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 "wimlib_internal.h"
  25 #include "dentry.h"
  26 #include "timestamp.h"
  27 #include "lookup_table.h"
  28 #include "xml.h"
  29 #include <string.h>
  30 #include <fnmatch.h>
  31 #include <stdlib.h>
  32 #include <ctype.h>
  33 #include <sys/stat.h>
  34 #include <dirent.h>
  35 #include <errno.h>
  36 #include <unistd.h>
  37
  38 #define WIMLIB_ADD_IMAGE_FLAG_ROOT      0x80000000
  39 #define WIMLIB_ADD_IMAGE_FLAG_SOURCE    0x40000000
  40
  41 /*
  42  * Adds the dentry tree and security data for a new image to the image metadata
  43  * array of the WIMStruct.
  44  */
  45 int add_new_dentry_tree(WIMStruct *w, struct wim_dentry *root_dentry,
  46                         struct wim_security_data *sd)
  47 {
  48         struct wim_lookup_table_entry *metadata_lte;
  49         struct wim_image_metadata *imd;
  50         struct wim_image_metadata *new_imd;
  51
  52         wimlib_assert(root_dentry != NULL);
  53
  54         DEBUG("Reallocating image metadata array for image_count = %u",
  55               w->hdr.image_count + 1);
  56         imd = CALLOC((w->hdr.image_count + 1), sizeof(struct wim_image_metadata));
  57
  58         if (!imd) {
  59                 ERROR("Failed to allocate memory for new image metadata array");
  60                 goto err;
  61         }
  62
  63         memcpy(imd, w->image_metadata,
  64                w->hdr.image_count * sizeof(struct wim_image_metadata));
  65
  66         metadata_lte = new_lookup_table_entry();
  67         if (!metadata_lte)
  68                 goto err_free_imd;
  69
  70         metadata_lte->resource_entry.flags = WIM_RESHDR_FLAG_METADATA;
  71         random_hash(metadata_lte->hash);
  72         lookup_table_insert(w->lookup_table, metadata_lte);
  73
  74         new_imd = &imd[w->hdr.image_count];
  75
  76         new_imd->root_dentry    = root_dentry;
  77         new_imd->metadata_lte   = metadata_lte;
  78         new_imd->security_data  = sd;
  79         new_imd->modified       = 1;
  80
  81         FREE(w->image_metadata);
  82         w->image_metadata = imd;
  83         w->hdr.image_count++;
  84         return 0;
  85 err_free_imd:
  86         FREE(imd);
  87 err:
  88         return WIMLIB_ERR_NOMEM;
  89
  90 }
  91
  92
  93 /*
  94  * build_dentry_tree():
  95  *      Recursively builds a tree of WIM dentries from an on-disk directory
  96  *      tree.
  97  *
  98  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
  99  *              modified if successful.  Set to NULL if the file or directory was
 100  *              excluded from capture.
 101  *
 102  * @root_disk_path:  The path to the root of the directory tree on disk.
 103  *
 104  * @lookup_table: The lookup table for the WIM file.  For each file added to the
 105  *              dentry tree being built, an entry is added to the lookup table,
 106  *              unless an identical stream is already in the lookup table.
 107  *              These lookup table entries that are added point to the path of
 108  *              the file on disk.
 109  *
 110  * @sd:         Ignored.  (Security data only captured in NTFS mode.)
 111  *
 112  * @capture_config:
 113  *              Configuration for files to be excluded from capture.
 114  *
 115  * @add_flags:  Bitwise or of WIMLIB_ADD_IMAGE_FLAG_*
 116  *
 117  * @extra_arg:  Ignored. (Only used in NTFS mode.)
 118  *
 119  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 120  *              the files cannot be `stat'ed, or if any of the needed
 121  *              directories cannot be opened or read.  Failure to add the files
 122  *              to the WIM may still occur later when trying to actually read
 123  *              the on-disk files during a call to wimlib_write() or
 124  *              wimlib_overwrite().
 125  */
 126 static int build_dentry_tree(struct wim_dentry **root_ret,
 127                              const char *root_disk_path,
 128                              struct wim_lookup_table *lookup_table,
 129                              struct wim_security_data *sd,
 130                              const struct capture_config *config,
 131                              int add_image_flags,
 132                              wimlib_progress_func_t progress_func,
 133                              void *extra_arg)
 134 {
 135         struct stat root_stbuf;
 136         int ret = 0;
 137         int (*stat_fn)(const char *restrict, struct stat *restrict);
 138         struct wim_dentry *root;
 139         struct wim_inode *inode;
 140
 141         if (exclude_path(root_disk_path, config, true)) {
 142                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) {
 143                         ERROR("Cannot exclude the root directory from capture");
 144                         return WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 145                 }
 146                 if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 147                     && progress_func)
 148                 {
 149                         union wimlib_progress_info info;
 150                         info.scan.cur_path = root_disk_path;
 151                         info.scan.excluded = true;
 152                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 153                 }
 154                 *root_ret = NULL;
 155                 return 0;
 156         }
 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 = false;
 164                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 165         }
 166
 167         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)
 168                 stat_fn = stat;
 169         else
 170                 stat_fn = lstat;
 171
 172         ret = (*stat_fn)(root_disk_path, &root_stbuf);
 173         if (ret != 0) {
 174                 ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 175                 return WIMLIB_ERR_STAT;
 176         }
 177
 178         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) &&
 179               !S_ISDIR(root_stbuf.st_mode))
 180         {
 181                 /* Do a dereference-stat in case the root is a symbolic link.
