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 <string.h>
  45
  46 #include <unistd.h>
  47
  48 #ifdef HAVE_ALLOCA_H
  49 #  include <alloca.h>
  50 #endif
  51
  52 /*
  53  * Adds the dentry tree and security data for a new image to the image metadata
  54  * array of the WIMStruct.
  55  */
  56 int
  57 add_new_dentry_tree(WIMStruct *w, struct wim_dentry *root_dentry,
  58                     struct wim_security_data *sd)
  59 {
  60         struct wim_lookup_table_entry *metadata_lte;
  61         struct wim_image_metadata *imd;
  62         struct wim_image_metadata *new_imd;
  63
  64         wimlib_assert(root_dentry != NULL);
  65
  66         DEBUG("Reallocating image metadata array for image_count = %u",
  67               w->hdr.image_count + 1);
  68         imd = CALLOC((w->hdr.image_count + 1), sizeof(struct wim_image_metadata));
  69
  70         if (!imd) {
  71                 ERROR("Failed to allocate memory for new image metadata array");
  72                 goto err;
  73         }
  74
  75         memcpy(imd, w->image_metadata,
  76                w->hdr.image_count * sizeof(struct wim_image_metadata));
  77
  78         metadata_lte = new_lookup_table_entry();
  79         if (!metadata_lte)
  80                 goto err_free_imd;
  81
  82         metadata_lte->resource_entry.flags = WIM_RESHDR_FLAG_METADATA;
  83         random_hash(metadata_lte->hash);
  84         lookup_table_insert(w->lookup_table, metadata_lte);
  85
  86         new_imd = &imd[w->hdr.image_count];
  87
  88         new_imd->root_dentry    = root_dentry;
  89         new_imd->metadata_lte   = metadata_lte;
  90         new_imd->security_data  = sd;
  91         new_imd->modified       = 1;
  92
  93         FREE(w->image_metadata);
  94         w->image_metadata = imd;
  95         w->hdr.image_count++;
  96         return 0;
  97 err_free_imd:
  98         FREE(imd);
  99 err:
 100         return WIMLIB_ERR_NOMEM;
 101
 102 }
 103
 104 #ifndef __WIN32__
 105 /*
 106  * unix_build_dentry_tree():
 107  *      Recursively builds a tree of WIM dentries from an on-disk directory
 108  *      tree (UNIX version; no NTFS-specific data is captured).
 109  *
 110  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 111  *              modified if successful.  Set to NULL if the file or directory was
 112  *              excluded from capture.
 113  *
 114  * @root_disk_path:  The path to the root of the directory tree on disk.
 115  *
 116  * @lookup_table: The lookup table for the WIM file.  For each file added to the
 117  *              dentry tree being built, an entry is added to the lookup table,
 118  *              unless an identical stream is already in the lookup table.
 119  *              These lookup table entries that are added point to the path of
 120  *              the file on disk.
 121  *
 122  * @sd_set:     Ignored.  (Security data only captured in NTFS mode.)
 123  *
 124  * @capture_config:
 125  *              Configuration for files to be excluded from capture.
 126  *
 127  * @add_flags:  Bitwise or of WIMLIB_ADD_IMAGE_FLAG_*
 128  *
 129  * @extra_arg:  Ignored
 130  *
 131  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 132  *              the files cannot be `stat'ed, or if any of the needed
 133  *              directories cannot be opened or read.  Failure to add the files
 134  *              to the WIM may still occur later when trying to actually read
 135  *              the on-disk files during a call to wimlib_write() or
 136  *              wimlib_overwrite().
 137  */
 138 static int
 139 unix_build_dentry_tree(struct wim_dentry **root_ret,
 140                        const char *root_disk_path,
 141                        struct wim_lookup_table *lookup_table,
 142                        struct sd_set *sd_set,
 143                        const struct capture_config *config,
 144                        int add_image_flags,
 145                        wimlib_progress_func_t progress_func,
 146                        void *extra_arg)
 147 {
 148         struct wim_dentry *root = NULL;
 149         int ret = 0;
 150         struct wim_inode *inode;
 151
 152         if (exclude_path(root_disk_path, config, true)) {
 153                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) {
 154                         ERROR("Cannot exclude the root directory from capture");
 155                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 156                         goto out;
 157                 }
 158                 if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 159                     && progress_func)
 160                 {
 161                         union wimlib_progress_info info;
 162                         info.scan.cur_path = root_disk_path;
 163                         info.scan.excluded = true;
 164                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 165                 }
 166                 goto out;
 167         }
 168
 169         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 170             && progress_func)
 171         {
 172                 union wimlib_progress_info info;
 173                 info.scan.cur_path = root_disk_path;
 174                 info.scan.excluded = false;
 175                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 176         }
 177
 178         /* UNIX version of capturing a directory tree */
 179         struct stat root_stbuf;
 180         int (*stat_fn)(const char *restrict, struct stat *restrict);
 181         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)
 182                 stat_fn = stat;
 183         else
 184                 stat_fn = lstat;
 185
 186         ret = (*stat_fn)(root_disk_path, &root_stbuf);
 187         if (ret != 0) {
 188                 ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 189                 goto out;
 190         }
 191
 192         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) &&
 193               !S_ISDIR(root_stbuf.st_mode))
 194         {
 195                 /* Do a dereference-stat in case the root is a symbolic link.
