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

   1 /*
   2  * extract_image.c
   3  *
   4  * Support for extracting WIM files.
   5  *
   6  * This code does NOT contain any filesystem-specific features.  In particular,
   7  * security information (i.e. file permissions) and alternate data streams are
   8  * ignored, except possibly to read an alternate data stream that contains
   9  * symbolic link data.
  10  */
  11
  12 /*
  13  * Copyright (C) 2012, 2013 Eric Biggers
  14  *
  15  * This file is part of wimlib, a library for working with WIM files.
  16  *
  17  * wimlib is free software; you can redistribute it and/or modify it under the
  18  * terms of the GNU General Public License as published by the Free
  19  * Software Foundation; either version 3 of the License, or (at your option)
  20  * any later version.
  21  *
  22  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  23  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  24  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  25  * details.
  26  *
  27  * You should have received a copy of the GNU General Public License
  28  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  29  */
  30
  31 #include "config.h"
  32
  33 #if defined(__CYGWIN__) || defined(__WIN32__)
  34 #include <windows.h>
  35 #       ifdef ERROR
  36 #               undef ERROR
  37 #       endif
  38 #include <wchar.h>
  39 #endif
  40
  41 #include <dirent.h>
  42 #include <errno.h>
  43 #include <fcntl.h>
  44 #include <string.h>
  45 #include <sys/stat.h>
  46 #include <stdlib.h>
  47 #include <sys/time.h>
  48
  49 #ifdef HAVE_UTIME_H
  50 #include <utime.h>
  51 #endif
  52
  53 #include <unistd.h>
  54
  55 #include "dentry.h"
  56 #include "lookup_table.h"
  57 #include "timestamp.h"
  58 #include "wimlib_internal.h"
  59 #include "xml.h"
  60
  61 #ifdef WITH_NTFS_3G
  62 #include <ntfs-3g/volume.h>
  63 #endif
  64
  65 #ifdef HAVE_ALLOCA_H
  66 #include <alloca.h>
  67 #else
  68 #include <stdlib.h>
  69 #endif
  70
  71 #if defined(__CYGWIN__) || defined(__WIN32__)
  72
  73 static int win32_set_reparse_data(HANDLE h,
  74                                   u32 reparse_tag,
  75                                   const struct wim_lookup_table_entry *lte,
  76                                   const wchar_t *path,
  77                                   const char *path_utf8)
  78 {
  79         int ret;
  80         u8 *buf;
  81         size_t len;
  82
  83         if (!lte) {
  84                 WARNING("\"%s\" is marked as a reparse point but had no reparse data",
  85                         path_utf8);
  86                 return 0;
  87         }
  88         len = wim_resource_size(lte);
  89         if (len > 16 * 1024 - 8) {
  90                 WARNING("\"%s\": reparse data too long!", path_utf8);
  91                 return 0;
  92         }
  93         buf = alloca(len + 8);
  94         ret = read_full_wim_resource(lte, buf + 8, 0);
  95         if (ret)
  96                 return ret;
  97         *(u32*)(buf + 0) = reparse_tag;
  98         *(u16*)(buf + 4) = len;
  99         if (!DeviceIoControl(h, FSCTL_SET_REPARSE_POINT, buf, len + 8,
 100                              NULL, 0, NULL, NULL))
 101         {
 102                 DWORD err = GetLastError();
 103                 ERROR("Failed to set reparse data on \"%s\"",
 104                       path_utf8);
 105                 win32_error(err);
 106                 ret = WIMLIB_ERR_WRITE;
 107         } else
 108                 ret = 0;
 109         return ret;
 110 }
 111
 112
 113 static int win32_extract_chunk(const u8 *buf, size_t len, u64 offset, void *arg)
 114 {
 115         HANDLE hStream = arg;
 116
 117         DWORD nbytes_written;
 118         wimlib_assert(len <= 0xffffffff);
 119
 120         if (!WriteFile(hStream, buf, len, &nbytes_written, NULL) ||
 121             nbytes_written != len)
 122         {
 123                 DWORD err = GetLastError();
 124                 ERROR("WriteFile(): write error");
 125                 win32_error(err);
 126                 return WIMLIB_ERR_WRITE;
 127         }
 128         return 0;
 129 }
 130
 131 static int do_win32_extract_stream(HANDLE hStream, struct wim_lookup_table_entry *lte)
 132 {
 133         return extract_wim_resource(lte, wim_resource_size(lte),
 134                                     win32_extract_chunk, hStream);
 135 }
 136
 137 static int win32_extract_stream(const struct wim_inode *inode,
 138                                 const wchar_t *path,
 139                                 const wchar_t *stream_name_utf16,
 140                                 struct wim_lookup_table_entry *lte,
 141                                 const char *path_utf8)
 142 {
 143         wchar_t *stream_path;
 144         HANDLE h;
 145         int ret;
 146
 147         if (stream_name_utf16) {
 148                 size_t stream_path_nchars;
 149                 size_t path_nchars = wcslen(path);
 150                 size_t stream_name_nchars = wcslen(stream_name_utf16);
 151
 152                 stream_path_nchars = path_nchars + 1 + stream_name_nchars;
 153
 154                 stream_path = alloca((stream_path_nchars + 1) * sizeof(wchar_t));
 155
 156                 memcpy(stream_path, path, path_nchars * sizeof(wchar_t));
 157                 stream_path[path_nchars] = L':';
 158                 memcpy(&stream_path[path_nchars + 1], stream_name_utf16,
 159                        stream_name_nchars * sizeof(wchar_t));
 160                 stream_path[stream_path_nchars] = L'\0';
 161
 162                 /*wsprintf(stream_path, stream_path_nchars, L"%ls:%ls",*/
 163                          /*path, stream_name_utf16);*/
 164         } else {
 165                 stream_path = (wchar_t*)path;
 166         }
 167
 168         h = CreateFileW(stream_path,
 169                         GENERIC_WRITE | WRITE_OWNER | WRITE_DAC,
 170                         0,
 171                         NULL,
 172                         CREATE_ALWAYS,
 173                         FILE_FLAG_OPEN_REPARSE_POINT |
 174                             FILE_FLAG_BACKUP_SEMANTICS |
 175                             inode->i_attributes,
 176                         NULL);
 177         if (!h) {
 178                 win32_error(GetLastError());
 179                 ret = WIMLIB_ERR_OPEN;
 180                 goto fail;
 181         }
 182
 183         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 184                 ret = win32_set_reparse_data(h, inode->i_reparse_tag,
 185                                              lte, path, path_utf8);
 186                 if (ret)
