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

   1 /*
   2  * add_image.c
   3  */
   4
   5 /*
   6  * Copyright (C) 2012, 2013 Eric Biggers
   7  *
   8  * This file is part of wimlib, a library for working with WIM files.
   9  *
  10  * wimlib is free software; you can redistribute it and/or modify it under the
  11  * terms of the GNU General Public License as published by the Free
  12  * Software Foundation; either version 3 of the License, or (at your option)
  13  * any later version.
  14  *
  15  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  16  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  17  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  18  * details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  22  */
  23
  24 #include "config.h"
  25
  26 #ifdef __WIN32__
  27 #  include "win32.h"
  28 #else
  29 #  include <dirent.h>
  30 #  include <sys/stat.h>
  31 #  include <fnmatch.h>
  32 #  include "timestamp.h"
  33 #endif
  34
  35 #include "wimlib_internal.h"
  36 #include "dentry.h"
  37 #include "lookup_table.h"
  38 #include "xml.h"
  39 #include "security.h"
  40
  41 #include <ctype.h>
  42 #include <errno.h>
  43 #include <stdlib.h>
  44 #include <limits.h>
  45 #include <string.h>
  46
  47 #include <unistd.h>
  48
  49 #ifdef HAVE_ALLOCA_H
  50 #  include <alloca.h>
  51 #endif
  52
  53 /*
  54  * Adds the dentry tree and security data for a new image to the image metadata
  55  * array of the WIMStruct.
  56  */
  57 int
  58 add_new_dentry_tree(WIMStruct *w, struct wim_dentry *root_dentry,
  59                     struct wim_security_data *sd)
  60 {
  61         struct wim_lookup_table_entry *metadata_lte;
  62         struct wim_image_metadata *imd;
  63         struct wim_image_metadata *new_imd;
  64
  65         wimlib_assert(root_dentry != NULL);
  66
  67         DEBUG("Reallocating image metadata array for image_count = %u",
  68               w->hdr.image_count + 1);
  69         imd = CALLOC((w->hdr.image_count + 1), sizeof(struct wim_image_metadata));
  70
  71         if (!imd) {
  72                 ERROR("Failed to allocate memory for new image metadata array");
  73                 goto err;
  74         }
  75
  76         memcpy(imd, w->image_metadata,
  77                w->hdr.image_count * sizeof(struct wim_image_metadata));
  78
  79         metadata_lte = new_lookup_table_entry();
  80         if (!metadata_lte)
  81                 goto err_free_imd;
  82
  83         metadata_lte->resource_entry.flags = WIM_RESHDR_FLAG_METADATA;
  84
  85         new_imd = &imd[w->hdr.image_count];
  86
  87         new_imd->root_dentry    = root_dentry;
  88         new_imd->metadata_lte   = metadata_lte;
  89         new_imd->security_data  = sd;
  90         new_imd->modified       = 1;
  91
  92         FREE(w->image_metadata);
  93         w->image_metadata = imd;
  94         w->hdr.image_count++;
  95         return 0;
  96 err_free_imd:
  97         FREE(imd);
  98 err:
  99         return WIMLIB_ERR_NOMEM;
 100
 101 }
 102
 103 #ifndef __WIN32__
 104
 105 static int
 106 unix_capture_regular_file(const char *path,
 107                           uint64_t size,
 108                           struct wim_inode *inode,
 109                           struct wim_lookup_table *lookup_table)
 110 {
 111         struct wim_lookup_table_entry *lte;
 112         u8 hash[SHA1_HASH_SIZE];
 113         int ret;
 114
 115         inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 116
 117         /* Empty files do not have to have a lookup table entry. */
 118         if (size == 0)
 119                 return 0;
 120
 121         /* For each regular file, we must check to see if the file is in
 122          * the lookup table already; if it is, we increment its refcnt;
 123          * otherwise, we create a new lookup table entry and insert it.