 196                  * This case is allowed, provided that the symbolic link points
 197                  * to a directory. */
 198                 ret = stat(root_disk_path, &root_stbuf);
 199                 if (ret != 0) {
 200                         ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 201                         ret = WIMLIB_ERR_STAT;
 202                         goto out;
 203                 }
 204                 if (!S_ISDIR(root_stbuf.st_mode)) {
 205                         ERROR("`%s' is not a directory", root_disk_path);
 206                         ret = WIMLIB_ERR_NOTDIR;
 207                         goto out;
 208                 }
 209         }
 210         if (!S_ISREG(root_stbuf.st_mode) && !S_ISDIR(root_stbuf.st_mode)
 211             && !S_ISLNK(root_stbuf.st_mode)) {
 212                 ERROR("`%s' is not a regular file, directory, or symbolic link.",
 213                       root_disk_path);
 214                 ret = WIMLIB_ERR_SPECIAL_FILE;
 215                 goto out;
 216         }
 217
 218         ret = new_dentry_with_timeless_inode(path_basename(root_disk_path),
 219                                              &root);
 220         if (ret)
 221                 goto out;
 222
 223         inode = root->d_inode;
 224
 225 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 226         inode->i_creation_time = timespec_to_wim_timestamp(root_stbuf.st_mtim);
 227         inode->i_last_write_time = timespec_to_wim_timestamp(root_stbuf.st_mtim);
 228         inode->i_last_access_time = timespec_to_wim_timestamp(root_stbuf.st_atim);
 229 #else
 230         inode->i_creation_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 231         inode->i_last_write_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 232         inode->i_last_access_time = unix_timestamp_to_wim(root_stbuf.st_atime);
 233 #endif
 234         /* Leave the inode number at 0 for directories. */
 235         if (!S_ISDIR(root_stbuf.st_mode)) {
 236                 if (sizeof(ino_t) >= 8)
 237                         inode->i_ino = (u64)root_stbuf.st_ino;
 238                 else
 239                         inode->i_ino = (u64)root_stbuf.st_ino |
 240                                            ((u64)root_stbuf.st_dev <<
 241                                                 ((sizeof(ino_t) * 8) & 63));
 242         }
 243         inode->i_resolved = 1;
 244         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 245                 ret = inode_set_unix_data(inode, root_stbuf.st_uid,
 246                                           root_stbuf.st_gid,
 247                                           root_stbuf.st_mode,
 248                                           lookup_table,
 249                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 250                 if (ret)
 251                         goto out;
 252         }
 253         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 254         if (S_ISREG(root_stbuf.st_mode)) { /* Archiving a regular file */
 255
 256                 struct wim_lookup_table_entry *lte;
 257                 u8 hash[SHA1_HASH_SIZE];
 258
 259                 inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 260
 261                 /* Empty files do not have to have a lookup table entry. */
 262                 if (root_stbuf.st_size == 0)
 263                         goto out;
 264
 265                 /* For each regular file, we must check to see if the file is in
 266                  * the lookup table already; if it is, we increment its refcnt;
 267                  * otherwise, we create a new lookup table entry and insert it.
 268                  * */
 269
 270                 ret = sha1sum(root_disk_path, hash);
 271                 if (ret != 0)
 272                         goto out;
 273
 274                 lte = __lookup_resource(lookup_table, hash);
 275                 if (lte) {
 276                         lte->refcnt++;
 277                         DEBUG("Add lte reference %u for `%s'", lte->refcnt,
 278                               root_disk_path);
 279                 } else {
 280                         char *file_on_disk = STRDUP(root_disk_path);
 281                         if (!file_on_disk) {
 282                                 ERROR("Failed to allocate memory for file path");
 283                                 ret = WIMLIB_ERR_NOMEM;
 284                                 goto out;
 285                         }
 286                         lte = new_lookup_table_entry();
 287                         if (!lte) {
 288                                 FREE(file_on_disk);
 289                                 ret = WIMLIB_ERR_NOMEM;
 290                                 goto out;
 291                         }
 292                         lte->file_on_disk = file_on_disk;
 293                         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 294                         lte->resource_entry.original_size = root_stbuf.st_size;
 295                         lte->resource_entry.size = root_stbuf.st_size;
 296                         copy_hash(lte->hash, hash);
 297                         lookup_table_insert(lookup_table, lte);
 298                 }
 299                 root->d_inode->i_lte = lte;
 300         } else if (S_ISDIR(root_stbuf.st_mode)) { /* Archiving a directory */
 301
 302                 inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 303
 304                 DIR *dir;
 305                 struct dirent entry, *result;
 306                 struct wim_dentry *child;
 307
 308                 dir = opendir(root_disk_path);
 309                 if (!dir) {
 310                         ERROR_WITH_ERRNO("Failed to open the directory `%s'",
 311                                          root_disk_path);
 312                         ret = WIMLIB_ERR_OPEN;
 313                         goto out;
 314                 }
 315
 316                 /* Buffer for names of files in directory. */
 317                 size_t len = strlen(root_disk_path);
 318                 char name[len + 1 + FILENAME_MAX + 1];
 319                 memcpy(name, root_disk_path, len);
 320                 name[len] = '/';
 321
 322                 /* Create a dentry for each entry in the directory on disk, and recurse
 323                  * to any subdirectories. */
 324                 while (1) {
 325                         errno = 0;
 326                         ret = readdir_r(dir, &entry, &result);
 327                         if (ret != 0) {
 328                                 ret = WIMLIB_ERR_READ;
 329                                 ERROR_WITH_ERRNO("Error reading the "
 330                                                  "directory `%s'",
 331                                                  root_disk_path);
 332                                 break;
 333                         }
 334                         if (result == NULL)
 335                                 break;
 336                         if (result->d_name[0] == '.' && (result->d_name[1] == '\0'
 337                               || (result->d_name[1] == '.' && result->d_name[2] == '\0')))
 338                                         continue;
 339                         strcpy(name + len + 1, result->d_name);
 340                         ret = unix_build_dentry_tree(&child, name,
 341                                                      lookup_table,
 342                                                      NULL, config,
 343                                                      add_image_flags,
 344                                                      progress_func, NULL);
 345                         if (ret != 0)
 346                                 break;
 347                         if (child)
 348                                 dentry_add_child(root, child);
 349                 }
 350                 closedir(dir);
 351         } else { /* Archiving a symbolic link */
 352                 inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 353                 inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 354
 355                 /* The idea here is to call readlink() to get the UNIX target of
 356                  * the symbolic link, then turn the target into a reparse point
 357                  * data buffer that contains a relative or absolute symbolic
 358                  * link (NOT a junction point or *full* path symbolic link with
 359                  * drive letter).