 187                         goto fail_close_handle;
 188         } else {
 189                 if (lte) {
 190                         ret = do_win32_extract_stream(h, lte);
 191                         if (ret)
 192                                 goto fail_close_handle;
 193                 }
 194         }
 195
 196         if (!CloseHandle(h)) {
 197                 win32_error(GetLastError());
 198                 ret = WIMLIB_ERR_WRITE;
 199                 goto fail;
 200         }
 201         ret = 0;
 202         goto out;
 203 fail_close_handle:
 204         CloseHandle(h);
 205 fail:
 206         ERROR("Error extracting %s", path_utf8);
 207 out:
 208         return ret;
 209 }
 210
 211 static int win32_extract_streams(struct wim_inode *inode,
 212                                  const wchar_t *path,
 213                                  const char *path_utf8)
 214 {
 215         struct wim_lookup_table_entry *unnamed_lte;
 216         int ret;
 217
 218         ret = win32_extract_stream(inode, path, NULL, unnamed_lte, path_utf8);
 219         if (ret)
 220                 goto out;
 221
 222         for (u16 i = 0; i < inode->i_num_ads; i++) {
 223                 const struct wim_ads_entry *ads_entry = &inode->i_ads_entries[i];
 224                 if (ads_entry->stream_name_len != 0) {
 225                         ret = win32_extract_stream(inode,
 226                                                    path,
 227                                                    (const wchar_t*)ads_entry->stream_name,
 228                                                    ads_entry->lte,
 229                                                    path_utf8);
 230                         if (ret)
 231                                 goto out;
 232                 }
 233         }
 234         ret = 0;
 235 out:
 236         return ret;
 237 }
 238
 239 static int win32_set_security_data(const struct wim_inode *inode,
 240                                    const wchar_t *path,
 241                                    const struct wim_security_data *sd,
 242                                    const char *path_utf8)
 243 {
 244         SECURITY_INFORMATION securityInformation = DACL_SECURITY_INFORMATION |
 245                                                    SACL_SECURITY_INFORMATION |
 246                                                    OWNER_SECURITY_INFORMATION |
 247                                                    GROUP_SECURITY_INFORMATION;
 248         if (!SetFileSecurityW(path, securityInformation,
 249                               (PSECURITY_DESCRIPTOR)sd->descriptors[inode->i_security_id]))
 250         {
 251                 DWORD err = GetLastError();
 252                 ERROR("Can't set security descriptor on \"%s\"", path_utf8);
 253                 win32_error(err);
 254                 return WIMLIB_ERR_WRITE;
 255         }
 256         return 0;
 257 }
 258
 259 #else
 260 static int extract_regular_file_linked(struct wim_dentry *dentry,
 261                                        const char *output_path,
 262                                        struct apply_args *args,
 263                                        struct wim_lookup_table_entry *lte)
 264 {
 265         /* This mode overrides the normal hard-link extraction and
 266          * instead either symlinks or hardlinks *all* identical files in
 267          * the WIM, even if they are in a different image (in the case
 268          * of a multi-image extraction) */
 269
 270         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
 271                 if (link(lte->extracted_file, output_path) != 0) {
 272                         ERROR_WITH_ERRNO("Failed to hard link "
 273                                          "`%s' to `%s'",
 274                                          output_path, lte->extracted_file);
 275                         return WIMLIB_ERR_LINK;
 276                 }
 277         } else {
 278                 int num_path_components;
 279                 int num_output_dir_path_components;
 280                 size_t extracted_file_len;
 281                 char *p;
 282                 const char *p2;
 283                 size_t i;
 284
 285                 num_path_components =
 286                         get_num_path_components(dentry->full_path_utf8) - 1;
 287                 num_output_dir_path_components =
 288                         get_num_path_components(args->target);
 289
 290                 if (args->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
 291                         num_path_components++;
 292                         num_output_dir_path_components--;
 293                 }
 294                 extracted_file_len = strlen(lte->extracted_file);
 295
 296                 char buf[extracted_file_len + 3 * num_path_components + 1];
 297                 p = &buf[0];
 298
 299                 for (i = 0; i < num_path_components; i++) {
 300                         *p++ = '.';
 301                         *p++ = '.';
 302                         *p++ = '/';
 303                 }
 304                 p2 = lte->extracted_file;
 305                 while (*p2 == '/')
 306                         p2++;
 307                 while (num_output_dir_path_components--)
 308                         p2 = path_next_part(p2, NULL);
 309                 strcpy(p, p2);
 310                 if (symlink(buf, output_path) != 0) {
 311                         ERROR_WITH_ERRNO("Failed to symlink `%s' to "
 312                                          "`%s'",
 313                                          buf, lte->extracted_file);
 314                         return WIMLIB_ERR_LINK;
 315                 }
 316         }
 317         return 0;
 318 }
 319
 320 static int symlink_apply_unix_data(const char *link,
 321                                    const struct wimlib_unix_data *unix_data)
 322 {
 323         if (lchown(link, unix_data->uid, unix_data->gid)) {
 324                 if (errno == EPERM) {
 325                         /* Ignore */
 326                         WARNING_WITH_ERRNO("failed to set symlink UNIX owner/group");
 327                 } else {
 328                         ERROR_WITH_ERRNO("failed to set symlink UNIX owner/group");
 329                         return WIMLIB_ERR_INVALID_DENTRY;
 330                 }
 331         }
 332         return 0;
 333 }
 334
 335 static int fd_apply_unix_data(int fd, const struct wimlib_unix_data *unix_data)
 336 {
 337         if (fchown(fd, unix_data->uid, unix_data->gid)) {
 338                 if (errno == EPERM) {
 339                         WARNING_WITH_ERRNO("failed to set file UNIX owner/group");