 182                  * This case is allowed, provided that the symbolic link points
 183                  * to a directory. */
 184                 ret = stat(root_disk_path, &root_stbuf);
 185                 if (ret != 0) {
 186                         ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 187                         return WIMLIB_ERR_STAT;
 188                 }
 189                 if (!S_ISDIR(root_stbuf.st_mode)) {
 190                         ERROR("`%s' is not a directory", root_disk_path);
 191                         return WIMLIB_ERR_NOTDIR;
 192                 }
 193         }
 194         if (!S_ISREG(root_stbuf.st_mode) && !S_ISDIR(root_stbuf.st_mode)
 195             && !S_ISLNK(root_stbuf.st_mode)) {
 196                 ERROR("`%s' is not a regular file, directory, or symbolic link.",
 197                       root_disk_path);
 198                 return WIMLIB_ERR_SPECIAL_FILE;
 199         }
 200
 201         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 202         if (!root) {
 203                 if (errno == EILSEQ)
 204                         return WIMLIB_ERR_INVALID_UTF8_STRING;
 205                 else if (errno == ENOMEM)
 206                         return WIMLIB_ERR_NOMEM;
 207                 else
 208                         return WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 209         }
 210
 211         inode = root->d_inode;
 212
 213 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 214         inode->i_creation_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 215         inode->i_last_write_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 216         inode->i_last_access_time = timespec_to_wim_timestamp(&root_stbuf.st_atim);
 217 #else
 218         inode->i_creation_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 219         inode->i_last_write_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 220         inode->i_last_access_time = unix_timestamp_to_wim(root_stbuf.st_atime);
 221 #endif
 222         if (sizeof(ino_t) >= 8)
 223                 inode->i_ino = (u64)root_stbuf.st_ino;
 224         else
 225                 inode->i_ino = (u64)root_stbuf.st_ino |
 226                                    ((u64)root_stbuf.st_dev << ((sizeof(ino_t) * 8) & 63));
 227         inode->i_resolved = 1;
 228         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 229                 ret = inode_set_unix_data(inode, root_stbuf.st_uid,
 230                                           root_stbuf.st_gid,
 231                                           root_stbuf.st_mode,
 232                                           lookup_table,
 233                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 234                 if (ret)
 235                         goto out;
 236         }
 237         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 238         if (S_ISREG(root_stbuf.st_mode)) { /* Archiving a regular file */
 239
 240                 struct wim_lookup_table_entry *lte;
 241                 u8 hash[SHA1_HASH_SIZE];
 242
 243                 inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 244
 245                 /* Empty files do not have to have a lookup table entry. */
 246                 if (root_stbuf.st_size == 0)
 247                         goto out;
 248
 249                 /* For each regular file, we must check to see if the file is in
 250                  * the lookup table already; if it is, we increment its refcnt;
 251                  * otherwise, we create a new lookup table entry and insert it.