 360                  */
 361
 362                 char deref_name_buf[4096];
 363                 ssize_t deref_name_len;
 364
 365                 deref_name_len = readlink(root_disk_path, deref_name_buf,
 366                                           sizeof(deref_name_buf) - 1);
 367                 if (deref_name_len >= 0) {
 368                         deref_name_buf[deref_name_len] = '\0';
 369                         DEBUG("Read symlink `%s'", deref_name_buf);
 370                         ret = inode_set_symlink(root->d_inode, deref_name_buf,
 371                                                 lookup_table, NULL);
 372                         if (ret == 0) {
 373                                 /*
 374                                  * Unfortunately, Windows seems to have the
 375                                  * concept of "file" symbolic links as being
 376                                  * different from "directory" symbolic links...
 377                                  * so FILE_ATTRIBUTE_DIRECTORY needs to be set
 378                                  * on the symbolic link if the *target* of the
 379                                  * symbolic link is a directory.
 380                                  */
 381                                 struct stat stbuf;
 382                                 if (stat(root_disk_path, &stbuf) == 0 &&
 383                                     S_ISDIR(stbuf.st_mode))
 384                                 {
 385                                         inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 386                                 }
 387                         }
 388                 } else {
 389                         ERROR_WITH_ERRNO("Failed to read target of "
 390                                          "symbolic link `%s'", root_disk_path);
 391                         ret = WIMLIB_ERR_READLINK;
 392                 }
 393         }
 394 out:
 395         if (ret == 0)
 396                 *root_ret = root;
 397         else
 398                 free_dentry_tree(root, lookup_table);
 399         return ret;
 400 }
 401 #endif /* !__WIN32__ */
 402
 403 enum pattern_type {
 404         NONE = 0,
 405         EXCLUSION_LIST,
 406         EXCLUSION_EXCEPTION,
 407         COMPRESSION_EXCLUSION_LIST,
 408         ALIGNMENT_LIST,
 409 };
 410
 411 #define COMPAT_DEFAULT_CONFIG
 412
 413 /* Default capture configuration file when none is specified. */
 414 static const tchar *default_config =
 415 #ifdef COMPAT_DEFAULT_CONFIG /* XXX: This policy is being moved to library
 416                                 users.  The next ABI-incompatible library
 417                                 version will default to the empty string here. */
 418 T(
 419 "[ExclusionList]\n"
 420 "\\$ntfs.log\n"
 421 "\\hiberfil.sys\n"
 422 "\\pagefile.sys\n"
 423 "\\System Volume Information\n"
 424 "\\RECYCLER\n"
 425 "\\Windows\\CSC\n"
 426 "\n"
 427 "[CompressionExclusionList]\n"
 428 "*.mp3\n"
 429 "*.zip\n"
 430 "*.cab\n"
 431 "\\WINDOWS\\inf\\*.pnf\n"
 432 );
 433 #else
 434 T("");
 435 #endif
 436
 437 static void
 438 destroy_pattern_list(struct pattern_list *list)
 439 {
 440         FREE(list->pats);
 441 }
 442
 443 static void
 444 destroy_capture_config(struct capture_config *config)
 445 {
 446         destroy_pattern_list(&config->exclusion_list);
 447         destroy_pattern_list(&config->exclusion_exception);
 448         destroy_pattern_list(&config->compression_exclusion_list);
 449         destroy_pattern_list(&config->alignment_list);
 450         FREE(config->config_str);
 451         FREE(config->prefix);
 452         memset(config, 0, sizeof(*config));
 453 }
 454
 455 static int
 456 pattern_list_add_pattern(struct pattern_list *list, const tchar *pattern)
 457 {
 458         const tchar **pats;
 459         if (list->num_pats >= list->num_allocated_pats) {
 460                 pats = REALLOC(list->pats,
 461                                sizeof(list->pats[0]) * (list->num_allocated_pats + 8));
 462                 if (!pats)
 463                         return WIMLIB_ERR_NOMEM;
 464                 list->num_allocated_pats += 8;
 465                 list->pats = pats;
 466         }
 467         list->pats[list->num_pats++] = pattern;
 468         return 0;
 469 }
 470
 471 /* Parses the contents of the image capture configuration file and fills in a
 472  * `struct capture_config'. */
 473 static int
 474 init_capture_config(struct capture_config *config,
 475                     const tchar *_config_str,
 476                     size_t config_num_tchars)
 477 {
 478         tchar *config_str;
 479         tchar *p;
 480         tchar *eol;
 481         tchar *next_p;
 482         size_t num_tchars_remaining;
 483         enum pattern_type type = NONE;
 484         int ret;
 485         unsigned long line_no = 0;
 486
 487         DEBUG("config_num_tchars = %zu", config_num_tchars);
 488         num_tchars_remaining = config_num_tchars;
 489         memset(config, 0, sizeof(*config));
 490         config_str = TMALLOC(config_num_tchars);
 491         if (!config_str) {
 492                 ERROR("Could not duplicate capture config string");
 493                 return WIMLIB_ERR_NOMEM;
 494         }
 495
 496         tmemcpy(config_str, _config_str, config_num_tchars);
 497         next_p = config_str;
 498         config->config_str = config_str;
 499         while (num_tchars_remaining != 0) {
 500                 line_no++;
 501                 p = next_p;
 502                 eol = tmemchr(p, T('\n'), num_tchars_remaining);
 503                 if (!eol) {