 124          * */
 125
 126         ret = sha1sum(path, hash);
 127         if (ret)
 128                 return ret;
 129
 130         lte = __lookup_resource(lookup_table, hash);
 131         if (lte) {
 132                 lte->refcnt++;
 133                 DEBUG("Add lte reference %u for `%s'", lte->refcnt,
 134                       path);
 135         } else {
 136                 char *file_on_disk = STRDUP(path);
 137                 if (!file_on_disk) {
 138                         ERROR("Failed to allocate memory for file path");
 139                         return WIMLIB_ERR_NOMEM;
 140                 }
 141                 lte = new_lookup_table_entry();
 142                 if (!lte) {
 143                         FREE(file_on_disk);
 144                         return WIMLIB_ERR_NOMEM;
 145                 }
 146                 lte->file_on_disk = file_on_disk;
 147                 lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 148                 lte->resource_entry.original_size = size;
 149                 lte->resource_entry.size = size;
 150                 copy_hash(lte->hash, hash);
 151                 lookup_table_insert(lookup_table, lte);
 152         }
 153         inode->i_lte = lte;
 154         return 0;
 155 }
 156
 157 static int
 158 unix_build_dentry_tree_recursive(struct wim_dentry **root_ret,
 159                                  char *path,
 160                                  size_t path_len,
 161                                  struct wim_lookup_table *lookup_table,
 162                                  const struct wimlib_capture_config *config,
 163                                  int add_image_flags,
 164                                  wimlib_progress_func_t progress_func);
 165
 166 static int
 167 unix_capture_directory(struct wim_dentry *dir_dentry,
 168                        char *path,
 169                        size_t path_len,
 170                        struct wim_lookup_table *lookup_table,
 171                        const struct wimlib_capture_config *config,
 172                        int add_image_flags,
 173                        wimlib_progress_func_t progress_func)
 174 {
 175
 176         DIR *dir;
 177         struct dirent entry, *result;
 178         struct wim_dentry *child;
 179         int ret;
 180
 181         dir_dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 182         dir = opendir(path);
 183         if (!dir) {
 184                 ERROR_WITH_ERRNO("Failed to open the directory `%s'",
 185                                  path);
 186                 return WIMLIB_ERR_OPEN;
 187         }
 188
 189         /* Recurse on directory contents */
 190         while (1) {
 191                 errno = 0;
 192                 ret = readdir_r(dir, &entry, &result);
 193                 if (ret != 0) {
 194                         ret = WIMLIB_ERR_READ;
 195                         ERROR_WITH_ERRNO("Error reading the "
 196                                          "directory `%s'",
 197                                          path);
 198                         break;
 199                 }
 200                 if (result == NULL)
 201                         break;
 202                 if (result->d_name[0] == '.' && (result->d_name[1] == '\0'
 203                       || (result->d_name[1] == '.' && result->d_name[2] == '\0')))
 204                                 continue;
 205
 206                 size_t name_len = strlen(result->d_name);
 207
 208                 path[path_len] = '/';
 209                 memcpy(&path[path_len + 1], result->d_name, name_len + 1);
 210                 ret = unix_build_dentry_tree_recursive(&child,
 211                                                        path,
 212                                                        path_len + 1 + name_len,
 213                                                        lookup_table,
 214                                                        config,
 215                                                        add_image_flags,
 216                                                        progress_func);
 217                 if (ret)
 218                         break;
 219                 if (child)
 220                         dentry_add_child(dir_dentry, child);
 221         }
 222         closedir(dir);
 223         return ret;
 224 }
 225
 226 static int
 227 unix_capture_symlink(const char *path,
 228                      struct wim_inode *inode,
 229                      struct wim_lookup_table *lookup_table)
 230 {
 231         char deref_name_buf[4096];
 232         ssize_t deref_name_len;
 233         int ret;
 234
 235         inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 236         inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 237
 238         /* The idea here is to call readlink() to get the UNIX target of
 239          * the symbolic link, then turn the target into a reparse point
 240          * data buffer that contains a relative or absolute symbolic
 241          * link (NOT a junction point or *full* path symbolic link with
 242          * drive letter).
 243          */
 244         deref_name_len = readlink(path, deref_name_buf,
 245                                   sizeof(deref_name_buf) - 1);
 246         if (deref_name_len >= 0) {
 247                 deref_name_buf[deref_name_len] = '\0';
 248                 DEBUG("Read symlink `%s'", deref_name_buf);
 249                 ret = inode_set_symlink(inode, deref_name_buf,
 250                                         lookup_table, NULL);
 251                 if (ret == 0) {
 252                         /* Unfortunately, Windows seems to have the concept of
 253                          * "file" symbolic links as being different from
 254                          * "directory" symbolic links...  so
 255                          * FILE_ATTRIBUTE_DIRECTORY needs to be set on the
 256                          * symbolic link if the *target* of the symbolic link is
 257                          * a directory.  */
 258                         struct stat stbuf;
 259                         if (stat(path, &stbuf) == 0 && S_ISDIR(stbuf.st_mode))
 260                                 inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 261                 }
 262         } else {
 263                 ERROR_WITH_ERRNO("Failed to read target of "
 264                                  "symbolic link `%s'", path);
 265                 ret = WIMLIB_ERR_READLINK;
 266         }
 267         return ret;
 268 }
 269
 270 static int
 271 unix_build_dentry_tree_recursive(struct wim_dentry **root_ret,
 272                                  char *path,
 273                                  size_t path_len,
 274                                  struct wim_lookup_table *lookup_table,
 275                                  const struct wimlib_capture_config *config,
 276                                  int add_image_flags,
 277                                  wimlib_progress_func_t progress_func)
 278 {
 279         struct wim_dentry *root = NULL;
 280         int ret = 0;
 281         struct wim_inode *inode;
 282
 283         if (exclude_path(path, path_len, config, true)) {
 284                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) {
 285                         ERROR("Cannot exclude the root directory from capture");
 286                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 287                         goto out;
 288                 }
 289                 if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_EXCLUDE_VERBOSE)
 290                     && progress_func)
 291                 {
 292                         union wimlib_progress_info info;
 293                         info.scan.cur_path = path;
 294                         info.scan.excluded = true;
 295                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 296                 }
 297                 goto out;
 298         }
 299
 300         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 301             && progress_func)
 302         {
 303                 union wimlib_progress_info info;
 304                 info.scan.cur_path = path;
 305                 info.scan.excluded = false;
 306                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 307         }
 308
 309         /* UNIX version of capturing a directory tree */
 310         struct stat stbuf;
 311         int (*stat_fn)(const char *restrict, struct stat *restrict);
 312         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)
 313                 stat_fn = stat;
 314         else
 315                 stat_fn = lstat;
 316
 317         ret = (*stat_fn)(path, &stbuf);
 318         if (ret != 0) {
 319                 ERROR_WITH_ERRNO("Failed to stat `%s'", path);
 320                 goto out;
 321         }
 322
 323         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) &&
 324               !S_ISDIR(stbuf.st_mode))
 325         {
 326                 /* Do a dereference-stat in case the root is a symbolic link.