 252                  * */
 253
 254                 ret = sha1sum(root_disk_path, hash);
 255                 if (ret != 0)
 256                         goto out;
 257
 258                 lte = __lookup_resource(lookup_table, hash);
 259                 if (lte) {
 260                         lte->refcnt++;
 261                         DEBUG("Add lte reference %u for `%s'", lte->refcnt,
 262                               root_disk_path);
 263                 } else {
 264                         char *file_on_disk = STRDUP(root_disk_path);
 265                         if (!file_on_disk) {
 266                                 ERROR("Failed to allocate memory for file path");
 267                                 ret = WIMLIB_ERR_NOMEM;
 268                                 goto out;
 269                         }
 270                         lte = new_lookup_table_entry();
 271                         if (!lte) {
 272                                 FREE(file_on_disk);
 273                                 ret = WIMLIB_ERR_NOMEM;
 274                                 goto out;
 275                         }
 276                         lte->file_on_disk = file_on_disk;
 277                         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 278                         lte->resource_entry.original_size = root_stbuf.st_size;
 279                         lte->resource_entry.size = root_stbuf.st_size;
 280                         copy_hash(lte->hash, hash);
 281                         lookup_table_insert(lookup_table, lte);
 282                 }
 283                 root->d_inode->i_lte = lte;
 284         } else if (S_ISDIR(root_stbuf.st_mode)) { /* Archiving a directory */
 285
 286                 inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 287
 288                 DIR *dir;
 289                 struct dirent entry, *result;
 290                 struct wim_dentry *child;
 291
 292                 dir = opendir(root_disk_path);
 293                 if (!dir) {
 294                         ERROR_WITH_ERRNO("Failed to open the directory `%s'",
 295                                          root_disk_path);
 296                         ret = WIMLIB_ERR_OPEN;
 297                         goto out;
 298                 }
 299
 300                 /* Buffer for names of files in directory. */
 301                 size_t len = strlen(root_disk_path);
 302                 char name[len + 1 + FILENAME_MAX + 1];
 303                 memcpy(name, root_disk_path, len);
 304                 name[len] = '/';
 305
 306                 /* Create a dentry for each entry in the directory on disk, and recurse
 307                  * to any subdirectories. */
 308                 while (1) {
 309                         errno = 0;
 310                         ret = readdir_r(dir, &entry, &result);
 311                         if (ret != 0) {
 312                                 ret = WIMLIB_ERR_READ;
 313                                 ERROR_WITH_ERRNO("Error reading the "
 314                                                  "directory `%s'",
 315                                                  root_disk_path);
 316                                 break;
 317                         }
 318                         if (result == NULL)
 319                                 break;
 320                         if (result->d_name[0] == '.' && (result->d_name[1] == '\0'
 321                               || (result->d_name[1] == '.' && result->d_name[2] == '\0')))
 322                                         continue;
 323                         strcpy(name + len + 1, result->d_name);
 324                         ret = build_dentry_tree(&child, name, lookup_table,
 325                                                 NULL, config, add_image_flags,
 326                                                 progress_func, NULL);
 327                         if (ret != 0)
 328                                 break;
 329                         if (child)
 330                                 dentry_add_child(root, child);
 331                 }
 332                 closedir(dir);
 333         } else { /* Archiving a symbolic link */
 334                 inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 335                 inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 336
 337                 /* The idea here is to call readlink() to get the UNIX target of
 338                  * the symbolic link, then turn the target into a reparse point
 339                  * data buffer that contains a relative or absolute symbolic
 340                  * link (NOT a junction point or *full* path symbolic link with
 341                  * drive letter).
 342                  */
 343
 344                 char deref_name_buf[4096];
 345                 ssize_t deref_name_len;
 346
 347                 deref_name_len = readlink(root_disk_path, deref_name_buf,
 348                                           sizeof(deref_name_buf) - 1);
 349                 if (deref_name_len >= 0) {
 350                         deref_name_buf[deref_name_len] = '\0';
 351                         DEBUG("Read symlink `%s'", deref_name_buf);
 352                         ret = inode_set_symlink(root->d_inode, deref_name_buf,
 353                                                 lookup_table, NULL);
 354                         if (ret == 0) {
 355                                 /*
 356                                  * Unfortunately, Windows seems to have the
 357                                  * concept of "file" symbolic links as being
 358                                  * different from "directory" symbolic links...
 359                                  * so FILE_ATTRIBUTE_DIRECTORY needs to be set
 360                                  * on the symbolic link if the *target* of the
 361                                  * symbolic link is a directory.
 362                                  */
 363                                 struct stat stbuf;
 364                                 if (stat(root_disk_path, &stbuf) == 0 &&