 504                         ERROR("Expected end-of-line in capture config file on "
 505                               "line %lu", line_no);
 506                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 507                         goto out_destroy;
 508                 }
 509
 510                 next_p = eol + 1;
 511                 num_tchars_remaining -= (next_p - p);
 512                 if (eol == p)
 513                         continue;
 514
 515                 if (*(eol - 1) == T('\r'))
 516                         eol--;
 517                 *eol = T('\0');
 518
 519                 /* Translate backslash to forward slash */
 520                 for (tchar *pp = p; pp != eol; pp++)
 521                         if (*pp == T('\\'))
 522                                 *pp = T('/');
 523
 524                 /* Remove drive letter (UNIX only) */
 525         #ifndef __WIN32__
 526                 if (eol - p > 2 && istalpha(*p) && *(p + 1) == T(':'))
 527                         p += 2;
 528         #endif
 529
 530                 ret = 0;
 531                 if (!tstrcmp(p, T("[ExclusionList]")))
 532                         type = EXCLUSION_LIST;
 533                 else if (!tstrcmp(p, T("[ExclusionException]")))
 534                         type = EXCLUSION_EXCEPTION;
 535                 else if (!tstrcmp(p, T("[CompressionExclusionList]")))
 536                         type = COMPRESSION_EXCLUSION_LIST;
 537                 else if (!tstrcmp(p, T("[AlignmentList]")))
 538                         type = ALIGNMENT_LIST;
 539                 else if (p[0] == T('[') && tstrrchr(p, T(']'))) {
 540                         ERROR("Unknown capture configuration section \"%"TS"\"", p);
 541                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 542                 } else switch (type) {
 543                 case EXCLUSION_LIST:
 544                         DEBUG("Adding pattern \"%"TS"\" to exclusion list", p);
 545                         ret = pattern_list_add_pattern(&config->exclusion_list, p);
 546                         break;
 547                 case EXCLUSION_EXCEPTION:
 548                         DEBUG("Adding pattern \"%"TS"\" to exclusion exception list", p);
 549                         ret = pattern_list_add_pattern(&config->exclusion_exception, p);
 550                         break;
 551                 case COMPRESSION_EXCLUSION_LIST:
 552                         DEBUG("Adding pattern \"%"TS"\" to compression exclusion list", p);
 553                         ret = pattern_list_add_pattern(&config->compression_exclusion_list, p);
 554                         break;
 555                 case ALIGNMENT_LIST:
 556                         DEBUG("Adding pattern \"%"TS"\" to alignment list", p);
 557                         ret = pattern_list_add_pattern(&config->alignment_list, p);
 558                         break;
 559                 default:
 560                         ERROR("Line %lu of capture configuration is not "
 561                               "in a block (such as [ExclusionList])",
 562                               line_no);
 563                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 564                         break;
 565                 }
 566                 if (ret != 0)
 567                         goto out_destroy;
 568         }
 569         return 0;
 570 out_destroy:
 571         destroy_capture_config(config);
 572         return ret;
 573 }
 574
 575 static int capture_config_set_prefix(struct capture_config *config,
 576                                      const tchar *_prefix)
 577 {
 578         tchar *prefix = TSTRDUP(_prefix);
 579
 580         if (!prefix)
 581                 return WIMLIB_ERR_NOMEM;
 582         FREE(config->prefix);
 583         config->prefix = prefix;
 584         config->prefix_num_tchars = tstrlen(prefix);
 585         return 0;
 586 }
 587
 588 static bool match_pattern(const tchar *path,
 589                           const tchar *path_basename,
 590                           const struct pattern_list *list)
 591 {
 592         for (size_t i = 0; i < list->num_pats; i++) {
 593                 const tchar *pat = list->pats[i];
 594                 const tchar *string;
 595                 if (pat[0] == '/')
 596                         /* Absolute path from root of capture */
 597                         string = path;
 598                 else {
 599                         if (tstrchr(pat, T('/')))
 600                                 /* Relative path from root of capture */
 601                                 string = path + 1;
 602                         else
 603                                 /* A file name pattern */
 604                                 string = path_basename;
 605                 }
 606
 607                 /* Warning: on Windows native builds, fnmatch() calls the
 608                  * replacement function in win32.c. */
 609                 if (fnmatch(pat, string, FNM_PATHNAME
 610                                 #ifdef FNM_CASEFOLD
 611                                         | FNM_CASEFOLD
 612                                 #endif
 613                             ) == 0)
 614                 {
 615                         DEBUG("\"%"TS"\" matches the pattern \"%"TS"\"",
 616                               string, pat);
 617                         return true;
 618                 }
 619         }
 620         return false;
 621 }
 622
 623 /* Return true if the image capture configuration file indicates we should
 624  * exclude the filename @path from capture.