 340                         /* Ignore? */
 341                 } else {
 342                         ERROR_WITH_ERRNO("failed to set file UNIX owner/group");
 343                         return WIMLIB_ERR_INVALID_DENTRY;
 344                 }
 345         }
 346
 347         if (fchmod(fd, unix_data->mode)) {
 348                 if (errno == EPERM) {
 349                         WARNING_WITH_ERRNO("failed to set UNIX file mode");
 350                         /* Ignore? */
 351                 } else {
 352                         ERROR_WITH_ERRNO("failed to set UNIX file mode");
 353                         return WIMLIB_ERR_INVALID_DENTRY;
 354                 }
 355         }
 356         return 0;
 357 }
 358
 359 static int dir_apply_unix_data(const char *dir,
 360                                const struct wimlib_unix_data *unix_data)
 361 {
 362         int dfd = open(dir, O_RDONLY);
 363         int ret;
 364         if (dfd >= 0) {
 365                 ret = fd_apply_unix_data(dfd, unix_data);
 366                 if (close(dfd)) {
 367                         ERROR_WITH_ERRNO("can't close directory `%s'", dir);
 368                         ret = WIMLIB_ERR_MKDIR;
 369                 }
 370         } else {
 371                 ERROR_WITH_ERRNO("can't open directory `%s'", dir);
 372                 ret = WIMLIB_ERR_MKDIR;
 373         }
 374         return ret;
 375 }
 376
 377 static int extract_regular_file_unlinked(struct wim_dentry *dentry,
 378                                          struct apply_args *args,
 379                                          const char *output_path,
 380                                          struct wim_lookup_table_entry *lte)
 381 {
 382         /* Normal mode of extraction.  Regular files and hard links are
 383          * extracted in the way that they appear in the WIM. */
 384
 385         int out_fd;
 386         int ret;
 387         struct wim_inode *inode = dentry->d_inode;
 388
 389         if (!((args->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE)
 390                 && (args->extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 391                                      WIMLIB_EXTRACT_FLAG_HARDLINK))))
 392         {
 393                 /* If the dentry is part of a hard link set of at least 2
 394                  * dentries and one of the other dentries has already been
 395                  * extracted, make a hard link to the file corresponding to this
 396                  * already-extracted directory.  Otherwise, extract the file and
 397                  * set the inode->i_extracted_file field so that other dentries
 398                  * in the hard link group can link to it. */
 399                 if (inode->i_nlink > 1) {
 400                         if (inode->i_extracted_file) {
 401                                 DEBUG("Extracting hard link `%s' => `%s'",
 402                                       output_path, inode->i_extracted_file);
 403                                 if (link(inode->i_extracted_file, output_path) != 0) {
 404                                         ERROR_WITH_ERRNO("Failed to hard link "
 405                                                          "`%s' to `%s'",
 406                                                          output_path,
 407                                                          inode->i_extracted_file);
 408                                         return WIMLIB_ERR_LINK;
 409                                 }
 410                                 return 0;
 411                         }
 412                         FREE(inode->i_extracted_file);
 413                         inode->i_extracted_file = STRDUP(output_path);
 414                         if (!inode->i_extracted_file) {
 415                                 ERROR("Failed to allocate memory for filename");
 416                                 return WIMLIB_ERR_NOMEM;
 417                         }
 418                 }
 419         }
 420
 421         /* Extract the contents of the file to @output_path. */
 422
 423         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
 424         if (out_fd == -1) {
 425                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
 426                                  output_path);
 427                 return WIMLIB_ERR_OPEN;
 428         }
 429
 430         if (!lte) {
 431                 /* Empty file with no lookup table entry */
 432                 DEBUG("Empty file `%s'.", output_path);
 433                 ret = 0;
 434                 goto out_extract_unix_data;
 435         }
 436
 437         ret = extract_wim_resource_to_fd(lte, out_fd, wim_resource_size(lte));
 438         if (ret != 0) {
 439                 ERROR("Failed to extract resource to `%s'", output_path);
 440                 goto out;
 441         }
 442
 443 out_extract_unix_data:
 444         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 445                 struct wimlib_unix_data unix_data;
 446                 ret = inode_get_unix_data(inode, &unix_data, NULL);
 447                 if (ret > 0)
 448                         ;
 449                 else if (ret < 0)
 450                         ret = 0;
 451                 else
 452                         ret = fd_apply_unix_data(out_fd, &unix_data);
 453                 if (ret != 0)
 454                         goto out;
 455         }
 456         if (lte)
 457                 args->progress.extract.completed_bytes += wim_resource_size(lte);
 458 out:
 459         if (close(out_fd) != 0) {
 460                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
 461                 if (ret == 0)
 462                         ret = WIMLIB_ERR_WRITE;
 463         }
 464         return ret;
 465 }
 466
 467 static int extract_regular_file(struct wim_dentry *dentry,
 468                                 struct apply_args *args,
 469                                 const char *output_path)
 470 {
 471         struct wim_lookup_table_entry *lte;
 472         const struct wim_inode *inode = dentry->d_inode;
 473
 474         lte = inode_unnamed_lte_resolved(inode);
 475
 476         if (lte && (args->extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 477                                            WIMLIB_EXTRACT_FLAG_HARDLINK)))
 478         {
 479                 if (lte->extracted_file) {
 480                         return extract_regular_file_linked(dentry, output_path, args, lte);
 481                 } else {
 482                         lte->extracted_file = STRDUP(output_path);
 483                         if (!lte->extracted_file)