 365                                     S_ISDIR(stbuf.st_mode))
 366                                 {
 367                                         inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 368                                 }
 369                         }
 370                 } else {
 371                         ERROR_WITH_ERRNO("Failed to read target of "
 372                                          "symbolic link `%s'", root_disk_path);
 373                         ret = WIMLIB_ERR_READLINK;
 374                 }
 375         }
 376 out:
 377         if (ret == 0)
 378                 *root_ret = root;
 379         else
 380                 free_dentry_tree(root, lookup_table);
 381         return ret;
 382 }
 383
 384
 385 enum pattern_type {
 386         NONE = 0,
 387         EXCLUSION_LIST,
 388         EXCLUSION_EXCEPTION,
 389         COMPRESSION_EXCLUSION_LIST,
 390         ALIGNMENT_LIST,
 391 };
 392
 393 #define COMPAT_DEFAULT_CONFIG
 394
 395 /* Default capture configuration file when none is specified. */
 396 static const char *default_config =
 397 #ifdef COMPAT_DEFAULT_CONFIG /* XXX: This policy is being moved to library
 398                                 users.  The next ABI-incompatible library
 399                                 version will default to the empty string here. */
 400 "[ExclusionList]\n"
 401 "\\$ntfs.log\n"
 402 "\\hiberfil.sys\n"
 403 "\\pagefile.sys\n"
 404 "\\System Volume Information\n"
 405 "\\RECYCLER\n"
 406 "\\Windows\\CSC\n"
 407 "\n"
 408 "[CompressionExclusionList]\n"
 409 "*.mp3\n"
 410 "*.zip\n"
 411 "*.cab\n"
 412 "\\WINDOWS\\inf\\*.pnf\n";
 413 #else
 414 "";
 415 #endif
 416
 417 static void destroy_pattern_list(struct pattern_list *list)
 418 {
 419         FREE(list->pats);
 420 }
 421
 422 static void destroy_capture_config(struct capture_config *config)
 423 {
 424         destroy_pattern_list(&config->exclusion_list);
 425         destroy_pattern_list(&config->exclusion_exception);
 426         destroy_pattern_list(&config->compression_exclusion_list);
 427         destroy_pattern_list(&config->alignment_list);
 428         FREE(config->config_str);
 429         FREE(config->prefix);
 430         memset(config, 0, sizeof(*config));
 431 }
 432
 433 static int pattern_list_add_pattern(struct pattern_list *list,
 434                                     const char *pattern)
 435 {
 436         const char **pats;
 437         if (list->num_pats >= list->num_allocated_pats) {
 438                 pats = REALLOC(list->pats,
 439                                sizeof(list->pats[0]) * (list->num_allocated_pats + 8));
 440                 if (!pats)
 441                         return WIMLIB_ERR_NOMEM;
 442                 list->num_allocated_pats += 8;
 443                 list->pats = pats;
 444         }
 445         list->pats[list->num_pats++] = pattern;
 446         return 0;
 447 }
 448
 449 /* Parses the contents of the image capture configuration file and fills in a
 450  * `struct capture_config'. */
 451 static int init_capture_config(const char *_config_str, size_t config_len,
 452                                const char *_prefix, struct capture_config *config)
 453 {
 454         char *config_str;
 455         char *prefix;
 456         char *p;
 457         char *eol;
 458         char *next_p;
 459         size_t bytes_remaining;
 460         enum pattern_type type = NONE;
 461         int ret;
 462         unsigned long line_no = 0;
 463
 464         DEBUG("config_len = %zu", config_len);
 465         bytes_remaining = config_len;
 466         memset(config, 0, sizeof(*config));
 467         config_str = MALLOC(config_len);
 468         if (!config_str) {
 469                 ERROR("Could not duplicate capture config string");
 470                 return WIMLIB_ERR_NOMEM;
 471         }
 472         prefix = STRDUP(_prefix);
 473         if (!prefix) {
 474                 FREE(config_str);
 475                 return WIMLIB_ERR_NOMEM;
 476         }
 477
 478         memcpy(config_str, _config_str, config_len);
 479         next_p = config_str;
 480         config->config_str = config_str;
 481         config->prefix = prefix;
 482         config->prefix_len = strlen(prefix);
 483         while (bytes_remaining) {
 484                 line_no++;
 485                 p = next_p;
 486                 eol = memchr(p, '\n', bytes_remaining);
 487                 if (!eol) {
 488                         ERROR("Expected end-of-line in capture config file on "
 489                               "line %lu", line_no);
 490                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 491                         goto out_destroy;
 492                 }
 493
 494                 next_p = eol + 1;
 495                 bytes_remaining -= (next_p - p);
 496                 if (eol == p)
 497                         continue;
 498
 499                 if (*(eol - 1) == '\r')
 500                         eol--;
 501                 *eol = '\0';
 502
 503                 /* Translate backslash to forward slash */
 504                 for (char *pp = p; pp != eol; pp++)
 505                         if (*pp == '\\')
 506                                 *pp = '/';
 507
 508                 /* Remove drive letter */
 509                 if (eol - p > 2 && isalpha(*p) && *(p + 1) == ':')
 510                         p += 2;
 511
 512                 ret = 0;
 513                 if (strcmp(p, "[ExclusionList]") == 0)
 514                         type = EXCLUSION_LIST;
 515                 else if (strcmp(p, "[ExclusionException]") == 0)
 516                         type = EXCLUSION_EXCEPTION;
 517                 else if (strcmp(p, "[CompressionExclusionList]") == 0)
 518                         type = COMPRESSION_EXCLUSION_LIST;
 519                 else if (strcmp(p, "[AlignmentList]") == 0)
 520                         type = ALIGNMENT_LIST;
 521                 else if (p[0] == '[' && strrchr(p, ']')) {
 522                         ERROR("Unknown capture configuration section `%s'", p);