 625  *
 626  * If @exclude_prefix is %true, the part of the path up and including the name
 627  * of the directory being captured is not included in the path for matching
 628  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
 629  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
 630  * directory.
 631  */
 632 bool
 633 exclude_path(const tchar *path, const struct capture_config *config,
 634              bool exclude_prefix)
 635 {
 636         const tchar *basename = path_basename(path);
 637         if (exclude_prefix) {
 638                 wimlib_assert(tstrlen(path) >= config->prefix_num_tchars);
 639                 if (!tmemcmp(config->prefix, path, config->prefix_num_tchars) &&
 640                     path[config->prefix_num_tchars] == T('/'))
 641                 {
 642                         path += config->prefix_num_tchars;
 643                 }
 644         }
 645         return match_pattern(path, basename, &config->exclusion_list) &&
 646                 !match_pattern(path, basename, &config->exclusion_exception);
 647
 648 }
 649
 650 /* Strip leading and trailing forward slashes from a string.  Modifies it in
 651  * place and returns the stripped string. */
 652 static const tchar *
 653 canonicalize_target_path(tchar *target_path)
 654 {
 655         tchar *p;
 656         if (target_path == NULL)
 657                 return T("");
 658         for (;;) {
 659                 if (*target_path == T('\0'))
 660                         return target_path;
 661                 else if (*target_path == T('/'))
 662                         target_path++;
 663                 else
 664                         break;
 665         }
 666
 667         p = tstrchr(target_path, T('\0')) - 1;
 668         while (*p == T('/'))
 669                 *p-- = T('\0');
 670         return target_path;
 671 }
 672
 673 /* Strip leading and trailing slashes from the target paths */
 674 static void
 675 canonicalize_targets(struct wimlib_capture_source *sources, size_t num_sources)
 676 {
 677         while (num_sources--) {
 678                 DEBUG("Canonicalizing { source: \"%"TS"\", target=\"%"TS"\"}",
 679                       sources->fs_source_path,
 680                       sources->wim_target_path);
 681
 682                 /* The Windows API can handle forward slashes.  Just get rid of
 683                  * backslashes to avoid confusing other parts of the library
 684                  * code. */
 685                 zap_backslashes(sources->fs_source_path);
 686                 if (sources->wim_target_path)
 687                         zap_backslashes(sources->wim_target_path);
 688
 689                 sources->wim_target_path =
 690                         (tchar*)canonicalize_target_path(sources->wim_target_path);
 691                 DEBUG("Canonical target: \"%"TS"\"", sources->wim_target_path);
 692                 sources++;
 693         }
 694 }
 695
 696 static int
 697 capture_source_cmp(const void *p1, const void *p2)
 698 {
 699         const struct wimlib_capture_source *s1 = p1, *s2 = p2;
 700         return tstrcmp(s1->wim_target_path, s2->wim_target_path);
 701 }
 702
 703 /* Sorts the capture sources lexicographically by target path.  This occurs
 704  * after leading and trailing forward slashes are stripped.
