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
  40 #include <ctype.h>
  41 #include <errno.h>
  42 #include <stdlib.h>
  43 #include <string.h>
  44 #include <unistd.h>
  45
  46 #ifdef HAVE_ALLOCA_H
  47 #include <alloca.h>
  48 #endif
  49
  50 /*
  51  * Adds the dentry tree and security data for a new image to the image metadata
  52  * array of the WIMStruct.
  53  */
  54 int add_new_dentry_tree(WIMStruct *w, struct wim_dentry *root_dentry,
  55                         struct wim_security_data *sd)
  56 {
  57         struct wim_lookup_table_entry *metadata_lte;
  58         struct wim_image_metadata *imd;
  59         struct wim_image_metadata *new_imd;
  60
  61         wimlib_assert(root_dentry != NULL);
  62
  63         DEBUG("Reallocating image metadata array for image_count = %u",
  64               w->hdr.image_count + 1);
  65         imd = CALLOC((w->hdr.image_count + 1), sizeof(struct wim_image_metadata));
  66
  67         if (!imd) {
  68                 ERROR("Failed to allocate memory for new image metadata array");
  69                 goto err;
  70         }
  71
  72         memcpy(imd, w->image_metadata,
  73                w->hdr.image_count * sizeof(struct wim_image_metadata));
  74
  75         metadata_lte = new_lookup_table_entry();
  76         if (!metadata_lte)
  77                 goto err_free_imd;
  78
  79         metadata_lte->resource_entry.flags = WIM_RESHDR_FLAG_METADATA;
  80         random_hash(metadata_lte->hash);
  81         lookup_table_insert(w->lookup_table, metadata_lte);
  82
  83         new_imd = &imd[w->hdr.image_count];
  84
  85         new_imd->root_dentry    = root_dentry;
  86         new_imd->metadata_lte   = metadata_lte;
  87         new_imd->security_data  = sd;
  88         new_imd->modified       = 1;
  89
  90         FREE(w->image_metadata);
  91         w->image_metadata = imd;
  92         w->hdr.image_count++;
  93         return 0;
  94 err_free_imd:
  95         FREE(imd);
  96 err:
  97         return WIMLIB_ERR_NOMEM;
  98
  99 }
 100
 101 #ifndef __WIN32__
 102 /*
 103  * build_dentry_tree():
 104  *      Recursively builds a tree of WIM dentries from an on-disk directory
 105  *      tree.
 106  *
 107  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 108  *              modified if successful.  Set to NULL if the file or directory was
 109  *              excluded from capture.
 110  *
 111  * @root_disk_path:  The path to the root of the directory tree on disk (UTF-8).
 112  *
 113  * @lookup_table: The lookup table for the WIM file.  For each file added to the
 114  *              dentry tree being built, an entry is added to the lookup table,
 115  *              unless an identical stream is already in the lookup table.
 116  *              These lookup table entries that are added point to the path of
 117  *              the file on disk.
 118  *
 119  * @sd:         Ignored.  (Security data only captured in NTFS mode.)
 120  *
 121  * @capture_config:
 122  *              Configuration for files to be excluded from capture.
 123  *
 124  * @add_flags:  Bitwise or of WIMLIB_ADD_IMAGE_FLAG_*
 125  *
 126  * @extra_arg:  Ignored in UNIX builds; used to pass sd_set pointer in Windows
 127  *              builds.
 128  *
 129  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 130  *              the files cannot be `stat'ed, or if any of the needed
 131  *              directories cannot be opened or read.  Failure to add the files
 132  *              to the WIM may still occur later when trying to actually read
 133  *              the on-disk files during a call to wimlib_write() or
 134  *              wimlib_overwrite().