 484                                 return WIMLIB_ERR_NOMEM;
 485                 }
 486         }
 487         return extract_regular_file_unlinked(dentry, args, output_path, lte);
 488 }
 489
 490 static int extract_symlink(struct wim_dentry *dentry,
 491                            struct apply_args *args,
 492                            const char *output_path)
 493 {
 494         char target[4096];
 495         ssize_t ret = inode_readlink(dentry->d_inode, target,
 496                                      sizeof(target), args->w, 0);
 497         struct wim_lookup_table_entry *lte;
 498
 499         if (ret <= 0) {
 500                 ERROR("Could not read the symbolic link from dentry `%s'",
 501                       dentry->full_path_utf8);
 502                 return WIMLIB_ERR_INVALID_DENTRY;
 503         }
 504         ret = symlink(target, output_path);
 505         if (ret != 0) {
 506                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 507                                  output_path, target);
 508                 return WIMLIB_ERR_LINK;
 509         }
 510         lte = inode_unnamed_lte_resolved(dentry->d_inode);
 511         wimlib_assert(lte != NULL);
 512         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
 513                 struct wimlib_unix_data unix_data;
 514                 ret = inode_get_unix_data(dentry->d_inode, &unix_data, NULL);
 515                 if (ret > 0)
 516                         ;
 517                 else if (ret < 0)
 518                         ret = 0;
 519                 else
 520                         ret = symlink_apply_unix_data(output_path, &unix_data);
 521                 if (ret != 0)
 522                         return ret;
 523         }
 524         args->progress.extract.completed_bytes += wim_resource_size(lte);
 525         return 0;
 526 }
 527
 528 #endif /* !__CYGWIN__ && !__WIN32__ */
 529
 530 static int extract_directory(struct wim_dentry *dentry,
 531                              const char *output_path, bool is_root)
 532 {
 533         int ret;
 534         struct stat stbuf;
 535
 536         ret = stat(output_path, &stbuf);
 537         if (ret == 0) {
 538                 if (S_ISDIR(stbuf.st_mode)) {
 539                         /*if (!is_root)*/
 540                                 /*WARNING("`%s' already exists", output_path);*/
 541                         goto dir_exists;
 542                 } else {
 543                         ERROR("`%s' is not a directory", output_path);
 544                         return WIMLIB_ERR_MKDIR;
 545                 }
 546         } else {
 547                 if (errno != ENOENT) {
 548                         ERROR_WITH_ERRNO("Failed to stat `%s'", output_path);
 549                         return WIMLIB_ERR_STAT;
 550                 }
 551         }
 552         if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP |
 553                                S_IROTH | S_IXOTH) != 0) {
 554                 ERROR_WITH_ERRNO("Cannot create directory `%s'",
 555                                  output_path);
 556                 return WIMLIB_ERR_MKDIR;
 557         }
 558 dir_exists:
 559         if (dentry) {
 560                 struct wimlib_unix_data unix_data;
 561                 ret = inode_get_unix_data(dentry->d_inode, &unix_data, NULL);
 562                 if (ret > 0)
 563                         ;
 564                 else if (ret < 0)
 565                         ret = 0;
 566                 else
 567                         ret = dir_apply_unix_data(output_path, &unix_data);
 568         } else {
 569                 ret = 0;
 570         }
 571         return ret;
 572 }
 573
 574 /* Extracts a file, directory, or symbolic link from the WIM archive. */
 575 static int apply_dentry_normal(struct wim_dentry *dentry, void *arg)
 576 {
 577         struct apply_args *args = arg;
 578         struct wim_inode *inode = dentry->d_inode;
 579         size_t len;
 580
 581         len = strlen(args->target);
 582         char output_path[len + dentry->full_path_utf8_len + 1];
 583         memcpy(output_path, args->target, len);
 584         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 585         output_path[len + dentry->full_path_utf8_len] = '\0';
 586
 587 #if defined(__CYGWIN__) || defined(__WIN32__)
 588         char *utf16_path;
 589         size_t utf16_path_len;
 590         DWORD err;
 591         int ret;
 592         ret = utf8_to_utf16(output_path, len + dentry->full_path_utf8_len,
 593                             &utf16_path, &utf16_path_len);
 594         if (ret)
 595                 return ret;
 596
 597         if (inode->i_nlink > 1 && inode->i_extracted_file != NULL) {
 598                 /* Linked file, with another name already extracted */
 599                 if (!CreateHardLinkW((const wchar_t*)inode->i_extracted_file,
 600                                      (const wchar_t*)utf16_path,
 601                                      NULL))
 602                 {
 603                         err = GetLastError();
 604                         ERROR("Can't create hard link \"%s\"", output_path);
 605                         ret = WIMLIB_ERR_LINK;
 606                         win32_error(err);
 607                         goto out_free_utf16_path;
 608                 }
 609                 ret = 0;
 610                 goto out_free_utf16_path;
 611         }
 612         ret = win32_extract_streams(inode, (const wchar_t*)utf16_path,
 613                                     output_path);
 614         if (ret)
 615                 goto out_free_utf16_path;
 616
 617         if (inode->i_security_id != -1) {
 618                 ret = win32_set_security_data(inode,
 619                                               (const wchar_t*)utf16_path,
 620                                               wim_const_security_data(args->w),
 621                                               output_path);
 622                 if (ret)
 623                         goto out_free_utf16_path;
 624         }
 625
 626         /*if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {*/
 627                 /*ret = win32_set_reparse_data(inode, path, output_path);*/
 628                 /*if (ret)*/
 629                         /*goto out_free_utf16_path;*/
 630         /*}*/
 631
 632         if (inode->i_nlink > 1) {
 633                 inode->i_extracted_file = utf16_path;
 634                 utf16_path = NULL;
 635         }
 636         ret = 0;
 637 out_free_utf16_path:
 638         FREE(utf16_path);
 639         return ret;
 640 #else
 641         if (inode_is_symlink(inode))