 705  *
 706  * One purpose of this is to make sure that target paths that are inside other
 707  * target paths are added after the containing target paths. */
 708 static void
 709 sort_sources(struct wimlib_capture_source *sources, size_t num_sources)
 710 {
 711         qsort(sources, num_sources, sizeof(sources[0]), capture_source_cmp);
 712 }
 713
 714 static int
 715 check_sorted_sources(struct wimlib_capture_source *sources, size_t num_sources,
 716                      int add_image_flags)
 717 {
 718         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 719                 if (num_sources != 1) {
 720                         ERROR("Must specify exactly 1 capture source "
 721                               "(the NTFS volume) in NTFS mode!");
 722                         return WIMLIB_ERR_INVALID_PARAM;
 723                 }
 724                 if (sources[0].wim_target_path[0] != T('\0')) {
 725                         ERROR("In NTFS capture mode the target path inside "
 726                               "the image must be the root directory!");
 727                         return WIMLIB_ERR_INVALID_PARAM;
 728                 }
 729         } else if (num_sources != 0) {
 730                 /* This code is disabled because the current code
 731                  * unconditionally attempts to do overlays.  So, duplicate
 732                  * target paths are OK. */
 733         #if 0
 734                 if (num_sources > 1 && sources[0].wim_target_path[0] == '\0') {
 735                         ERROR("Cannot specify root target when using multiple "
 736                               "capture sources!");
 737                         return WIMLIB_ERR_INVALID_PARAM;
 738                 }
 739                 for (size_t i = 0; i < num_sources - 1; i++) {
 740                         size_t len = strlen(sources[i].wim_target_path);
 741                         size_t j = i + 1;
 742                         const char *target1 = sources[i].wim_target_path;
 743                         do {
 744                                 const char *target2 = sources[j].wim_target_path;
 745                                 DEBUG("target1=%s, target2=%s",
 746                                       target1,target2);
 747                                 if (strncmp(target1, target2, len) ||
 748                                     target2[len] > '/')
 749                                         break;
 750                                 if (target2[len] == '/') {
 751                                         ERROR("Invalid target `%s': is a prefix of `%s'",
 752                                               target1, target2);
 753                                         return WIMLIB_ERR_INVALID_PARAM;
 754                                 }
 755                                 if (target2[len] == '\0') {
 756                                         ERROR("Invalid target `%s': is a duplicate of `%s'",
 757                                               target1, target2);
 758                                         return WIMLIB_ERR_INVALID_PARAM;
 759                                 }
 760                         } while (++j != num_sources);
 761                 }
 762         #endif
 763         }
 764         return 0;
 765
 766 }
 767
 768 /* Creates a new directory to place in the WIM image.  This is to create parent
 769  * directories that are not part of any target as needed.  */
 770 static int
 771 new_filler_directory(const tchar *name, struct wim_dentry **dentry_ret)
 772 {
 773         int ret;
 774         struct wim_dentry *dentry;
 775
 776         DEBUG("Creating filler directory \"%"TS"\"", name);
 777         ret = new_dentry_with_inode(name, &dentry);
 778         if (ret == 0) {
 779                 /* Leave the inode number as 0 for now.  The final inode number
 780                  * will be assigned later by assign_inode_numbers(). */
 781                 dentry->d_inode->i_resolved = 1;
 782                 dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 783                 *dentry_ret = dentry;
 784         }
 785         return ret;
 786 }
 787
 788 /* Transfers the children of @branch to @target.  It is an error if @target is
 789  * not a directory or if both @branch and @target contain a child dentry with
 790  * the same name. */
 791 static int
 792 do_overlay(struct wim_dentry *target, struct wim_dentry *branch)
 793 {
 794         struct rb_root *rb_root;
 795
 796         DEBUG("Doing overlay \"%"WS"\" => \"%"WS"\"",
 797               branch->file_name, target->file_name);
 798
 799         if (!dentry_is_directory(target)) {
 800                 ERROR("Cannot overlay directory \"%"WS"\" "
 801                       "over non-directory", branch->file_name);
 802                 return WIMLIB_ERR_INVALID_OVERLAY;
 803         }
 804
 805         rb_root = &branch->d_inode->i_children;
 806         while (rb_root->rb_node) { /* While @branch has children... */
 807                 struct wim_dentry *child = rbnode_dentry(rb_root->rb_node);
 808                 /* Move @child to the directory @target */
 809                 unlink_dentry(child);
 810                 if (!dentry_add_child(target, child)) {
 811                         /* Revert the change to avoid leaking the directory tree
 812                          * rooted at @child */
 813                         dentry_add_child(branch, child);
 814                         ERROR("Overlay error: file \"%"WS"\" already exists "
 815                               "as a child of \"%"WS"\"",
 816                               child->file_name, target->file_name);
 817                         return WIMLIB_ERR_INVALID_OVERLAY;
 818                 }
 819         }
 820         free_dentry(branch);
 821         return 0;
 822
 823 }
 824
 825 /* Attach or overlay a branch onto the WIM image.
 826  *
 827  * @root_p:
 828  *      Pointer to the root of the WIM image, or pointer to NULL if it has not
 829  *      been created yet.
 830  * @branch
 831  *      Branch to add.
 832  * @target_path:
 833  *      Path in the WIM image to add the branch, with leading and trailing
 834  *      slashes stripped.
 835  */
 836 static int
 837 attach_branch(struct wim_dentry **root_p, struct wim_dentry *branch,
 838               tchar *target_path)
 839 {
 840         tchar *slash;
 841         struct wim_dentry *dentry, *parent, *target;
 842         int ret;
 843
 844         DEBUG("Attaching branch \"%"WS"\" => \"%"TS"\"",
 845               branch->file_name, target_path);
 846
 847         if (*target_path == T('\0')) {
 848                 /* Target: root directory */
 849                 if (*root_p) {
 850                         /* Overlay on existing root */
 851                         return do_overlay(*root_p, branch);
 852                 } else  {
 853                         /* Set as root */
 854                         *root_p = branch;
 855                         return 0;
 856                 }
 857         }
 858
 859         /* Adding a non-root branch.  Create root if it hasn't been created
 860          * already. */
 861         if (!*root_p) {
 862                 ret  = new_filler_directory(T(""), root_p);
 863                 if (ret)
 864                         return ret;
 865         }
 866
 867         /* Walk the path to the branch, creating filler directories as needed.