 523                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 524                 } else switch (type) {
 525                 case EXCLUSION_LIST:
 526                         DEBUG("Adding pattern \"%s\" to exclusion list", p);
 527                         ret = pattern_list_add_pattern(&config->exclusion_list, p);
 528                         break;
 529                 case EXCLUSION_EXCEPTION:
 530                         DEBUG("Adding pattern \"%s\" to exclusion exception list", p);
 531                         ret = pattern_list_add_pattern(&config->exclusion_exception, p);
 532                         break;
 533                 case COMPRESSION_EXCLUSION_LIST:
 534                         DEBUG("Adding pattern \"%s\" to compression exclusion list", p);
 535                         ret = pattern_list_add_pattern(&config->compression_exclusion_list, p);
 536                         break;
 537                 case ALIGNMENT_LIST:
 538                         DEBUG("Adding pattern \"%s\" to alignment list", p);
 539                         ret = pattern_list_add_pattern(&config->alignment_list, p);
 540                         break;
 541                 default:
 542                         ERROR("Line %lu of capture configuration is not "
 543                               "in a block (such as [ExclusionList])",
 544                               line_no);
 545                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 546                         break;
 547                 }
 548                 if (ret != 0)
 549                         goto out_destroy;
 550         }
 551         return 0;
 552 out_destroy:
 553         destroy_capture_config(config);
 554         return ret;
 555 }
 556
 557 static bool match_pattern(const char *path, const char *path_basename,
 558                           const struct pattern_list *list)
 559 {
 560         for (size_t i = 0; i < list->num_pats; i++) {
 561                 const char *pat = list->pats[i];
 562                 const char *string;
 563                 if (pat[0] == '/')
 564                         /* Absolute path from root of capture */
 565                         string = path;
 566                 else {
 567                         if (strchr(pat, '/'))
 568                                 /* Relative path from root of capture */
 569                                 string = path + 1;
 570                         else
 571                                 /* A file name pattern */
 572                                 string = path_basename;
 573                 }
 574                 if (fnmatch(pat, string, FNM_PATHNAME
 575                         #ifdef FNM_CASEFOLD
 576                                         | FNM_CASEFOLD
 577                         #endif
 578                         ) == 0)
 579                 {
 580                         DEBUG("`%s' matches the pattern \"%s\"",
 581                               string, pat);
 582                         return true;
 583                 }
 584         }
 585         return false;
 586 }
 587
 588 /* Return true if the image capture configuration file indicates we should
 589  * exclude the filename @path from capture.
 590  *
 591  * If @exclude_prefix is %true, the part of the path up and including the name
 592  * of the directory being captured is not included in the path for matching
 593  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
 594  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
 595  * directory.
 596  */
 597 bool exclude_path(const char *path, const struct capture_config *config,
 598                   bool exclude_prefix)
 599 {
 600         const char *basename = path_basename(path);
 601         if (exclude_prefix) {
 602                 wimlib_assert(strlen(path) >= config->prefix_len);
 603                 if (memcmp(config->prefix, path, config->prefix_len) == 0
 604                      && path[config->prefix_len] == '/')
 605                         path += config->prefix_len;
 606         }
 607         return match_pattern(path, basename, &config->exclusion_list) &&
 608                 !match_pattern(path, basename, &config->exclusion_exception);
 609
 610 }
 611
 612 /* Strip leading and trailing forward slashes from a string.  Modifies it in
 613  * place and returns the stripped string. */
 614 static const char *canonicalize_target_path(char *target_path)
 615 {
 616         char *p;
 617         if (target_path == NULL)
 618                 target_path = "";
 619         for (;;) {
 620                 if (*target_path == '\0')
 621                         return target_path;
 622                 else if (*target_path == '/')
 623                         target_path++;
 624                 else
 625                         break;
 626         }
 627
 628         p = target_path + strlen(target_path) - 1;
 629         while (*p == '/')
 630                 *p-- = '\0';
 631         return target_path;
 632 }
 633
 634 /* Strip leading and trailing slashes from the target paths */
 635 static void canonicalize_targets(struct wimlib_capture_source *sources,
 636                                  size_t num_sources)
 637 {
 638         while (num_sources--) {
 639                 DEBUG("Canonicalizing { source: \"%s\", target=\"%s\"}",
 640                       sources->fs_source_path,
 641                       sources->wim_target_path);
 642                 sources->wim_target_path =
 643                         (char*)canonicalize_target_path(sources->wim_target_path);
 644                 DEBUG("\"%s\"", sources->wim_target_path);
 645                 sources++;
 646         }
 647 }
 648
 649 static int capture_source_cmp(const void *p1, const void *p2)
 650 {
 651         const struct wimlib_capture_source *s1, *s2;
 652
 653         s1 = p1;
 654         s2 = p2;
 655         return strcmp(s1->wim_target_path, s2->wim_target_path);
 656 }
 657
 658 /* Sorts the capture sources lexicographically by target path.  This occurs
 659  * after leading and trailing forward slashes are stripped.