 135  */
 136 static int unix_build_dentry_tree(struct wim_dentry **root_ret,
 137                                   const char *root_disk_path,
 138                                   struct wim_lookup_table *lookup_table,
 139                                   struct wim_security_data *sd,
 140                                   const struct capture_config *config,
 141                                   int add_image_flags,
 142                                   wimlib_progress_func_t progress_func,
 143                                   void *extra_arg)
 144 {
 145         struct wim_dentry *root = NULL;
 146         int ret = 0;
 147         struct wim_inode *inode;
 148
 149         if (exclude_path(root_disk_path, config, true)) {
 150                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) {
 151                         ERROR("Cannot exclude the root directory from capture");
 152                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 153                         goto out;
 154                 }
 155                 if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 156                     && progress_func)
 157                 {
 158                         union wimlib_progress_info info;
 159                         info.scan.cur_path = root_disk_path;
 160                         info.scan.excluded = true;
 161                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 162                 }
 163                 goto out;
 164         }
 165
 166         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 167             && progress_func)
 168         {
 169                 union wimlib_progress_info info;
 170                 info.scan.cur_path = root_disk_path;
 171                 info.scan.excluded = false;
 172                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 173         }
 174
 175         /* UNIX version of capturing a directory tree */
 176         struct stat root_stbuf;
 177         int (*stat_fn)(const char *restrict, struct stat *restrict);
 178         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)
 179                 stat_fn = stat;
 180         else
 181                 stat_fn = lstat;
 182
 183         ret = (*stat_fn)(root_disk_path, &root_stbuf);
 184         if (ret != 0) {
 185                 ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 186                 goto out;
 187         }
 188
 189         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) &&
 190               !S_ISDIR(root_stbuf.st_mode))
 191         {
 192                 /* Do a dereference-stat in case the root is a symbolic link.
 193                  * This case is allowed, provided that the symbolic link points
 194                  * to a directory. */
 195                 ret = stat(root_disk_path, &root_stbuf);
 196                 if (ret != 0) {
 197                         ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 198                         ret = WIMLIB_ERR_STAT;
 199                         goto out;
 200                 }
 201                 if (!S_ISDIR(root_stbuf.st_mode)) {
 202                         ERROR("`%s' is not a directory", root_disk_path);
 203                         ret = WIMLIB_ERR_NOTDIR;
 204                         goto out;
 205                 }
 206         }
 207         if (!S_ISREG(root_stbuf.st_mode) && !S_ISDIR(root_stbuf.st_mode)
 208             && !S_ISLNK(root_stbuf.st_mode)) {
 209                 ERROR("`%s' is not a regular file, directory, or symbolic link.",
 210                       root_disk_path);
 211                 ret = WIMLIB_ERR_SPECIAL_FILE;
 212                 goto out;
 213         }
 214
 215         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 216         if (!root) {
 217                 if (errno == EILSEQ)
 218                         ret = WIMLIB_ERR_INVALID_UTF8_STRING;
 219                 else if (errno == ENOMEM)
 220                         ret = WIMLIB_ERR_NOMEM;
 221                 else
 222                         ret = WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 223                 goto out;
 224         }
 225
 226         inode = root->d_inode;
 227
 228 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 229         inode->i_creation_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 230         inode->i_last_write_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 231         inode->i_last_access_time = timespec_to_wim_timestamp(&root_stbuf.st_atim);
 232 #else
 233         inode->i_creation_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 234         inode->i_last_write_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 235         inode->i_last_access_time = unix_timestamp_to_wim(root_stbuf.st_atime);
 236 #endif
 237         /* Leave the inode number at 0 for directories. */
 238         if (!S_ISDIR(root_stbuf.st_mode)) {
 239                 if (sizeof(ino_t) >= 8)
 240                         inode->i_ino = (u64)root_stbuf.st_ino;
 241                 else
 242                         inode->i_ino = (u64)root_stbuf.st_ino |
 243                                            ((u64)root_stbuf.st_dev <<
 244                                                 ((sizeof(ino_t) * 8) & 63));
 245         }
 246         inode->i_resolved = 1;
 247         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 248                 ret = inode_set_unix_data(inode, root_stbuf.st_uid,
 249                                           root_stbuf.st_gid,
 250                                           root_stbuf.st_mode,
 251                                           lookup_table,
 252                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 253                 if (ret)
 254                         goto out;
 255         }
 256         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 257         if (S_ISREG(root_stbuf.st_mode)) { /* Archiving a regular file */
 258
 259                 struct wim_lookup_table_entry *lte;
 260                 u8 hash[SHA1_HASH_SIZE];
 261
 262                 inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 263
 264                 /* Empty files do not have to have a lookup table entry. */
 265                 if (root_stbuf.st_size == 0)
 266                         goto out;
 267
 268                 /* For each regular file, we must check to see if the file is in
 269                  * the lookup table already; if it is, we increment its refcnt;
 270                  * otherwise, we create a new lookup table entry and insert it.