 327                  * This case is allowed, provided that the symbolic link points
 328                  * to a directory. */
 329                 ret = stat(path, &stbuf);
 330                 if (ret != 0) {
 331                         ERROR_WITH_ERRNO("Failed to stat `%s'", path);
 332                         ret = WIMLIB_ERR_STAT;
 333                         goto out;
 334                 }
 335                 if (!S_ISDIR(stbuf.st_mode)) {
 336                         ERROR("`%s' is not a directory", path);
 337                         ret = WIMLIB_ERR_NOTDIR;
 338                         goto out;
 339                 }
 340         }
 341         if (!S_ISREG(stbuf.st_mode) && !S_ISDIR(stbuf.st_mode)
 342             && !S_ISLNK(stbuf.st_mode)) {
 343                 ERROR("`%s' is not a regular file, directory, or symbolic link.",
 344                       path);
 345                 ret = WIMLIB_ERR_SPECIAL_FILE;
 346                 goto out;
 347         }
 348
 349         ret = new_dentry_with_timeless_inode(path_basename_with_len(path, path_len),
 350                                              &root);
 351         if (ret)
 352                 goto out;
 353
 354         inode = root->d_inode;
 355
 356 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 357         inode->i_creation_time = timespec_to_wim_timestamp(stbuf.st_mtim);
 358         inode->i_last_write_time = timespec_to_wim_timestamp(stbuf.st_mtim);
 359         inode->i_last_access_time = timespec_to_wim_timestamp(stbuf.st_atim);
 360 #else
 361         inode->i_creation_time = unix_timestamp_to_wim(stbuf.st_mtime);
 362         inode->i_last_write_time = unix_timestamp_to_wim(stbuf.st_mtime);
 363         inode->i_last_access_time = unix_timestamp_to_wim(stbuf.st_atime);
 364 #endif
 365         /* Leave the inode number at 0 for directories.  Otherwise grab the
 366          * inode number from the `stat' buffer, including the device number if
 367          * possible. */
 368         if (!S_ISDIR(stbuf.st_mode)) {
 369                 if (sizeof(ino_t) >= 8)
 370                         inode->i_ino = (u64)stbuf.st_ino;
 371                 else
 372                         inode->i_ino = (u64)stbuf.st_ino |
 373                                            ((u64)stbuf.st_dev <<
 374                                                 ((sizeof(ino_t) * 8) & 63));
 375         }
 376         inode->i_resolved = 1;
 377         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 378                 ret = inode_set_unix_data(inode, stbuf.st_uid,
 379                                           stbuf.st_gid,
 380                                           stbuf.st_mode,
 381                                           lookup_table,
 382                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 383                 if (ret)
 384                         goto out;
 385         }
 386         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 387         if (S_ISREG(stbuf.st_mode))
 388                 ret = unix_capture_regular_file(path, stbuf.st_size,
 389                                                 inode, lookup_table);
 390         else if (S_ISDIR(stbuf.st_mode))
 391                 ret = unix_capture_directory(root, path, path_len,
 392                                              lookup_table, config,
 393                                              add_image_flags, progress_func);
 394         else
 395                 ret = unix_capture_symlink(path, inode, lookup_table);
 396 out:
 397         if (ret == 0)
 398                 *root_ret = root;
 399         else
 400                 free_dentry_tree(root, lookup_table);
 401         return ret;
 402 }
 403
 404 /*
 405  * unix_build_dentry_tree():
 406  *      Builds a tree of WIM dentries from an on-disk directory tree (UNIX
 407  *      version; no NTFS-specific data is captured).
 408  *
 409  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 410  *              modified if successful.  Set to NULL if the file or directory was
 411  *              excluded from capture.
 412  *
 413  * @root_disk_path:  The path to the root of the directory tree on disk.
 414  *
 415  * @lookup_table: The lookup table for the WIM file.  For each file added to the
 416  *              dentry tree being built, an entry is added to the lookup table,
 417  *              unless an identical stream is already in the lookup table.
 418  *              These lookup table entries that are added point to the path of
 419  *              the file on disk.
 420  *
 421  * @sd_set:     Ignored.  (Security data only captured in NTFS mode.)