 642                 return extract_symlink(dentry, args, output_path);
 643         else if (inode_is_directory(inode))
 644                 return extract_directory((args->extract_flags &
 645                                            WIMLIB_EXTRACT_FLAG_UNIX_DATA) ? dentry : NULL,
 646                                          output_path, false);
 647         else
 648                 return extract_regular_file(dentry, args, output_path);
 649 #endif
 650 }
 651
 652 /* Apply timestamps to an extracted file or directory */
 653 static int apply_dentry_timestamps_normal(struct wim_dentry *dentry, void *arg)
 654 {
 655         struct apply_args *args = arg;
 656         size_t len = strlen(args->target);
 657         char output_path[len + dentry->full_path_utf8_len + 1];
 658         const struct wim_inode *inode = dentry->d_inode;
 659         int ret;
 660
 661         memcpy(output_path, args->target, len);
 662         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 663         output_path[len + dentry->full_path_utf8_len] = '\0';
 664
 665 #if defined(__CYGWIN__) || defined(__WIN32__)
 666         /* Win32 */
 667         char *utf16_path;
 668         size_t utf16_path_len;
 669         DWORD err;
 670         HANDLE h;
 671         BOOL bret1, bret2;
 672
 673         ret = utf8_to_utf16(output_path, len + dentry->full_path_utf8_len,
 674                             &utf16_path, &utf16_path_len);
 675         if (ret)
 676                 return ret;
 677         h = CreateFile(utf16_path, GENERIC_WRITE, 0, NULL, OPEN_EXISTING,
 678                        FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OPEN_REPARSE_POINT,
 679                        NULL);
 680
 681         if (!h)
 682                 err = GetLastError();
 683         FREE(utf16_path);
 684         if (!h)
 685                 goto fail;
 686
 687         FILETIME creationTime = {.dwLowDateTime = dentry->d_inode->i_creation_time & 0xffffffff,
 688                                  .dwHighDateTime = dentry->d_inode->i_creation_time >> 32};
 689         FILETIME lastAccessTime = {.dwLowDateTime = dentry->d_inode->i_last_access_time & 0xffffffff,
 690                                   .dwHighDateTime = dentry->d_inode->i_last_access_time >> 32};
 691         FILETIME lastWriteTime = {.dwLowDateTime = dentry->d_inode->i_last_write_time & 0xffffffff,
 692                                   .dwHighDateTime = dentry->d_inode->i_last_write_time >> 32};
 693
 694         if (!SetFileTime(h, &creationTime, &lastAccessTime, &lastWriteTime)) {
 695                 err = GetLastError();
 696                 CloseHandle(h);
 697                 goto fail;
 698         }
 699         if (!CloseHandle(h)) {
 700                 err = GetLastError();
 701                 goto fail;
 702         }
 703         return 0;
 704 fail:
 705         err = GetLastError();
 706         ERROR("Can't set timestamps on \"%s\"", output_path);
 707         win32_error(err);
 708         return WIMLIB_ERR_WRITE;
 709 #else
 710         /* UNIX */
 711
 712         /* Convert the WIM timestamps, which are accurate to 100 nanoseconds,
 713          * into struct timeval's. */
 714         struct timeval tv[2];
 715         wim_timestamp_to_timeval(inode->i_last_access_time, &tv[0]);
 716         wim_timestamp_to_timeval(inode->i_last_write_time, &tv[1]);
 717         #ifdef HAVE_LUTIMES
 718         ret = lutimes(output_path, tv);
 719         #else
 720         ret = -1;
 721         errno = ENOSYS;
 722         #endif
 723         if (ret != 0) {
 724                 #ifdef HAVE_UTIME
 725                 if (errno == ENOSYS) {
 726                         struct utimbuf buf;
 727                         buf.actime = wim_timestamp_to_unix(inode->i_last_access_time);
 728                         buf.modtime = wim_timestamp_to_unix(inode->i_last_write_time);
 729                         if (utime(output_path, &buf) == 0)
 730                                 return 0;
 731                 }
 732                 #endif
 733                 if (errno != ENOSYS || args->num_lutimes_warnings < 10) {
 734                         /*WARNING_WITH_ERRNO("Failed to set timestamp on file `%s',*/
 735                                             /*output_path");*/
 736                         args->num_lutimes_warnings++;
 737                 }
 738         }
 739         return 0;
 740 #endif
 741 }
 742
 743 /* Extract a dentry if it hasn't already been extracted, and either the dentry
 744  * has no streams or WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified. */
 745 static int maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
 746 {
 747         struct apply_args *args = arg;
 748         int ret;
 749
 750         if (dentry->is_extracted)
 751                 return 0;
 752
 753         if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS)
 754                 if (inode_unnamed_lte_resolved(dentry->d_inode))
 755                         return 0;
 756
 757         if ((args->extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) &&
 758              args->progress_func) {
 759                 args->progress.extract.cur_path = dentry->full_path_utf8;
 760                 args->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY,
 761                                     &args->progress);
 762         }
 763         ret = args->apply_dentry(dentry, args);
 764         if (ret == 0)
 765                 dentry->is_extracted = 1;
 766         return ret;
 767 }
 768
 769 static int cmp_streams_by_wim_position(const void *p1, const void *p2)
 770 {
 771         const struct wim_lookup_table_entry *lte1, *lte2;
 772         lte1 = *(const struct wim_lookup_table_entry**)p1;
 773         lte2 = *(const struct wim_lookup_table_entry**)p2;
 774         if (lte1->resource_entry.offset < lte2->resource_entry.offset)
 775                 return -1;
 776         else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
 777                 return 1;
 778         else
 779                 return 0;
 780 }
 781
 782 static int sort_stream_list_by_wim_position(struct list_head *stream_list)
 783 {
 784         struct list_head *cur;
 785         size_t num_streams;
 786         struct wim_lookup_table_entry **array;
 787         size_t i;
 788         size_t array_size;
 789
 790         num_streams = 0;
 791         list_for_each(cur, stream_list)
 792                 num_streams++;
 793         array_size = num_streams * sizeof(array[0]);
 794         array = MALLOC(array_size);
 795         if (!array) {
 796                 ERROR("Failed to allocate %zu bytes to sort stream entries",
 797                       array_size);