 271                  * */
 272
 273                 ret = sha1sum(root_disk_path, hash);
 274                 if (ret != 0)
 275                         goto out;
 276
 277                 lte = __lookup_resource(lookup_table, hash);
 278                 if (lte) {
 279                         lte->refcnt++;
 280                         DEBUG("Add lte reference %u for `%s'", lte->refcnt,
 281                               root_disk_path);
 282                 } else {
 283                         char *file_on_disk = STRDUP(root_disk_path);
 284                         if (!file_on_disk) {
 285                                 ERROR("Failed to allocate memory for file path");
 286                                 ret = WIMLIB_ERR_NOMEM;
 287                                 goto out;
 288                         }
 289                         lte = new_lookup_table_entry();
 290                         if (!lte) {
 291                                 FREE(file_on_disk);
 292                                 ret = WIMLIB_ERR_NOMEM;
 293                                 goto out;
 294                         }
 295                         lte->file_on_disk = file_on_disk;
 296                         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 297                         lte->resource_entry.original_size = root_stbuf.st_size;
 298                         lte->resource_entry.size = root_stbuf.st_size;
 299                         copy_hash(lte->hash, hash);
 300                         lookup_table_insert(lookup_table, lte);
 301                 }
 302                 root->d_inode->i_lte = lte;
 303         } else if (S_ISDIR(root_stbuf.st_mode)) { /* Archiving a directory */
 304
 305                 inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 306
 307                 DIR *dir;
 308                 struct dirent entry, *result;
 309                 struct wim_dentry *child;
 310
 311                 dir = opendir(root_disk_path);
 312                 if (!dir) {
 313                         ERROR_WITH_ERRNO("Failed to open the directory `%s'",
 314                                          root_disk_path);
 315                         ret = WIMLIB_ERR_OPEN;
 316                         goto out;
 317                 }
 318
 319                 /* Buffer for names of files in directory. */
 320                 size_t len = strlen(root_disk_path);
 321                 char name[len + 1 + FILENAME_MAX + 1];
 322                 memcpy(name, root_disk_path, len);
 323                 name[len] = '/';
 324
 325                 /* Create a dentry for each entry in the directory on disk, and recurse
 326                  * to any subdirectories. */
 327                 while (1) {
 328                         errno = 0;
 329                         ret = readdir_r(dir, &entry, &result);
 330                         if (ret != 0) {
 331                                 ret = WIMLIB_ERR_READ;
 332                                 ERROR_WITH_ERRNO("Error reading the "
 333                                                  "directory `%s'",
 334                                                  root_disk_path);
 335                                 break;
 336                         }
 337                         if (result == NULL)
 338                                 break;
 339                         if (result->d_name[0] == '.' && (result->d_name[1] == '\0'
 340                               || (result->d_name[1] == '.' && result->d_name[2] == '\0')))
 341                                         continue;
 342                         strcpy(name + len + 1, result->d_name);
 343                         ret = unix_build_dentry_tree(&child, name,
 344                                                      lookup_table,
 345                                                      NULL, config,
 346                                                      add_image_flags,
 347                                                      progress_func, NULL);
 348                         if (ret != 0)
 349                                 break;
 350                         if (child)
 351                                 dentry_add_child(root, child);
 352                 }
 353                 closedir(dir);
 354         } else { /* Archiving a symbolic link */
 355                 inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 356                 inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 357
 358                 /* The idea here is to call readlink() to get the UNIX target of
 359                  * the symbolic link, then turn the target into a reparse point
 360                  * data buffer that contains a relative or absolute symbolic
 361                  * link (NOT a junction point or *full* path symbolic link with
 362                  * drive letter).