 422  *
 423  * @config:
 424  *              Configuration for files to be excluded from capture.
 425  *
 426  * @add_flags:  Bitwise or of WIMLIB_ADD_IMAGE_FLAG_*
 427  *
 428  * @extra_arg:  Ignored
 429  *
 430  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 431  *              the files cannot be `stat'ed, or if any of the needed
 432  *              directories cannot be opened or read.  Failure to add the files
 433  *              to the WIM may still occur later when trying to actually read
 434  *              the on-disk files during a call to wimlib_write() or
 435  *              wimlib_overwrite().
 436  */
 437 static int
 438 unix_build_dentry_tree(struct wim_dentry **root_ret,
 439                        const char *root_disk_path,
 440                        struct wim_lookup_table *lookup_table,
 441                        struct sd_set *sd_set,
 442                        const struct wimlib_capture_config *config,
 443                        int add_image_flags,
 444                        wimlib_progress_func_t progress_func,
 445                        void *extra_arg)
 446 {
 447         char *path_buf;
 448         int ret;
 449         size_t path_len;
 450         size_t path_bufsz;
 451
 452         path_bufsz = min(32790, PATH_MAX + 1);
 453         path_len = strlen(root_disk_path);
 454
 455         if (path_len >= path_bufsz)
 456                 return WIMLIB_ERR_INVALID_PARAM;
 457
 458         path_buf = MALLOC(path_bufsz);
 459         if (!path_buf)
 460                 return WIMLIB_ERR_NOMEM;
 461         memcpy(path_buf, root_disk_path, path_len + 1);
 462         ret = unix_build_dentry_tree_recursive(root_ret,
 463                                                path_buf,
 464                                                path_len,
 465                                                lookup_table,
 466                                                config,
 467                                                add_image_flags,
 468                                                progress_func);
 469         FREE(path_buf);
 470         return ret;
 471 }
 472 #endif /* !__WIN32__ */
 473
 474 static bool
 475 match_pattern(const tchar *path,
 476               const tchar *path_basename,
 477               const struct wimlib_pattern_list *list)
 478 {
 479         for (size_t i = 0; i < list->num_pats; i++) {
 480
 481                 const tchar *pat = list->pats[i];
 482                 const tchar *string;
 483
 484                 if (*pat == T('/')) {
 485                         /* Absolute path from root of capture */
 486                         string = path;
 487                 } else {
 488                         if (tstrchr(pat, T('/')))
 489                                 /* Relative path from root of capture */
 490                                 string = path + 1;
 491                         else
 492                                 /* A file name pattern */
 493                                 string = path_basename;
 494                 }
 495
 496                 /* Warning: on Windows native builds, fnmatch() calls the
 497                  * replacement function in win32.c. */
 498                 if (fnmatch(pat, string, FNM_PATHNAME | FNM_NOESCAPE
 499                                 #ifdef FNM_CASEFOLD
 500                                         | FNM_CASEFOLD
 501                                 #endif
 502                             ) == 0)
 503                 {
 504                         DEBUG("\"%"TS"\" matches the pattern \"%"TS"\"",
 505                               string, pat);
 506                         return true;
 507                 } else {
 508                         DEBUG2("\"%"TS"\" does not match the pattern \"%"TS"\"",
 509                                string, pat);
 510                 }
 511         }
 512         return false;
 513 }
 514
 515 /* Return true if the image capture configuration file indicates we should
 516  * exclude the filename @path from capture.
 517  *
 518  * If @exclude_prefix is %true, the part of the path up and including the name
 519  * of the directory being captured is not included in the path for matching
 520  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
 521  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
 522  * directory.
 523  */
 524 bool
 525 exclude_path(const tchar *path, size_t path_len,
 526              const struct wimlib_capture_config *config, bool exclude_prefix)
 527 {
 528         const tchar *basename = path_basename_with_len(path, path_len);
 529         if (exclude_prefix) {
 530                 wimlib_assert(path_len >= config->_prefix_num_tchars);
 531                 if (!tmemcmp(config->_prefix, path, config->_prefix_num_tchars) &&
 532                     path[config->_prefix_num_tchars] == T('/'))
 533                 {
 534                         path += config->_prefix_num_tchars;
 535                 }
 536         }
 537         return match_pattern(path, basename, &config->exclusion_pats) &&
 538                 !match_pattern(path, basename, &config->exclusion_exception_pats);
 539
 540 }
 541
 542 /* Strip leading and trailing forward slashes from a string.  Modifies it in
 543  * place and returns the stripped string. */
 544 static const tchar *
 545 canonicalize_target_path(tchar *target_path)
 546 {
 547         tchar *p;
 548         if (target_path == NULL)
 549                 return T("");
 550         for (;;) {
 551                 if (*target_path == T('\0'))
 552                         return target_path;
 553                 else if (*target_path == T('/'))
 554                         target_path++;
 555                 else
 556                         break;
 557         }
 558
 559         p = tstrchr(target_path, T('\0')) - 1;
 560         while (*p == T('/'))
 561                 *p-- = T('\0');
 562         return target_path;
 563 }
 564
 565 /* Strip leading and trailing slashes from the target paths, and translate all
 566  * backslashes in the source and target paths into forward slashes. */
 567 static void
 568 canonicalize_sources_and_targets(struct wimlib_capture_source *sources,
 569                                  size_t num_sources)
 570 {
 571         while (num_sources--) {
 572                 DEBUG("Canonicalizing { source: \"%"TS"\", target=\"%"TS"\"}",
 573                       sources->fs_source_path,
 574                       sources->wim_target_path);
 575
 576                 /* The Windows API can handle forward slashes.  Just get rid of
 577                  * backslashes to avoid confusing other parts of the library
 578                  * code. */
 579                 zap_backslashes(sources->fs_source_path);
 580                 if (sources->wim_target_path)
 581                         zap_backslashes(sources->wim_target_path);
 582
 583                 sources->wim_target_path =
 584                         (tchar*)canonicalize_target_path(sources->wim_target_path);
 585                 DEBUG("Canonical target: \"%"TS"\"", sources->wim_target_path);
 586                 sources++;
 587         }
 588 }
 589
 590 static int
 591 capture_source_cmp(const void *p1, const void *p2)
 592 {
 593         const struct wimlib_capture_source *s1 = p1, *s2 = p2;
 594         return tstrcmp(s1->wim_target_path, s2->wim_target_path);
 595 }
 596
 597 /* Sorts the capture sources lexicographically by target path.  This occurs
 598  * after leading and trailing forward slashes are stripped.