 660  *
 661  * One purpose of this is to make sure that target paths that are inside other
 662  * target paths are extracted after the containing target paths. */
 663 static void sort_sources(struct wimlib_capture_source *sources,
 664                          size_t num_sources)
 665 {
 666         qsort(sources, num_sources, sizeof(sources[0]), capture_source_cmp);
 667 }
 668
 669 static int check_sorted_sources(struct wimlib_capture_source *sources,
 670                                 size_t num_sources, int add_image_flags)
 671 {
 672         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 673                 if (num_sources != 1) {
 674                         ERROR("Must specify exactly 1 capture source "
 675                               "(the NTFS volume) in NTFS mode!");
 676                         return WIMLIB_ERR_INVALID_PARAM;
 677                 }
 678                 if (sources[0].wim_target_path[0] != '\0') {
 679                         ERROR("In NTFS capture mode the target path inside "
 680                               "the image must be the root directory!");
 681                         return WIMLIB_ERR_INVALID_PARAM;
 682                 }
 683         } else if (num_sources != 0) {
 684                 /* This code is disabled because the current code
 685                  * unconditionally attempts to do overlays.  So, duplicate
 686                  * target paths are OK. */
 687         #if 0
 688                 if (num_sources > 1 && sources[0].wim_target_path[0] == '\0') {
 689                         ERROR("Cannot specify root target when using multiple "
 690                               "capture sources!");
 691                         return WIMLIB_ERR_INVALID_PARAM;
 692                 }
 693                 for (size_t i = 0; i < num_sources - 1; i++) {
 694                         size_t len = strlen(sources[i].wim_target_path);
 695                         size_t j = i + 1;
 696                         const char *target1 = sources[i].wim_target_path;
 697                         do {
 698                                 const char *target2 = sources[j].wim_target_path;
 699                                 DEBUG("target1=%s, target2=%s",
 700                                       target1,target2);
 701                                 if (strncmp(target1, target2, len) ||
 702                                     target2[len] > '/')
 703                                         break;
 704                                 if (target2[len] == '/') {
 705                                         ERROR("Invalid target `%s': is a prefix of `%s'",
 706                                               target1, target2);
 707                                         return WIMLIB_ERR_INVALID_PARAM;
 708                                 }
 709                                 if (target2[len] == '\0') {
 710                                         ERROR("Invalid target `%s': is a duplicate of `%s'",
 711                                               target1, target2);
 712                                         return WIMLIB_ERR_INVALID_PARAM;
 713                                 }
 714                         } while (++j != num_sources);
 715                 }
 716         #endif
 717         }
 718         return 0;
 719
 720 }
 721
 722 /* Creates a new directory to place in the WIM image.  This is to create parent
 723  * directories that are not part of any target as needed.  */
 724 static struct wim_dentry *
 725 new_filler_directory(const char *name)
 726 {
 727         struct wim_dentry *dentry;
 728         DEBUG("Creating filler directory \"%s\"", name);
 729         dentry = new_dentry_with_inode(name);
 730         if (dentry) {
 731                 dentry->d_inode->i_ino = 0;
 732                 dentry->d_inode->i_resolved = 1;
 733                 dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 734         }
 735         return dentry;
 736 }
 737
 738 /* Transfers the children of @branch to @target.  It is an error if @target is
 739  * not a directory or if both @branch and @target contain a child dentry with
 740  * the same name. */
 741 static int do_overlay(struct wim_dentry *target,
 742                       struct wim_dentry *branch)
 743 {
 744         struct rb_root *rb_root;
 745
 746         if (!dentry_is_directory(target)) {
 747                 ERROR("Cannot overlay directory `%s' over non-directory",
 748                       branch->file_name_utf8);
 749                 /* XXX Use a different error code */
 750                 return WIMLIB_ERR_INVALID_DENTRY;
 751         }
 752
 753         rb_root = &branch->d_inode->i_children;
 754         while (rb_root->rb_node) { /* While @branch has children... */
 755                 struct wim_dentry *child;
 756
 757                 child = container_of(rb_root->rb_node, struct wim_dentry, rb_node);
 758                 unlink_dentry(child);
 759                 if (!dentry_add_child(target, child)) {
 760                         dentry_add_child(branch, child);
 761                         ERROR("Overlay error: file `%s' already exists as child of `%s'",
 762                               child->file_name_utf8, target->file_name_utf8);
 763                         return WIMLIB_ERR_INVALID_DENTRY;
 764                 }
 765         }
 766         return 0;
 767
 768 }
 769
 770 /* Attach or overlay a branch onto the WIM image.
 771  *
 772  * @root_p:
 773  *      Pointer to the root of the WIM image, or pointer to NULL if it has not
 774  *      been created yet.
 775  * @branch
 776  *      Branch to add.
 777  * @target_path:
 778  *      Path in the WIM image to add the branch.
 779  */
 780 static int attach_branch(struct wim_dentry **root_p,
 781                          struct wim_dentry *branch,
 782                          char *target_path)
 783 {
 784         char *slash;
 785         struct wim_dentry *dentry, *parent, *target;
 786
 787         if (*target_path == '\0') {
 788                 /* Target: root directory */
 789                 if (*root_p) {
 790                         /* Overlay on existing root */
 791                         return do_overlay(*root_p, branch);
 792                 } else  {
 793                         /* Set as root */
 794                         *root_p = branch;
 795                         return 0;
 796                 }
 797         }
 798
 799         /* Adding a non-root branch.  Create root if it hasn't been created
 800          * already. */
 801         if (!*root_p) {
 802                 *root_p = new_filler_directory("");
 803                 if (!*root_p)
 804                         return WIMLIB_ERR_NOMEM;
 805         }
 806
 807         /* Walk the path to the branch, creating filler directories as needed.