 868          * */
 869         parent = *root_p;
 870         while ((slash = tstrchr(target_path, T('/')))) {
 871                 *slash = T('\0');
 872                 dentry = get_dentry_child_with_name(parent, target_path);
 873                 if (!dentry) {
 874                         ret = new_filler_directory(target_path, &dentry);
 875                         if (ret)
 876                                 return ret;
 877                         dentry_add_child(parent, dentry);
 878                 }
 879                 parent = dentry;
 880                 target_path = slash;
 881                 /* Skip over slashes.  Note: this cannot overrun the length of
 882                  * the string because the last character cannot be a slash, as
 883                  * trailing slashes were tripped.  */
 884                 do {
 885                         ++target_path;
 886                 } while (*target_path == T('/'));
 887         }
 888
 889         /* If the target path already existed, overlay the branch onto it.
 890          * Otherwise, set the branch as the target path. */
 891         target = get_dentry_child_with_utf16le_name(parent, branch->file_name,
 892                                                     branch->file_name_nbytes);
 893         if (target) {
 894                 return do_overlay(target, branch);
 895         } else {
 896                 dentry_add_child(parent, branch);
 897                 return 0;
 898         }
 899 }
 900
 901 WIMLIBAPI int
 902 wimlib_add_image_multisource(WIMStruct *w,
 903                              struct wimlib_capture_source *sources,
 904                              size_t num_sources,
 905                              const tchar *name,
 906                              const tchar *config_str,
 907                              size_t config_len,
 908                              int add_image_flags,
 909                              wimlib_progress_func_t progress_func)
 910 {
 911         int (*capture_tree)(struct wim_dentry **,
 912                             const tchar *,
 913                             struct wim_lookup_table *,
 914                             struct sd_set *,
 915                             const struct capture_config *,
 916                             int,
 917                             wimlib_progress_func_t,
 918                             void *);
 919         void *extra_arg;
 920         struct wim_dentry *root_dentry;
 921         struct wim_dentry *branch;
 922         struct wim_security_data *sd;
 923         struct capture_config config;
 924         struct wim_image_metadata *imd;
 925         int ret;
 926         struct sd_set sd_set;
 927
 928         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 929 #ifdef WITH_NTFS_3G
 930                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
 931                         ERROR("Cannot dereference files when capturing directly from NTFS");
 932                         return WIMLIB_ERR_INVALID_PARAM;
 933                 }
 934                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 935                         ERROR("Capturing UNIX owner and mode not supported "
 936                               "when capturing directly from NTFS");
 937                         return WIMLIB_ERR_INVALID_PARAM;
 938                 }
 939                 capture_tree = build_dentry_tree_ntfs;
 940                 extra_arg = &w->ntfs_vol;
 941 #else
 942                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
 943                       "        cannot capture a WIM image directly from a NTFS volume!");
 944                 return WIMLIB_ERR_UNSUPPORTED;
 945 #endif
 946         } else {
 947         #ifdef __WIN32__
 948                 capture_tree = win32_build_dentry_tree;
 949         #else
 950                 capture_tree = unix_build_dentry_tree;
 951         #endif
 952                 extra_arg = NULL;
 953         }
 954
 955 #ifdef __WIN32__
 956         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 957                 ERROR("Capturing UNIX-specific data is not supported on Windows");
 958                 return WIMLIB_ERR_INVALID_PARAM;
 959         }
 960         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
 961                 ERROR("Dereferencing symbolic links is not supported on Windows");
 962                 return WIMLIB_ERR_INVALID_PARAM;
 963         }
 964 #endif
 965
 966         if (!name || !*name) {
 967                 ERROR("Must specify a non-empty string for the image name");
 968                 return WIMLIB_ERR_INVALID_PARAM;
 969         }
 970
 971         if (w->hdr.total_parts != 1) {
 972                 ERROR("Cannot add an image to a split WIM");
 973                 return WIMLIB_ERR_SPLIT_UNSUPPORTED;
 974         }
 975
 976         if (wimlib_image_name_in_use(w, name)) {
 977                 ERROR("There is already an image named \"%"TS"\" in the WIM!",
 978                       name);
 979                 return WIMLIB_ERR_IMAGE_NAME_COLLISION;
 980         }
 981
 982         if (!config_str) {
 983                 DEBUG("Using default capture configuration");
 984                 config_str = default_config;
 985                 config_len = tstrlen(default_config);
 986         }
 987         ret = init_capture_config(&config, config_str, config_len);
 988         if (ret)
 989                 goto out;
 990
 991         DEBUG("Allocating security data");
 992         sd = CALLOC(1, sizeof(struct wim_security_data));
 993         if (!sd) {
 994                 ret = WIMLIB_ERR_NOMEM;
 995                 goto out_destroy_capture_config;
 996         }
 997         sd->total_length = 8;
 998         sd->refcnt = 1;
 999
1000         sd_set.sd = sd;
1001         sd_set.rb_root.rb_node = NULL;
1002
1003         DEBUG("Using %zu capture sources", num_sources);
1004         canonicalize_targets(sources, num_sources);
1005         sort_sources(sources, num_sources);