 599  *
 600  * One purpose of this is to make sure that target paths that are inside other
 601  * target paths are added after the containing target paths. */
 602 static void
 603 sort_sources(struct wimlib_capture_source *sources, size_t num_sources)
 604 {
 605         qsort(sources, num_sources, sizeof(sources[0]), capture_source_cmp);
 606 }
 607
 608 static int
 609 check_sorted_sources(struct wimlib_capture_source *sources, size_t num_sources,
 610                      int add_image_flags)
 611 {
 612         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 613                 if (num_sources != 1) {
 614                         ERROR("Must specify exactly 1 capture source "
 615                               "(the NTFS volume) in NTFS mode!");
 616                         return WIMLIB_ERR_INVALID_PARAM;
 617                 }
 618                 if (sources[0].wim_target_path[0] != T('\0')) {
 619                         ERROR("In NTFS capture mode the target path inside "
 620                               "the image must be the root directory!");
 621                         return WIMLIB_ERR_INVALID_PARAM;
 622                 }
 623         } else if (num_sources != 0) {
 624                 /* This code is disabled because the current code
 625                  * unconditionally attempts to do overlays.  So, duplicate
 626                  * target paths are OK. */
 627         #if 0
 628                 if (num_sources > 1 && sources[0].wim_target_path[0] == '\0') {
 629                         ERROR("Cannot specify root target when using multiple "
 630                               "capture sources!");
 631                         return WIMLIB_ERR_INVALID_PARAM;
 632                 }
 633                 for (size_t i = 0; i < num_sources - 1; i++) {
 634                         size_t len = strlen(sources[i].wim_target_path);
 635                         size_t j = i + 1;
 636                         const char *target1 = sources[i].wim_target_path;
 637                         do {
 638                                 const char *target2 = sources[j].wim_target_path;
 639                                 DEBUG("target1=%s, target2=%s",
 640                                       target1,target2);
 641                                 if (strncmp(target1, target2, len) ||
 642                                     target2[len] > '/')
 643                                         break;
 644                                 if (target2[len] == '/') {
 645                                         ERROR("Invalid target `%s': is a prefix of `%s'",
 646                                               target1, target2);
 647                                         return WIMLIB_ERR_INVALID_PARAM;
 648                                 }
 649                                 if (target2[len] == '\0') {
 650                                         ERROR("Invalid target `%s': is a duplicate of `%s'",
 651                                               target1, target2);
 652                                         return WIMLIB_ERR_INVALID_PARAM;
 653                                 }
 654                         } while (++j != num_sources);
 655                 }
 656         #endif
 657         }
 658         return 0;
 659
 660 }
 661
 662 /* Creates a new directory to place in the WIM image.  This is to create parent
 663  * directories that are not part of any target as needed.  */
 664 static int
 665 new_filler_directory(const tchar *name, struct wim_dentry **dentry_ret)
 666 {
 667         int ret;
 668         struct wim_dentry *dentry;
 669
 670         DEBUG("Creating filler directory \"%"TS"\"", name);
 671         ret = new_dentry_with_inode(name, &dentry);
 672         if (ret == 0) {
 673                 /* Leave the inode number as 0 for now.  The final inode number
 674                  * will be assigned later by assign_inode_numbers(). */
 675                 dentry->d_inode->i_resolved = 1;
 676                 dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 677                 *dentry_ret = dentry;
 678         }
 679         return ret;
 680 }
 681
 682 /* Transfers the children of @branch to @target.  It is an error if @target is
 683  * not a directory or if both @branch and @target contain a child dentry with
 684  * the same name. */
 685 static int
 686 do_overlay(struct wim_dentry *target, struct wim_dentry *branch)
 687 {
 688         struct rb_root *rb_root;
 689
 690         DEBUG("Doing overlay \"%"WS"\" => \"%"WS"\"",
 691               branch->file_name, target->file_name);
 692
 693         if (!dentry_is_directory(target)) {
 694                 ERROR("Cannot overlay directory \"%"WS"\" "
 695                       "over non-directory", branch->file_name);
 696                 return WIMLIB_ERR_INVALID_OVERLAY;
 697         }
 698
 699         rb_root = &branch->d_inode->i_children;
 700         while (rb_root->rb_node) { /* While @branch has children... */
 701                 struct wim_dentry *child = rbnode_dentry(rb_root->rb_node);
 702                 /* Move @child to the directory @target */
 703                 unlink_dentry(child);
 704                 if (!dentry_add_child(target, child)) {
 705                         /* Revert the change to avoid leaking the directory tree
 706                          * rooted at @child */
 707                         dentry_add_child(branch, child);
 708                         ERROR("Overlay error: file \"%"WS"\" already exists "
 709                               "as a child of \"%"WS"\"",
 710                               child->file_name, target->file_name);
 711                         return WIMLIB_ERR_INVALID_OVERLAY;
 712                 }
 713         }
 714         free_dentry(branch);
 715         return 0;
 716
 717 }
 718
 719 /* Attach or overlay a branch onto the WIM image.