 798                 return WIMLIB_ERR_NOMEM;
 799         }
 800         cur = stream_list->next;
 801         for (i = 0; i < num_streams; i++) {
 802                 array[i] = container_of(cur, struct wim_lookup_table_entry, staging_list);
 803                 cur = cur->next;
 804         }
 805
 806         qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
 807
 808         INIT_LIST_HEAD(stream_list);
 809         for (i = 0; i < num_streams; i++)
 810                 list_add_tail(&array[i]->staging_list, stream_list);
 811         FREE(array);
 812         return 0;
 813 }
 814
 815 static void calculate_bytes_to_extract(struct list_head *stream_list,
 816                                        int extract_flags,
 817                                        union wimlib_progress_info *progress)
 818 {
 819         struct wim_lookup_table_entry *lte;
 820         u64 total_bytes = 0;
 821         u64 num_streams = 0;
 822
 823         /* For each stream to be extracted... */
 824         list_for_each_entry(lte, stream_list, staging_list) {
 825                 if (extract_flags &
 826                     (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 827                 {
 828                         /* In the symlink or hard link extraction mode, each
 829                          * stream will be extracted one time regardless of how
 830                          * many dentries share the stream. */
 831                         wimlib_assert(!(extract_flags & WIMLIB_EXTRACT_FLAG_NTFS));
 832                         if (!lte->extracted_file) {
 833                                 num_streams++;
 834                                 total_bytes += wim_resource_size(lte);
 835                         }
 836                 } else {
 837                         num_streams += lte->out_refcnt;
 838                         total_bytes += lte->out_refcnt * wim_resource_size(lte);
 839                 }
 840         }
 841         progress->extract.num_streams = num_streams;
 842         progress->extract.total_bytes = total_bytes;
 843         progress->extract.completed_bytes = 0;
 844 }
 845
 846 static void maybe_add_stream_for_extraction(struct wim_lookup_table_entry *lte,
 847                                             struct list_head *stream_list)
 848 {
 849         if (++lte->out_refcnt == 1) {
 850                 INIT_LIST_HEAD(&lte->inode_list);
 851                 list_add_tail(&lte->staging_list, stream_list);
 852         }
 853 }
 854
 855 static void inode_find_streams_for_extraction(struct wim_inode *inode,
 856                                               struct list_head *stream_list,
 857                                               int extract_flags)
 858 {
 859         struct wim_lookup_table_entry *lte;
 860         bool inode_added = false;
 861
 862         lte = inode_unnamed_lte_resolved(inode);
 863         if (lte) {
 864                 maybe_add_stream_for_extraction(lte, stream_list);
 865                 list_add_tail(&inode->i_lte_inode_list, &lte->inode_list);
 866                 inode_added = true;
 867         }
 868 #ifdef WITH_NTFS_3G
 869         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
 870                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
 871                         if (inode->i_ads_entries[i].stream_name_len != 0) {
 872                                 lte = inode->i_ads_entries[i].lte;
 873                                 if (lte) {
 874                                         maybe_add_stream_for_extraction(lte,
 875                                                                         stream_list);
 876                                         if (!inode_added) {
 877                                                 list_add_tail(&inode->i_lte_inode_list,
 878                                                               &lte->inode_list);
 879                                                 inode_added = true;
 880                                         }
 881                                 }
 882                         }
 883                 }
 884         }
 885 #endif
 886 }
 887
 888 static void find_streams_for_extraction(struct hlist_head *inode_list,
 889                                         struct list_head *stream_list,
 890                                         struct wim_lookup_table *lookup_table,
 891                                         int extract_flags)
 892 {
 893         struct wim_inode *inode;
 894         struct hlist_node *cur;
 895         struct wim_dentry *dentry;
 896
 897         for_lookup_table_entry(lookup_table, lte_zero_out_refcnt, NULL);
 898         INIT_LIST_HEAD(stream_list);
 899         hlist_for_each_entry(inode, cur, inode_list, i_hlist) {
 900                 if (!inode->i_resolved)
 901                         inode_resolve_ltes(inode, lookup_table);
 902                 inode_for_each_dentry(dentry, inode)
 903                         dentry->is_extracted = 0;
 904                 inode_find_streams_for_extraction(inode, stream_list,
 905                                                   extract_flags);
 906         }
 907 }
 908
 909 struct apply_operations {
 910         int (*apply_dentry)(struct wim_dentry *dentry, void *arg);
 911         int (*apply_dentry_timestamps)(struct wim_dentry *dentry, void *arg);
 912 };
 913
 914 static const struct apply_operations normal_apply_operations = {
 915         .apply_dentry = apply_dentry_normal,
 916         .apply_dentry_timestamps = apply_dentry_timestamps_normal,
 917 };
 918
 919 #ifdef WITH_NTFS_3G
 920 static const struct apply_operations ntfs_apply_operations = {
 921         .apply_dentry = apply_dentry_ntfs,
 922         .apply_dentry_timestamps = apply_dentry_timestamps_ntfs,
 923 };
 924 #endif
 925
 926 static int apply_stream_list(struct list_head *stream_list,
 927                              struct apply_args *args,
 928                              const struct apply_operations *ops,
 929                              wimlib_progress_func_t progress_func)
 930 {
 931         uint64_t bytes_per_progress = args->progress.extract.total_bytes / 100;
 932         uint64_t next_progress = bytes_per_progress;
 933         struct wim_lookup_table_entry *lte;
 934         struct wim_inode *inode;
 935         struct wim_dentry *dentry;
 936         int ret;
 937
 938         /* This complicated loop is essentially looping through the dentries,
 939          * although dentries may be visited more than once (if a dentry contains
 940          * two different nonempty streams) or not at all (if a dentry contains
 941          * no non-empty streams).