 363                  */
 364
 365                 char deref_name_buf[4096];
 366                 ssize_t deref_name_len;
 367
 368                 deref_name_len = readlink(root_disk_path, deref_name_buf,
 369                                           sizeof(deref_name_buf) - 1);
 370                 if (deref_name_len >= 0) {
 371                         deref_name_buf[deref_name_len] = '\0';
 372                         DEBUG("Read symlink `%s'", deref_name_buf);
 373                         ret = inode_set_symlink(root->d_inode, deref_name_buf,
 374                                                 lookup_table, NULL);
 375                         if (ret == 0) {
 376                                 /*
 377                                  * Unfortunately, Windows seems to have the
 378                                  * concept of "file" symbolic links as being
 379                                  * different from "directory" symbolic links...
 380                                  * so FILE_ATTRIBUTE_DIRECTORY needs to be set
 381                                  * on the symbolic link if the *target* of the
 382                                  * symbolic link is a directory.
 383                                  */
 384                                 struct stat stbuf;
 385                                 if (stat(root_disk_path, &stbuf) == 0 &&
 386                                     S_ISDIR(stbuf.st_mode))
 387                                 {
 388                                         inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 389                                 }
 390                         }
 391                 } else {
 392                         ERROR_WITH_ERRNO("Failed to read target of "
 393                                          "symbolic link `%s'", root_disk_path);
 394                         ret = WIMLIB_ERR_READLINK;
 395                 }
 396         }
 397 out:
 398         if (ret == 0)
 399                 *root_ret = root;
 400         else
 401                 free_dentry_tree(root, lookup_table);
 402         return ret;
 403 }
 404 #endif /* !__WIN32__ */
 405
 406 enum pattern_type {
 407         NONE = 0,
 408         EXCLUSION_LIST,
 409         EXCLUSION_EXCEPTION,
 410         COMPRESSION_EXCLUSION_LIST,
 411         ALIGNMENT_LIST,
 412 };
 413
 414 #define COMPAT_DEFAULT_CONFIG
 415
 416 /* Default capture configuration file when none is specified. */
 417 static const char *default_config =
 418 #ifdef COMPAT_DEFAULT_CONFIG /* XXX: This policy is being moved to library
 419                                 users.  The next ABI-incompatible library
 420                                 version will default to the empty string here. */
 421 "[ExclusionList]\n"
 422 "\\$ntfs.log\n"
 423 "\\hiberfil.sys\n"
 424 "\\pagefile.sys\n"
 425 "\\System Volume Information\n"
 426 "\\RECYCLER\n"
 427 "\\Windows\\CSC\n"
 428 "\n"
 429 "[CompressionExclusionList]\n"
 430 "*.mp3\n"
 431 "*.zip\n"
 432 "*.cab\n"
 433 "\\WINDOWS\\inf\\*.pnf\n";
 434 #else
 435 "";
 436 #endif
 437
 438 static void destroy_pattern_list(struct pattern_list *list)
 439 {
 440         FREE(list->pats);
 441 }
 442
 443 static void destroy_capture_config(struct capture_config *config)
 444 {
 445         destroy_pattern_list(&config->exclusion_list);
 446         destroy_pattern_list(&config->exclusion_exception);
 447         destroy_pattern_list(&config->compression_exclusion_list);
 448         destroy_pattern_list(&config->alignment_list);
 449         FREE(config->config_str);
 450         FREE(config->prefix);
 451         memset(config, 0, sizeof(*config));
 452 }
 453
 454 static int pattern_list_add_pattern(struct pattern_list *list,
 455                                     const char *pattern)
 456 {
 457         const char **pats;
 458         if (list->num_pats >= list->num_allocated_pats) {
 459                 pats = REALLOC(list->pats,
 460                                sizeof(list->pats[0]) * (list->num_allocated_pats + 8));
 461                 if (!pats)
 462                         return WIMLIB_ERR_NOMEM;
 463                 list->num_allocated_pats += 8;
 464                 list->pats = pats;
 465         }
 466         list->pats[list->num_pats++] = pattern;
 467         return 0;
 468 }
 469
 470 /* Parses the contents of the image capture configuration file and fills in a
 471  * `struct capture_config'. */
 472 static int init_capture_config(struct capture_config *config,
 473                                const char *_config_str, size_t config_len)
 474 {
 475         char *config_str;
 476         char *p;
 477         char *eol;
 478         char *next_p;
 479         size_t bytes_remaining;
 480         enum pattern_type type = NONE;
 481         int ret;
 482         unsigned long line_no = 0;
 483
 484         DEBUG("config_len = %zu", config_len);
 485         bytes_remaining = config_len;
 486         memset(config, 0, sizeof(*config));
 487         config_str = MALLOC(config_len);
 488         if (!config_str) {
 489                 ERROR("Could not duplicate capture config string");
 490                 return WIMLIB_ERR_NOMEM;
 491         }
 492
 493         memcpy(config_str, _config_str, config_len);
 494         next_p = config_str;
 495         config->config_str = config_str;
 496         while (bytes_remaining) {
 497                 line_no++;
 498                 p = next_p;
 499                 eol = memchr(p, '\n', bytes_remaining);
