) {
2908                 int cmd_flags = (cmds[i].extract_flags |
2909                                  default_extract_flags);
2910
2911                 if (cmd_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
2912                         return WIMLIB_ERR_INVALID_PARAM;
2913
2914                 int cmd_link_flags = (cmd_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2915                                                    WIMLIB_EXTRACT_FLAG_HARDLINK));
2916                 if (i == 0) {
2917                         link_flags = cmd_link_flags;
2918                 } else {
2919                         if (cmd_link_flags != link_flags) {
2920                                 ERROR("The same symlink or hardlink extraction mode "
2921                                       "must be set on all extraction commands!");
2922                                 return WIMLIB_ERR_INVALID_PARAM;
2923                         }
2924                 }
2925                 all_flags |= cmd_flags;
2926         }
2927         if (all_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
2928                 ERROR("Glob paths not supported for wimlib_extract_files(). "
2929                       "Use wimlib_extract_paths() instead.");
2930                 return WIMLIB_ERR_INVALID_PARAM;
2931         }
2932
2933         for (size_t i = 0; i < num_cmds; i++) {
2934                 int extract_flags = (cmds[i].extract_flags |
2935                                      default_extract_flags);
2936                 const tchar *target = cmds[i].fs_dest_path;
2937                 const tchar *wim_source_path = cmds[i].wim_source_path;
2938
2939                 ret = do_wimlib_extract_paths(wim, image, target,
2940                                               &wim_source_path, 1,
2941                                               extract_flags | WIMLIB_EXTRACT_FLAG_FILEMODE,
2942                                               progress_func);
2943                 if (ret)
2944                         break;
2945         }
2946
2947         clear_lte_extracted_file(wim, all_flags);
2948         return ret;
2949 }
2950
2951 WIMLIBAPI int
2952 wimlib_extract_paths(WIMStruct *wim,
2953                      int image,
2954                      const tchar *target,
2955                      const tchar * const *paths,
2956                      size_t num_paths,
2957                      int extract_flags,
2958                      wimlib_progress_func_t progress_func)
2959 {
2960         int ret;
2961
2962         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
2963                 return WIMLIB_ERR_INVALID_PARAM;
2964
2965         ret = do_wimlib_extract_paths(wim, image, target, paths, num_paths,
2966                                       extract_flags, progress_func);
2967         clear_lte_extracted_file(wim, extract_flags);
2968         return ret;
2969 }
2970
2971 WIMLIBAPI int
2972 wimlib_extract_pathlist(WIMStruct *wim, int image,
2973                         const tchar *target,
2974                         const tchar *path_list_file,
2975                         int extract_flags,
2976                         wimlib_progress_func_t progress_func)
2977 {
2978         int ret;
2979         tchar **paths;
2980         size_t num_paths;
2981         void *mem;
2982
2983         ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
2984         if (ret) {
2985                 ERROR("Failed to read path list file \"%"TS"\"",
2986                       path_list_file);
2987                 return ret;
2988         }
2989
2990         ret = wimlib_extract_paths(wim, image, target,
2991                                    (const tchar * const *)paths, num_paths,
2992                                    extract_flags, progress_func);
2993         FREE(paths);
2994         FREE(mem);
2995         return ret;
2996 }
2997
2998 WIMLIBAPI int
2999 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
3000                                const tchar *target, int extract_flags,
3001                                wimlib_progress_func_t progress_func)
3002 {
3003         int ret;
3004         WIMStruct *pwm;
3005         struct filedes *in_fd;
3006         int image;
3007         unsigned i;
3008
3009         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
3010                 return WIMLIB_ERR_INVALID_PARAM;
3011
3012         if (extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)
3013                 return WIMLIB_ERR_INVALID_PARAM;
3014
3015         /* Read the WIM header from the pipe and get a WIMStruct to represent
3016          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
3017          * wimlib_open_wim(), getting a WIMStruct in this way will result in
3018          * an empty lookup table, no XML data read, and no filename set.  */
3019         ret = open_wim_as_WIMStruct(&pipe_fd,
3020                                     WIMLIB_OPEN_FLAG_FROM_PIPE,
3021                                     &pwm, progress_func);
3022         if (ret)
3023                 return ret;
3024
3025         /* Sanity check to make sure this is a pipable WIM.  */
3026         if (pwm->hdr.magic != PWM_MAGIC) {
3027                 ERROR("The WIM being read from file descriptor %d "
3028                       "is not pipable!", pipe_fd);
3029                 ret = WIMLIB_ERR_NOT_PIPABLE;
3030                 goto out_wimlib_free;
3031         }
3032
3033         /* Sanity check to make sure the first part of a pipable split WIM is
3034          * sent over the pipe first.  */
3035         if (pwm->hdr.part_number != 1) {
3036                 ERROR("The first part of the split WIM must be "
3037                       "sent over the pipe first.");
3038                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3039                 goto out_wimlib_free;
3040         }
3041
3042         in_fd = &pwm->in_fd;
3043         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
3044