 942          *
 943          * The outer loop is over the distinct streams to be extracted so that
 944          * sequential reading of the WIM can be implemented. */
 945
 946         /* For each distinct stream to be extracted */
 947         list_for_each_entry(lte, stream_list, staging_list) {
 948                 /* For each inode that contains the stream */
 949                 list_for_each_entry(inode, &lte->inode_list, i_lte_inode_list) {
 950                         /* For each dentry that points to the inode */
 951                         inode_for_each_dentry(dentry, inode) {
 952                                 /* Extract the dentry if it was not already
 953                                  * extracted */
 954                                 ret = maybe_apply_dentry(dentry, args);
 955                                 if (ret != 0)
 956                                         return ret;
 957                                 if (progress_func &&
 958                                     args->progress.extract.completed_bytes >= next_progress)
 959                                 {
 960                                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS,
 961                                                       &args->progress);
 962                                         if (args->progress.extract.completed_bytes >=
 963                                             args->progress.extract.total_bytes)
 964                                         {
 965                                                 next_progress = ~0ULL;
 966                                         } else {
 967                                                 next_progress =
 968                                                         min (args->progress.extract.completed_bytes +
 969                                                              bytes_per_progress,
 970                                                              args->progress.extract.total_bytes);
 971                                         }
 972                                 }
 973                         }
 974                 }
 975         }
 976         return 0;
 977 }
 978
 979 /* Extracts the image @image from the WIM @w to the directory or NTFS volume
 980  * @target. */
 981 static int extract_single_image(WIMStruct *w, int image,
 982                                 const char *target, int extract_flags,
 983                                 wimlib_progress_func_t progress_func)
 984 {
 985         int ret;
 986         struct list_head stream_list;
 987         struct hlist_head *inode_list;
 988
 989         struct apply_args args;
 990         const struct apply_operations *ops;
 991
 992         args.w                    = w;
 993         args.target               = target;
 994         args.extract_flags        = extract_flags;
 995         args.num_lutimes_warnings = 0;
 996         args.stream_list          = &stream_list;
 997         args.progress_func        = progress_func;
 998
 999         if (progress_func) {
1000                 args.progress.extract.wimfile_name = w->filename;
1001                 args.progress.extract.image = image;
1002                 args.progress.extract.extract_flags = (extract_flags &
1003                                                        WIMLIB_EXTRACT_MASK_PUBLIC);
1004                 args.progress.extract.image_name = wimlib_get_image_name(w, image);
1005                 args.progress.extract.target = target;
1006         }
1007
1008 #ifdef WITH_NTFS_3G
1009         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1010                 args.vol = ntfs_mount(target, 0);
1011                 if (!args.vol) {
1012                         ERROR_WITH_ERRNO("Failed to mount NTFS volume `%s'", target);
1013                         return WIMLIB_ERR_NTFS_3G;
1014                 }
1015                 ops = &ntfs_apply_operations;
1016         } else
1017 #endif
1018                 ops = &normal_apply_operations;
1019
1020         ret = select_wim_image(w, image);
1021         if (ret != 0)
1022                 goto out;
1023
1024         inode_list = &w->image_metadata[image - 1].inode_list;
1025
1026         /* Build a list of the streams that need to be extracted */
1027         find_streams_for_extraction(inode_list, &stream_list,
1028                                     w->lookup_table, extract_flags);
1029
1030         /* Calculate the number of bytes of data that will be extracted */
1031         calculate_bytes_to_extract(&stream_list, extract_flags,
1032                                    &args.progress);
1033
1034         if (progress_func) {
1035                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
1036                               &args.progress);
1037         }
1038
1039         /* If a sequential extraction was specified, sort the streams to be
1040          * extracted by their position in the WIM file, so that the WIM file can
1041          * be read sequentially. */
1042         if (extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL) {
1043                 ret = sort_stream_list_by_wim_position(&stream_list);
1044                 if (ret != 0) {
1045                         WARNING("Falling back to non-sequential extraction");
1046                         extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
1047                 }
1048         }
1049
1050         if (progress_func) {
1051                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
1052                               &args.progress);
1053         }
1054
1055         /* Make the directory structure and extract empty files */
1056         args.extract_flags |= WIMLIB_EXTRACT_FLAG_NO_STREAMS;
1057         args.apply_dentry = ops->apply_dentry;
1058         ret = for_dentry_in_tree(wim_root_dentry(w), maybe_apply_dentry, &args);
1059         args.extract_flags &= ~WIMLIB_EXTRACT_FLAG_NO_STREAMS;
1060         if (ret != 0)
1061                 goto out;
1062
1063         if (progress_func) {
1064                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
1065                               &args.progress);
1066         }
1067
1068         /* Extract non-empty files */
1069         ret = apply_stream_list(&stream_list, &args, ops, progress_func);
1070         if (ret != 0)
1071                 goto out;
1072
1073         if (progress_func) {
1074                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
1075                               &args.progress);
1076         }
1077
1078         /* Apply timestamps */