 720  *
 721  * @root_p:
 722  *      Pointer to the root of the WIM image, or pointer to NULL if it has not
 723  *      been created yet.
 724  * @branch
 725  *      Branch to add.
 726  * @target_path:
 727  *      Path in the WIM image to add the branch, with leading and trailing
 728  *      slashes stripped.
 729  */
 730 static int
 731 attach_branch(struct wim_dentry **root_p, struct wim_dentry *branch,
 732               tchar *target_path)
 733 {
 734         tchar *slash;
 735         struct wim_dentry *dentry, *parent, *target;
 736         int ret;
 737
 738         DEBUG("Attaching branch \"%"WS"\" => \"%"TS"\"",
 739               branch->file_name, target_path);
 740
 741         if (*target_path == T('\0')) {
 742                 /* Target: root directory */
 743                 if (*root_p) {
 744                         /* Overlay on existing root */
 745                         return do_overlay(*root_p, branch);
 746                 } else  {
 747                         /* Set as root */
 748                         *root_p = branch;
 749                         return 0;
 750                 }
 751         }
 752
 753         /* Adding a non-root branch.  Create root if it hasn't been created
 754          * already. */
 755         if (!*root_p) {
 756                 ret  = new_filler_directory(T(""), root_p);
 757                 if (ret)
 758                         return ret;
 759         }
 760
 761         /* Walk the path to the branch, creating filler directories as needed.
 762          * */
 763         parent = *root_p;
 764         while ((slash = tstrchr(target_path, T('/')))) {
 765                 *slash = T('\0');
 766                 dentry = get_dentry_child_with_name(parent, target_path);
 767                 if (!dentry) {
 768                         ret = new_filler_directory(target_path, &dentry);
 769                         if (ret)
 770                                 return ret;
 771                         dentry_add_child(parent, dentry);
 772                 }
 773                 parent = dentry;
 774                 target_path = slash;
 775                 /* Skip over slashes.  Note: this cannot overrun the length of
 776                  * the string because the last character cannot be a slash, as
 777                  * trailing slashes were tripped.  */
 778                 do {
 779                         ++target_path;
 780                 } while (*target_path == T('/'));
 781         }
 782
 783         /* If the target path already existed, overlay the branch onto it.