3045         /* As mentioned, the WIMStruct we created from the pipe does not have
3046          * XML data yet.  Fix this by reading the extra copy of the XML data
3047          * that directly follows the header in pipable WIMs.  (Note: see
3048          * write_pipable_wim() for more details about the format of pipable
3049          * WIMs.)  */
3050         {
3051                 struct wim_lookup_table_entry xml_lte;
3052                 struct wim_resource_spec xml_rspec;
3053                 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
3054                 if (ret)
3055                         goto out_wimlib_free;
3056
3057                 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
3058                 {
3059                         ERROR("Expected XML data, but found non-metadata "
3060                               "stream.");
3061                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3062                         goto out_wimlib_free;
3063                 }
3064
3065                 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
3066
3067                 ret = read_wim_xml_data(pwm);
3068                 if (ret)
3069                         goto out_wimlib_free;
3070
3071                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
3072                         ERROR("Image count in XML data is not the same as in WIM header.");
3073                         ret = WIMLIB_ERR_XML;
3074                         goto out_wimlib_free;
3075                 }
3076         }
3077
3078         /* Get image index (this may use the XML data that was just read to
3079          * resolve an image name).  */
3080         if (image_num_or_name) {
3081                 image = wimlib_resolve_image(pwm, image_num_or_name);
3082                 if (image == WIMLIB_NO_IMAGE) {
3083                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
3084                               image_num_or_name);
3085                         ret = WIMLIB_ERR_INVALID_IMAGE;
3086                         goto out_wimlib_free;
3087                 } else if (image == WIMLIB_ALL_IMAGES) {
3088                         ERROR("Applying all images from a pipe is not supported.");
3089                         ret = WIMLIB_ERR_INVALID_IMAGE;
3090                         goto out_wimlib_free;
3091                 }
3092         } else {
3093                 if (pwm->hdr.image_count != 1) {
3094                         ERROR("No image was specified, but the pipable WIM "
3095                               "did not contain exactly 1 image");
3096                         ret = WIMLIB_ERR_INVALID_IMAGE;
3097                         goto out_wimlib_free;
3098                 }
3099                 image = 1;
3100         }
3101
3102         /* Load the needed metadata resource.  */
3103         for (i = 1; i <= pwm->hdr.image_count; i++) {
3104                 struct wim_lookup_table_entry *metadata_lte;
3105                 struct wim_image_metadata *imd;
3106                 struct wim_resource_spec *metadata_rspec;
3107
3108                 metadata_lte = new_lookup_table_entry();
3109                 if (metadata_lte == NULL) {
3110                         ret = WIMLIB_ERR_NOMEM;
3111                         goto out_wimlib_free;
3112                 }
3113                 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
3114                 if (metadata_rspec == NULL) {
3115                         ret = WIMLIB_ERR_NOMEM;
3116                         free_lookup_table_entry(metadata_lte);
3117                         goto out_wimlib_free;
3118                 }
3119
3120                 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
3121                 imd = pwm->image_metadata[i - 1];
3122                 imd->metadata_lte = metadata_lte;
3123                 if (ret) {
3124                         FREE(metadata_rspec);
3125                         goto out_wimlib_free;
3126                 }
3127
3128                 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
3129                         ERROR("Expected metadata resource, but found "
3130                               "non-metadata stream.");
3131                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
3132                         goto out_wimlib_free;
3133                 }
3134
3135                 if (i == image) {
3136                         /* Metadata resource is for the images being extracted.
3137                          * Parse it and save the metadata in memory.  */
3138                         ret = read_metadata_resource(pwm, imd);
3139                         if (ret)
3140                                 goto out_wimlib_free;
3141                         imd->modified = 1;
3142                 } else {
3143                         /* Metadata resource is not for the image being
3144                          * extracted.  Skip over it.  */
3145                         ret = skip_wim_stream(metadata_lte);
3146                         if (ret)
3147                                 goto out_wimlib_free;
3148                 }
3149         }
3150         /* Extract the image.  */
3151         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
3152         ret = do_wimlib_extract_image(pwm, image, target,
3153                                       extract_flags, progress_func);
3154         /* Clean up and return.  */
3155 out_wimlib_free:
3156         wimlib_free(pwm);
3157         return ret;
3158 }
3159
3160 WIMLIBAPI int
3161 wimlib_extract_image(WIMStruct *wim,
3162                      int image,
3163                      const tchar *target,
3164                      int extract_flags,
3165                      wimlib_progress_func_t progress_func)
3166 {
3167         if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
3168                 return WIMLIB_ERR_INVALID_PARAM;
3169         return do_wimlib_extract_image(wim, image, target, extract_flags,
3170                                        progress_func);
3171 }