 808          * */
 809         parent = *root_p;
 810         while ((slash = strchr(target_path, '/'))) {
 811                 *slash = '\0';
 812                 dentry = get_dentry_child_with_name(parent, target_path);
 813                 if (!dentry) {
 814                         dentry = new_filler_directory(target_path);
 815                         if (!dentry)
 816                                 return WIMLIB_ERR_NOMEM;
 817                         dentry_add_child(parent, dentry);
 818                 }
 819                 parent = dentry;
 820                 target_path = slash;
 821                 do {
 822                         ++target_path;
 823                         wimlib_assert(*target_path != '\0');
 824                 } while (*target_path == '/');
 825         }
 826
 827         /* If the target path already existed, overlay the branch onto it.
 828          * Otherwise, set the branch as the target path. */
 829         target = get_dentry_child_with_name(parent, branch->file_name_utf8);
 830         if (target) {
 831                 return do_overlay(target, branch);
 832         } else {
 833                 dentry_add_child(parent, branch);
 834                 return 0;
 835         }
 836 }
 837
 838 WIMLIBAPI int wimlib_add_image_multisource(WIMStruct *w,
 839                                            struct wimlib_capture_source *sources,
 840                                            size_t num_sources,
 841                                            const char *name,
 842                                            const char *config_str,
 843                                            size_t config_len,
 844                                            int add_image_flags,
 845                                            wimlib_progress_func_t progress_func)
 846 {
 847         int (*capture_tree)(struct wim_dentry **, const char *,
 848                             struct wim_lookup_table *,
 849                             struct wim_security_data *,
 850                             const struct capture_config *,
 851                             int, wimlib_progress_func_t, void *);
 852         void *extra_arg;
 853         struct wim_dentry *root_dentry = NULL;
 854         struct wim_security_data *sd;
 855         struct capture_config config;
 856         struct wim_image_metadata *imd;
 857         int ret;
 858
 859         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 860 #ifdef WITH_NTFS_3G
 861                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
 862                         ERROR("Cannot dereference files when capturing directly from NTFS");
 863                         return WIMLIB_ERR_INVALID_PARAM;
 864                 }
 865                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 866                         ERROR("Capturing UNIX owner and mode not supported "
 867                               "when capturing directly from NTFS");
 868                         return WIMLIB_ERR_INVALID_PARAM;
 869                 }
 870                 capture_tree = build_dentry_tree_ntfs;
 871                 extra_arg = &w->ntfs_vol;
 872 #else
 873                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
 874                       "        cannot capture a WIM image directly from a NTFS volume!");
 875                 return WIMLIB_ERR_UNSUPPORTED;
 876 #endif
 877         } else {
 878                 capture_tree = build_dentry_tree;
 879                 extra_arg = NULL;
 880         }
 881
 882         if (!name || !*name) {
 883                 ERROR("Must specify a non-empty string for the image name");
 884                 return WIMLIB_ERR_INVALID_PARAM;
 885         }
 886
 887         if (w->hdr.total_parts != 1) {
 888                 ERROR("Cannot add an image to a split WIM");
 889                 return WIMLIB_ERR_SPLIT_UNSUPPORTED;
 890         }
 891
 892         if (wimlib_image_name_in_use(w, name)) {
 893                 ERROR("There is already an image named \"%s\" in `%s'",
 894                       name, w->filename);
 895                 return WIMLIB_ERR_IMAGE_NAME_COLLISION;
 896         }
 897
 898         if (!config_str) {
 899                 DEBUG("Using default capture configuration");
 900                 config_str = default_config;
 901                 config_len = strlen(default_config);
 902         }
 903         memset(&config, 0, sizeof(struct capture_config));
 904
 905         DEBUG("Allocating security data");
 906
 907         sd = CALLOC(1, sizeof(struct wim_security_data));
 908         if (!sd) {
 909                 ret = WIMLIB_ERR_NOMEM;
 910                 goto out_destroy_config;
 911         }
 912         sd->total_length = 8;
 913         sd->refcnt = 1;
 914
 915         DEBUG("Using %zu capture sources", num_sources);
 916         canonicalize_targets(sources, num_sources);
 917         sort_sources(sources, num_sources);
 918         ret = check_sorted_sources(sources, num_sources, add_image_flags);
 919         if (ret) {
 920                 ret = WIMLIB_ERR_INVALID_PARAM;
 921                 goto out_free_security_data;
 922         }
 923
 924         DEBUG("Building dentry tree.");
 925         if (num_sources == 0) {
 926                 root_dentry = new_filler_directory("");
 927                 if (!root_dentry)
 928                         goto out_free_security_data;
 929         } else {
 930                 size_t i = 0;
 931                 struct wim_dentry *branch;
 932                 int flags;
 933                 do {