 500                 if (!eol) {
 501                         ERROR("Expected end-of-line in capture config file on "
 502                               "line %lu", line_no);
 503                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 504                         goto out_destroy;
 505                 }
 506
 507                 next_p = eol + 1;
 508                 bytes_remaining -= (next_p - p);
 509                 if (eol == p)
 510                         continue;
 511
 512                 if (*(eol - 1) == '\r')
 513                         eol--;
 514                 *eol = '\0';
 515
 516                 /* Translate backslash to forward slash */
 517                 for (char *pp = p; pp != eol; pp++)
 518                         if (*pp == '\\')
 519                                 *pp = '/';
 520
 521                 /* Remove drive letter */
 522                 if (eol - p > 2 && isalpha(*p) && *(p + 1) == ':')
 523                         p += 2;
 524
 525                 ret = 0;
 526                 if (strcmp(p, "[ExclusionList]") == 0)
 527                         type = EXCLUSION_LIST;
 528                 else if (strcmp(p, "[ExclusionException]") == 0)
 529                         type = EXCLUSION_EXCEPTION;
 530                 else if (strcmp(p, "[CompressionExclusionList]") == 0)
 531                         type = COMPRESSION_EXCLUSION_LIST;
 532                 else if (strcmp(p, "[AlignmentList]") == 0)
 533                         type = ALIGNMENT_LIST;
 534                 else if (p[0] == '[' && strrchr(p, ']')) {
 535                         ERROR("Unknown capture configuration section `%s'", p);
 536                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 537                 } else switch (type) {
 538                 case EXCLUSION_LIST:
 539                         DEBUG("Adding pattern \"%s\" to exclusion list", p);
 540                         ret = pattern_list_add_pattern(&config->exclusion_list, p);
 541                         break;
 542                 case EXCLUSION_EXCEPTION:
 543                         DEBUG("Adding pattern \"%s\" to exclusion exception list", p);
 544                         ret = pattern_list_add_pattern(&config->exclusion_exception, p);
 545                         break;
 546                 case COMPRESSION_EXCLUSION_LIST:
 547                         DEBUG("Adding pattern \"%s\" to compression exclusion list", p);
 548                         ret = pattern_list_add_pattern(&config->compression_exclusion_list, p);
 549                         break;
 550                 case ALIGNMENT_LIST:
 551                         DEBUG("Adding pattern \"%s\" to alignment list", p);
 552                         ret = pattern_list_add_pattern(&config->alignment_list, p);
 553                         break;
 554                 default:
 555                         ERROR("Line %lu of capture configuration is not "
 556                               "in a block (such as [ExclusionList])",
 557                               line_no);
 558                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 559                         break;
 560                 }
 561                 if (ret != 0)
 562                         goto out_destroy;
 563         }
 564         return 0;
 565 out_destroy:
 566         destroy_capture_config(config);
 567         return ret;
 568 }
 569
 570 static int capture_config_set_prefix(struct capture_config *config,
 571                                      const char *_prefix)
 572 {
 573         char *prefix = STRDUP(_prefix);
 574
 575         if (!prefix)
 576                 return WIMLIB_ERR_NOMEM;
 577         FREE(config->prefix);
 578         config->prefix = prefix;
 579         config->prefix_len = strlen(prefix);
 580         return 0;
 581 }
 582
 583 static bool match_pattern(const char *path, const char *path_basename,
 584                           const struct pattern_list *list)
 585 {
 586         for (size_t i = 0; i < list->num_pats; i++) {
 587                 const char *pat = list->pats[i];
 588                 const char *string;
 589                 if (pat[0] == '/')
 590                         /* Absolute path from root of capture */
 591                         string = path;
 592                 else {
 593                         if (strchr(pat, '/'))
 594                                 /* Relative path from root of capture */
 595                                 string = path + 1;
 596                         else
 597                                 /* A file name pattern */
 598                                 string = path_basename;
 599                 }
 600
 601                 /* Warning: on Windows native builds, fnmatch() calls the
 602                  * replacement function in win32.c. */
 603                 if (fnmatch(pat, string, FNM_PATHNAME
 604                                 #ifdef FNM_CASEFOLD
 605                                         | FNM_CASEFOLD
 606                                 #endif
 607                             ) == 0)
 608                 {
 609                         DEBUG("`%s' matches the pattern \"%s\"",
 610                               string, pat);
 611                         return true;
 612                 }
 613         }
 614         return false;
 615 }
 616
 617 /* Return true if the image capture configuration file indicates we should
 618  * exclude the filename @path from capture.