 784          * Otherwise, set the branch as the target path. */
 785         target = get_dentry_child_with_utf16le_name(parent, branch->file_name,
 786                                                     branch->file_name_nbytes);
 787         if (target) {
 788                 return do_overlay(target, branch);
 789         } else {
 790                 dentry_add_child(parent, branch);
 791                 return 0;
 792         }
 793 }
 794
 795 static int
 796 canonicalize_pat(tchar **pat_p)
 797 {
 798         tchar *pat = *pat_p;
 799
 800         /* Turn all backslashes in the pattern into forward slashes. */
 801         zap_backslashes(pat);
 802
 803         if (*pat != T('/') && *pat != T('\0') && *(pat + 1) == T(':')) {
 804                 /* Pattern begins with drive letter */
 805                 if (*(pat + 2) != T('/')) {
 806                         /* Something like c:file, which is actually a path
 807                          * relative to the current working directory on the c:
 808                          * drive.  We require paths with drive letters to be
 809                          * absolute. */
 810                         ERROR("Invalid path \"%"TS"\"; paths including drive letters "
 811                               "must be absolute!", pat);
 812                         ERROR("Maybe try \"%"TC":/%"TS"\"?",
 813                               *pat, pat + 2);
 814                         return WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 815                 }
 816
 817                 WARNING("Pattern \"%"TS"\" starts with a drive letter, which is "
 818                         "being removed.", pat);
 819                 /* Strip the drive letter */
 820                 pat += 2;
 821                 *pat_p = pat;
 822         }
 823         return 0;
 824 }
 825
 826 static int
 827 canonicalize_pat_list(struct wimlib_pattern_list *pat_list)
 828 {
 829         int ret = 0;
 830         for (size_t i = 0; i < pat_list->num_pats; i++) {
 831                 ret = canonicalize_pat(&pat_list->pats[i]);
 832                 if (ret)
 833                         break;
 834         }
 835         return ret;
 836 }
 837
 838 static int
 839 canonicalize_capture_config(struct wimlib_capture_config *config)
 840 {
 841         int ret = canonicalize_pat_list(&config->exclusion_pats);
 842         if (ret)
 843                 return ret;
 844         return canonicalize_pat_list(&config->exclusion_exception_pats);
 845 }
 846
 847 WIMLIBAPI int
 848 wimlib_add_image_multisource(WIMStruct *w,
 849                              struct wimlib_capture_source *sources,
 850                              size_t num_sources,
 851                              const tchar *name,
 852                              struct wimlib_capture_config *config,
 853                              int add_image_flags,
 854                              wimlib_progress_func_t progress_func)
 855 {
 856         int (*capture_tree)(struct wim_dentry **,
 857                             const tchar *,
 858                             struct wim_lookup_table *,
 859                             struct sd_set *,
 860                             const struct wimlib_capture_config *,
 861                             int,
 862                             wimlib_progress_func_t,
 863                             void *);
 864         void *extra_arg;
 865         struct wim_dentry *root_dentry;
 866         struct wim_dentry *branch;
 867         struct wim_security_data *sd;
 868         struct wim_image_metadata *imd;
 869         int ret;
 870         struct sd_set sd_set;
 871
 872         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 873 #ifdef WITH_NTFS_3G
 874                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
 875                         ERROR("Cannot dereference files when capturing directly from NTFS");
 876                         return WIMLIB_ERR_INVALID_PARAM;
 877                 }
 878                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 879                         ERROR("Capturing UNIX owner and mode not supported "
 880                               "when capturing directly from NTFS");
 881                         return WIMLIB_ERR_INVALID_PARAM;
 882                 }
 883                 capture_tree = build_dentry_tree_ntfs;
 884                 extra_arg = &w->ntfs_vol;
 885 #else
 886                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
 887                       "        cannot capture a WIM image directly from a NTFS volume!");
 888                 return WIMLIB_ERR_UNSUPPORTED;
 889 #endif
 890         } else {
 891         #ifdef __WIN32__
 892                 capture_tree = win32_build_dentry_tree;
 893         #else
 894                 capture_tree = unix_build_dentry_tree;
 895         #endif
 896                 extra_arg = NULL;
 897         }
 898
 899 #ifdef __WIN32__
 900         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 901                 ERROR("Capturing UNIX-specific data is not supported on Windows");
 902                 return WIMLIB_ERR_INVALID_PARAM;
 903         }
 904         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
 905                 ERROR("Dereferencing symbolic links is not supported on Windows");
 906                 return WIMLIB_ERR_INVALID_PARAM;
 907         }
 908 #endif
 909
 910         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 911                 add_image_flags |= WIMLIB_ADD_IMAGE_FLAG_EXCLUDE_VERBOSE;
 912
 913         if (!name || !*name) {
 914                 ERROR("Must specify a non-empty string for the image name");
 915                 return WIMLIB_ERR_INVALID_PARAM;
 916         }
 917
 918         if (w->hdr.total_parts != 1) {
 919                 ERROR("Cannot add an image to a split WIM");
 920                 return WIMLIB_ERR_SPLIT_UNSUPPORTED;
 921         }
 922
 923         if (wimlib_image_name_in_use(w, name)) {
 924                 ERROR("There is already an image named \"%"TS"\" in the WIM!",
 925                       name);
 926                 return WIMLIB_ERR_IMAGE_NAME_COLLISION;
 927         }
 928
 929         if (!config) {
 930                 DEBUG("Capture config not provided; using empty config");
 931                 config = alloca(sizeof(*config));
 932                 memset(config, 0, sizeof(*config));
 933         }
 934
 935         ret = canonicalize_capture_config(config);