 934                         DEBUG("Building dentry tree for source %zu of %zu "
 935                               "(\"%s\" => \"%s\")", i + 1, num_sources,
 936                               sources[i].fs_source_path,
 937                               sources[i].wim_target_path);
 938                         union wimlib_progress_info progress;
 939                         if (progress_func) {
 940                                 memset(&progress, 0, sizeof(progress));
 941                                 progress.scan.source = sources[i].fs_source_path;
 942                                 progress.scan.wim_target_path = sources[i].wim_target_path;
 943                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_BEGIN, &progress);
 944                         }
 945                         ret = init_capture_config(config_str, config_len,
 946                                                   sources[i].fs_source_path,
 947                                                   &config);
 948                         if (ret)
 949                                 goto out_free_dentry_tree;
 950                         flags = add_image_flags | WIMLIB_ADD_IMAGE_FLAG_SOURCE;
 951                         if (!*sources[i].wim_target_path)
 952                                 flags |= WIMLIB_ADD_IMAGE_FLAG_ROOT;
 953                         ret = (*capture_tree)(&branch, sources[i].fs_source_path,
 954                                               w->lookup_table, sd,
 955                                               &config,
 956                                               flags,
 957                                               progress_func, extra_arg);
 958                         if (ret) {
 959                                 ERROR("Failed to build dentry tree for `%s'",
 960                                       sources[i].fs_source_path);
 961                                 goto out_free_dentry_tree;
 962                         }
 963                         if (branch) {
 964                                 ret = set_dentry_name(branch,
 965                                                       path_basename(sources[i].wim_target_path));
 966                                 if (ret) {
 967                                         free_dentry_tree(branch, w->lookup_table);
 968                                         goto out_free_dentry_tree;
 969                                 }
 970
 971                                 ret = attach_branch(&root_dentry, branch,
 972                                                     sources[i].wim_target_path);
 973                                 if (ret) {
 974                                         free_dentry_tree(branch, w->lookup_table);
 975                                         goto out_free_dentry_tree;
 976                                 }
 977                         }
 978                         destroy_capture_config(&config);
 979                         if (progress_func)
 980                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_END, &progress);
 981                 } while (++i != num_sources);
 982         }
 983
 984         DEBUG("Calculating full paths of dentries.");
 985         ret = for_dentry_in_tree(root_dentry, calculate_dentry_full_path, NULL);
 986         if (ret != 0)
 987                 goto out_free_dentry_tree;
 988
 989         ret = add_new_dentry_tree(w, root_dentry, sd);
 990         if (ret != 0)
 991                 goto out_free_dentry_tree;
 992
 993         imd = &w->image_metadata[w->hdr.image_count - 1];
 994
 995         ret = dentry_tree_fix_inodes(root_dentry, &imd->inode_list);
 996         if (ret != 0)
 997                 goto out_destroy_imd;
 998
 999         DEBUG("Assigning hard link group IDs");
1000         assign_inode_numbers(&imd->inode_list);
1001
1002         ret = xml_add_image(w, name);
1003         if (ret != 0)
1004                 goto out_destroy_imd;
1005
1006         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_BOOT)
1007                 wimlib_set_boot_idx(w, w->hdr.image_count);
1008         ret = 0;
1009         goto out;
1010 out_destroy_imd:
1011         destroy_image_metadata(&w->image_metadata[w->hdr.image_count - 1],
1012                                w->lookup_table);
1013         w->hdr.image_count--;
1014         return ret;
1015 out_free_dentry_tree:
1016         free_dentry_tree(root_dentry, w->lookup_table);
1017 out_free_security_data:
1018         free_security_data(sd);
1019 out_destroy_config:
1020         destroy_capture_config(&config);
1021 out:
1022         return ret;
1023 }
1024
1025 WIMLIBAPI int wimlib_add_image(WIMStruct *w, const char *source,
1026                                const char *name, const char *config_str,
1027                                size_t config_len, int add_image_flags,
1028                                wimlib_progress_func_t progress_func)
1029 {
1030         if (!source || !*source)
1031                 return WIMLIB_ERR_INVALID_PARAM;
1032
1033         char *fs_source_path = STRDUP(source);
1034         int ret;
1035         struct wimlib_capture_source capture_src = {
1036                 .fs_source_path = fs_source_path,
1037                 .wim_target_path = NULL,
1038                 .reserved = 0,
1039         };
1040         ret = wimlib_add_image_multisource(w, &capture_src, 1, name,
1041                                            config_str, config_len,
1042                                            add_image_flags, progress_func);
1043         FREE(fs_source_path);
1044         return ret;
1045 }