 619  *
 620  * If @exclude_prefix is %true, the part of the path up and including the name
 621  * of the directory being captured is not included in the path for matching
 622  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
 623  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
 624  * directory.
 625  */
 626 bool exclude_path(const char *path, const struct capture_config *config,
 627                   bool exclude_prefix)
 628 {
 629         const char *basename = path_basename(path);
 630         if (exclude_prefix) {
 631                 wimlib_assert(strlen(path) >= config->prefix_len);
 632                 if (memcmp(config->prefix, path, config->prefix_len) == 0
 633                      && path[config->prefix_len] == '/')
 634                         path += config->prefix_len;
 635         }
 636         return match_pattern(path, basename, &config->exclusion_list) &&
 637                 !match_pattern(path, basename, &config->exclusion_exception);
 638
 639 }
 640
 641 /* Strip leading and trailing forward slashes from a string.  Modifies it in
 642  * place and returns the stripped string. */
 643 static const char *canonicalize_target_path(char *target_path)
 644 {
 645         char *p;
 646         if (target_path == NULL)
 647                 return "";
 648         for (;;) {
 649                 if (*target_path == '\0')
 650                         return target_path;
 651                 else if (*target_path == '/')
 652                         target_path++;
 653                 else
 654                         break;
 655         }
 656
 657         p = target_path + strlen(target_path) - 1;
 658         while (*p == '/')
 659                 *p-- = '\0';
 660         return target_path;
 661 }
 662
 663 #ifdef __WIN32__
 664 static void zap_backslashes(char *s)
 665 {
 666         while (*s) {
 667                 if (*s == '\\')
 668                         *s = '/';
 669                 s++;
 670         }
 671 }
 672 #endif
 673
 674 /* Strip leading and trailing slashes from the target paths */
 675 static void canonicalize_targets(struct wimlib_capture_source *sources,
 676                                  size_t num_sources)
 677 {
 678         while (num_sources--) {
 679                 DEBUG("Canonicalizing { source: \"%s\", target=\"%s\"}",
 680                       sources->fs_source_path,
 681                       sources->wim_target_path);
 682 #ifdef __WIN32__
 683                 /* The Windows API can handle forward slashes.  Just get rid of
 684                  * backslashes to avoid confusing other parts of the library
 685                  * code. */
 686                 zap_backslashes(sources->fs_source_path);
 687                 if (sources->wim_target_path)
 688                         zap_backslashes(sources->wim_target_path);
 689 #endif
 690                 sources->wim_target_path =
 691                         (char*)canonicalize_target_path(sources->wim_target_path);
 692                 DEBUG("Canonical target: \"%s\"", sources->wim_target_path);
 693                 sources++;
 694         }
 695 }
 696
 697 static int capture_source_cmp(const void *p1, const void *p2)
 698 {
 699         const struct wimlib_capture_source *s1 = p1, *s2 = p2;
 700         return strcmp(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 sort_sources(struct wimlib_capture_source *sources,
 709                          size_t num_sources)
 710 {
 711         qsort(sources, num_sources, sizeof(sources[0]), capture_source_cmp);
 712 }
 713
 714 static int check_sorted_sources(struct wimlib_capture_source *sources,
 715                                 size_t num_sources, int add_image_flags)
 716 {
 717         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 718                 if (num_sources != 1) {
 719                         ERROR("Must specify exactly 1 capture source "
 720                               "(the NTFS volume) in NTFS mode!");
 721                         return WIMLIB_ERR_INVALID_PARAM;
 722                 }
 723                 if (sources[0].wim_target_path[0] != '\0') {
 724                         ERROR("In NTFS capture mode the target path inside "