 936         if (ret)
 937                 goto out;
 938
 939         DEBUG("Allocating security data");
 940         sd = CALLOC(1, sizeof(struct wim_security_data));
 941         if (!sd) {
 942                 ret = WIMLIB_ERR_NOMEM;
 943                 goto out;
 944         }
 945         sd->total_length = 8;
 946         sd->refcnt = 1;
 947
 948         sd_set.sd = sd;
 949         sd_set.rb_root.rb_node = NULL;
 950
 951         DEBUG("Using %zu capture sources", num_sources);
 952         canonicalize_sources_and_targets(sources, num_sources);
 953         sort_sources(sources, num_sources);
 954         ret = check_sorted_sources(sources, num_sources, add_image_flags);
 955         if (ret) {
 956                 ret = WIMLIB_ERR_INVALID_PARAM;
 957                 goto out_free_security_data;
 958         }
 959
 960         DEBUG("Building dentry tree.");
 961         root_dentry = NULL;
 962
 963         for (size_t i = 0; i < num_sources; i++) {
 964                 int flags;
 965                 union wimlib_progress_info progress;
 966
 967                 DEBUG("Building dentry tree for source %zu of %zu "
 968                       "(\"%"TS"\" => \"%"TS"\")", i + 1, num_sources,
 969                       sources[i].fs_source_path,
 970                       sources[i].wim_target_path);
 971                 if (progress_func) {
 972                         memset(&progress, 0, sizeof(progress));
 973                         progress.scan.source = sources[i].fs_source_path;
 974                         progress.scan.wim_target_path = sources[i].wim_target_path;
 975                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_BEGIN, &progress);
 976                 }
 977                 config->_prefix = sources[i].fs_source_path;
 978                 config->_prefix_num_tchars = tstrlen(sources[i].fs_source_path);
 979                 flags = add_image_flags | WIMLIB_ADD_IMAGE_FLAG_SOURCE;
 980                 if (!*sources[i].wim_target_path)
 981                         flags |= WIMLIB_ADD_IMAGE_FLAG_ROOT;
 982                 ret = (*capture_tree)(&branch,
 983                                       sources[i].fs_source_path,
 984                                       w->lookup_table,
 985                                       &sd_set,
 986                                       config,
 987                                       flags,
 988                                       progress_func, extra_arg);
 989                 if (ret) {
 990                         ERROR("Failed to build dentry tree for `%"TS"'",
 991                               sources[i].fs_source_path);
 992                         goto out_free_dentry_tree;
 993                 }
 994                 if (branch) {
 995                         /* Use the target name, not the source name, for
 996                          * the root of each branch from a capture
 997                          * source.  (This will also set the root dentry
 998                          * of the entire image to be unnamed.) */
 999                         ret = set_dentry_name(branch,
1000                                               path_basename(sources[i].wim_target_path));
1001                         if (ret)
1002                                 goto out_free_branch;
1003
1004                         ret = attach_branch(&root_dentry, branch,
1005                                             sources[i].wim_target_path);
1006                         if (ret)
1007                                 goto out_free_branch;
1008                 }
1009                 if (progress_func)
1010                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_END, &progress);
1011         }
1012
1013         if (root_dentry == NULL) {
1014                 ret = new_filler_directory(T(""), &root_dentry);
1015                 if (ret)
1016                         goto out_free_dentry_tree;
1017         }
1018
1019         DEBUG("Calculating full paths of dentries.");
1020         ret = for_dentry_in_tree(root_dentry, calculate_dentry_full_path, NULL);
1021         if (ret)
1022                 goto out_free_dentry_tree;
1023
1024         ret = add_new_dentry_tree(w, root_dentry, sd);
1025         if (ret)
1026                 goto out_free_dentry_tree;
1027
1028         imd = &w->image_metadata[w->hdr.image_count - 1];
1029
1030         ret = dentry_tree_fix_inodes(root_dentry, &imd->inode_list);
1031         if (ret)
1032                 goto out_destroy_imd;
1033
1034         DEBUG("Assigning hard link group IDs");
1035         assign_inode_numbers(&imd->inode_list);
1036
1037         ret = xml_add_image(w, name);
1038         if (ret)
1039                 goto out_destroy_imd;
1040
1041         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_BOOT)
1042                 wimlib_set_boot_idx(w, w->hdr.image_count);
1043         ret = 0;
1044         goto out_destroy_sd_set;
1045 out_destroy_imd:
1046         destroy_image_metadata(&w->image_metadata[w->hdr.image_count - 1],
1047                                w->lookup_table);
1048         w->hdr.image_count--;
1049         goto out_destroy_sd_set;
1050 out_free_branch:
1051         free_dentry_tree(branch, w->lookup_table);
1052 out_free_dentry_tree:
1053         free_dentry_tree(root_dentry, w->lookup_table);
1054 out_free_security_data:
1055         free_security_data(sd);
1056 out_destroy_sd_set:
1057         destroy_sd_set(&sd_set);
1058 out:
1059         return ret;
1060 }
1061
1062 WIMLIBAPI int
1063 wimlib_add_image(WIMStruct *w,
1064                  const tchar *source,
1065                  const tchar *name,
1066                  struct wimlib_capture_config *config,
1067                  int add_image_flags,
1068                  wimlib_progress_func_t progress_func)
1069 {
1070         if (!source || !*source)
1071                 return WIMLIB_ERR_INVALID_PARAM;
1072
1073         tchar *fs_source_path = TSTRDUP(source);
1074         int ret;
1075         struct wimlib_capture_source capture_src = {
1076                 .fs_source_path = fs_source_path,
1077                 .wim_target_path = NULL,
1078                 .reserved = 0,
1079         };
1080         ret = wimlib_add_image_multisource(w, &capture_src, 1, name,
1081                                            config, add_image_flags,
1082                                            progress_func);
1083         FREE(fs_source_path);
1084         return ret;
1085 }