1006         ret = check_sorted_sources(sources, num_sources, add_image_flags);
1007         if (ret) {
1008                 ret = WIMLIB_ERR_INVALID_PARAM;
1009                 goto out_free_security_data;
1010         }
1011
1012         DEBUG("Building dentry tree.");
1013         root_dentry = NULL;
1014
1015         for (size_t i = 0; i < num_sources; i++) {
1016                 int flags;
1017                 union wimlib_progress_info progress;
1018
1019                 DEBUG("Building dentry tree for source %zu of %zu "
1020                       "(\"%"TS"\" => \"%"TS"\")", i + 1, num_sources,
1021                       sources[i].fs_source_path,
1022                       sources[i].wim_target_path);
1023                 if (progress_func) {
1024                         memset(&progress, 0, sizeof(progress));
1025                         progress.scan.source = sources[i].fs_source_path;
1026                         progress.scan.wim_target_path = sources[i].wim_target_path;
1027                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_BEGIN, &progress);
1028                 }
1029                 ret = capture_config_set_prefix(&config,
1030                                                 sources[i].fs_source_path);
1031                 if (ret)
1032                         goto out_free_dentry_tree;
1033                 flags = add_image_flags | WIMLIB_ADD_IMAGE_FLAG_SOURCE;
1034                 if (!*sources[i].wim_target_path)
1035                         flags |= WIMLIB_ADD_IMAGE_FLAG_ROOT;
1036                 ret = (*capture_tree)(&branch,
1037                                       sources[i].fs_source_path,
1038                                       w->lookup_table,
1039                                       &sd_set,
1040                                       &config,
1041                                       flags,
1042                                       progress_func, extra_arg);
1043                 if (ret) {
1044                         ERROR("Failed to build dentry tree for `%"TS"'",
1045                               sources[i].fs_source_path);
1046                         goto out_free_dentry_tree;
1047                 }
1048                 if (branch) {
1049                         /* Use the target name, not the source name, for
1050                          * the root of each branch from a capture
1051                          * source.  (This will also set the root dentry
1052                          * of the entire image to be unnamed.) */
1053                         ret = set_dentry_name(branch,
1054                                               path_basename(sources[i].wim_target_path));
1055                         if (ret)
1056                                 goto out_free_branch;
1057
1058                         ret = attach_branch(&root_dentry, branch,
1059                                             sources[i].wim_target_path);
1060                         if (ret)
1061                                 goto out_free_branch;
1062                 }
1063                 if (progress_func)
1064                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_END, &progress);
1065         }
1066
1067         if (root_dentry == NULL) {
1068                 ret = new_filler_directory(T(""), &root_dentry);
1069                 if (ret)
1070                         goto out_free_dentry_tree;
1071         }
1072
1073         DEBUG("Calculating full paths of dentries.");
1074         ret = for_dentry_in_tree(root_dentry, calculate_dentry_full_path, NULL);
1075         if (ret)
1076                 goto out_free_dentry_tree;
1077
1078         ret = add_new_dentry_tree(w, root_dentry, sd);
1079         if (ret)
1080                 goto out_free_dentry_tree;
1081
1082         imd = &w->image_metadata[w->hdr.image_count - 1];
1083
1084         ret = dentry_tree_fix_inodes(root_dentry, &imd->inode_list);
1085         if (ret)
1086                 goto out_destroy_imd;
1087
1088         DEBUG("Assigning hard link group IDs");
1089         assign_inode_numbers(&imd->inode_list);
1090
1091         ret = xml_add_image(w, name);
1092         if (ret)
1093                 goto out_destroy_imd;
1094
1095         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_BOOT)
1096                 wimlib_set_boot_idx(w, w->hdr.image_count);
1097         ret = 0;
1098         goto out_destroy_sd_set;
1099 out_destroy_imd:
1100         destroy_image_metadata(&w->image_metadata[w->hdr.image_count - 1],
1101                                w->lookup_table);
1102         w->hdr.image_count--;
1103         goto out_destroy_sd_set;
1104 out_free_branch:
1105         free_dentry_tree(branch, w->lookup_table);
1106 out_free_dentry_tree:
1107         free_dentry_tree(root_dentry, w->lookup_table);
1108 out_free_security_data:
1109         free_security_data(sd);
1110 out_destroy_sd_set:
1111         destroy_sd_set(&sd_set);
1112 out_destroy_capture_config:
1113         destroy_capture_config(&config);
1114 out:
1115         return ret;
1116 }
1117
1118 WIMLIBAPI int
1119 wimlib_add_image(WIMStruct *w,
1120                  const tchar *source,
1121                  const tchar *name,
1122                  const tchar *config_str,
1123                  size_t config_len,
1124                  int add_image_flags,
1125                  wimlib_progress_func_t progress_func)
1126 {
1127         if (!source || !*source)
1128                 return WIMLIB_ERR_INVALID_PARAM;
1129
1130         tchar *fs_source_path = TSTRDUP(source);
1131         int ret;
1132         struct wimlib_capture_source capture_src = {
1133                 .fs_source_path = fs_source_path,
1134                 .wim_target_path = NULL,
1135                 .reserved = 0,
1136         };
1137         ret = wimlib_add_image_multisource(w, &capture_src, 1, name,
1138                                            config_str, config_len,
1139                                            add_image_flags, progress_func);
1140         FREE(fs_source_path);
1141         return ret;
1142 }