1079         ret = for_dentry_in_tree_depth(wim_root_dentry(w),
1080                                        ops->apply_dentry_timestamps, &args);
1081         if (ret != 0)
1082                 goto out;
1083
1084         if (progress_func) {
1085                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
1086                               &args.progress);
1087         }
1088 out:
1089 #ifdef WITH_NTFS_3G
1090         /* Unmount the NTFS volume */
1091         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1092                 if (ntfs_umount(args.vol, FALSE) != 0) {
1093                         ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%s'", args.target);
1094                         if (ret == 0)
1095                                 ret = WIMLIB_ERR_NTFS_3G;
1096                 }
1097         }
1098 #endif
1099         return ret;
1100 }
1101
1102
1103 /* Extracts all images from the WIM to the directory @target, with the images
1104  * placed in subdirectories named by their image names. */
1105 static int extract_all_images(WIMStruct *w, const char *target,
1106                               int extract_flags,
1107                               wimlib_progress_func_t progress_func)
1108 {
1109         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
1110         size_t output_path_len = strlen(target);
1111         char buf[output_path_len + 1 + image_name_max_len + 1];
1112         int ret;
1113         int image;
1114         const char *image_name;
1115
1116         ret = extract_directory(NULL, target, true);
1117         if (ret != 0)
1118                 return ret;
1119
1120         memcpy(buf, target, output_path_len);
1121         buf[output_path_len] = '/';
1122         for (image = 1; image <= w->hdr.image_count; image++) {
1123                 image_name = wimlib_get_image_name(w, image);
1124                 if (image_name && *image_name) {
1125                         strcpy(buf + output_path_len + 1, image_name);
1126                 } else {
1127                         /* Image name is empty. Use image number instead */
1128                         sprintf(buf + output_path_len + 1, "%d", image);
1129                 }
1130                 ret = extract_single_image(w, image, buf, extract_flags,
1131                                            progress_func);
1132                 if (ret != 0)
1133                         return ret;
1134         }
1135         return 0;
1136 }
1137
1138 /* Extracts a single image or all images from a WIM file to a directory or NTFS
1139  * volume. */
1140 WIMLIBAPI int wimlib_extract_image(WIMStruct *w,
1141                                    int image,
1142                                    const char *target,
1143                                    int extract_flags,
1144                                    WIMStruct **additional_swms,
1145                                    unsigned num_additional_swms,
1146                                    wimlib_progress_func_t progress_func)
1147 {
1148         struct wim_lookup_table *joined_tab, *w_tab_save;
1149         int ret;
1150
1151         if (!target)
1152                 return WIMLIB_ERR_INVALID_PARAM;
1153
1154         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
1155
1156         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
1157                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
1158                 return WIMLIB_ERR_INVALID_PARAM;
1159
1160 #if defined(__CYGWIN__) || defined(__WIN32__)
1161         if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
1162                 ERROR("Extracting UNIX data is not supported on Windows");
1163                 return WIMLIB_ERR_INVALID_PARAM;
1164         }
1165 #endif
1166
1167         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1168 #ifdef WITH_NTFS_3G
1169                 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))) {
1170                         ERROR("Cannot specify symlink or hardlink flags when applying\n"
1171                               "        directly to a NTFS volume");
1172                         return WIMLIB_ERR_INVALID_PARAM;
1173                 }
1174                 if (image == WIMLIB_ALL_IMAGES) {
1175                         ERROR("Can only apply a single image when applying "
1176                               "directly to a NTFS volume");
1177                         return WIMLIB_ERR_INVALID_PARAM;
1178                 }
1179                 if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
1180                         ERROR("Cannot restore UNIX-specific data in the NTFS extraction mode");
1181                         return WIMLIB_ERR_INVALID_PARAM;
1182                 }
1183 #else
1184                 ERROR("wimlib was compiled without support for NTFS-3g, so");
1185                 ERROR("we cannot apply a WIM image directly to a NTFS volume");
1186                 return WIMLIB_ERR_UNSUPPORTED;
1187 #endif
1188         }
1189
1190         ret = verify_swm_set(w, additional_swms, num_additional_swms);
1191         if (ret != 0)
1192                 return ret;
1193
1194         if (num_additional_swms) {
1195                 ret = new_joined_lookup_table(w, additional_swms,
1196                                               num_additional_swms, &joined_tab);
1197                 if (ret != 0)
1198                         return ret;
1199                 w_tab_save = w->lookup_table;
1200                 w->lookup_table = joined_tab;
1201         }
1202
1203         if (image == WIMLIB_ALL_IMAGES) {
1204                 extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
1205                 ret = extract_all_images(w, target, extract_flags,
1206                                          progress_func);
1207         } else {
1208                 extract_flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
1209                 ret = extract_single_image(w, image, target, extract_flags,
1210                                            progress_func);
1211         }
1212
1213         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
1214                              WIMLIB_EXTRACT_FLAG_HARDLINK))
1215         {
1216                 for_lookup_table_entry(w->lookup_table,
1217                                        lte_free_extracted_file,
1218                                        NULL);
1219         }
1220
1221         if (num_additional_swms) {
1222                 free_lookup_table(w->lookup_table);
1223                 w->lookup_table = w_tab_save;
1224         }
1225         return ret;
1226 }