 725                               "the image must be the root directory!");
 726                         return WIMLIB_ERR_INVALID_PARAM;
 727                 }
 728         } else if (num_sources != 0) {
 729                 /* This code is disabled because the current code
 730                  * unconditionally attempts to do overlays.  So, duplicate
 731                  * target paths are OK. */
 732         #if 0
 733                 if (num_sources > 1 && sources[0].wim_target_path[0] == '\0') {
 734                         ERROR("Cannot specify root target when using multiple "
 735                               "capture sources!");
 736                         return WIMLIB_ERR_INVALID_PARAM;
 737                 }
 738                 for (size_t i = 0; i < num_sources - 1; i++) {
 739                         size_t len = strlen(sources[i].wim_target_path);
 740                         size_t j = i + 1;
 741                         const char *target1 = sources[i].wim_target_path;
 742                         do {
 743                                 const char *target2 = sources[j].wim_target_path;
 744                                 DEBUG("target1=%s, target2=%s",
 745                                       target1,target2);
 746                                 if (strncmp(target1, target2, len) ||
 747                                     target2[len] > '/')
 748                                         break;
 749                                 if (target2[len] == '/') {
 750                                         ERROR("Invalid target `%s': is a prefix of `%s'",
 751                                               target1, target2);
 752                                         return WIMLIB_ERR_INVALID_PARAM;
 753                                 }
 754                                 if (target2[len] == '\0') {
 755                                         ERROR("Invalid target `%s': is a duplicate of `%s'",
 756                                               target1, target2);
 757                                         return WIMLIB_ERR_INVALID_PARAM;
 758                                 }
 759                         } while (++j != num_sources);
 760                 }
 761         #endif
 762         }
 763         return 0;
 764
 765 }
 766
 767 /* Creates a new directory to place in the WIM image.  This is to create parent
 768  * directories that are not part of any target as needed.  */
 769 static struct wim_dentry *
 770 new_filler_directory(const char *name)
 771 {
 772         struct wim_dentry *dentry;
 773         DEBUG("Creating filler directory \"%s\"", name);
 774         dentry = new_dentry_with_inode(name);
 775         if (dentry) {
 776                 /* Leave the inode number as 0 for now.  The final inode number
 777                  * will be assigned later by assign_inode_numbers(). */
 778                 dentry->d_inode->i_resolved = 1;
 779                 dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 780         }
 781         return dentry;
 782 }
 783
 784 /* Transfers the children of @branch to @target.  It is an error if @target is
 785  * not a directory or if both @branch and @target contain a child dentry with
 786  * the same name. */
 787 static int do_overlay(struct wim_dentry *target, struct wim_dentry *branch)
 788 {
 789         struct rb_root *rb_root;
 790
 791         DEBUG("Doing overlay %s => %s",
 792               branch->file_name_utf8, target->file_name_utf8);
 793
 794         if (!dentry_is_directory(target)) {
 795                 ERROR("Cannot overlay directory `%s' over non-directory",
 796                       branch->file_name_utf8);
 797                 return WIMLIB_ERR_INVALID_OVERLAY;
 798         }
 799
 800         rb_root = &branch->d_inode->i_children;
 801         while (rb_root->rb_node) { /* While @branch has children... */
 802                 struct wim_dentry *child = rbnode_dentry(rb_root->rb_node);
 803                 /* Move @child to the directory @target */
 804                 unlink_dentry(child);
 805                 if (!dentry_add_child(target, child)) {
 806                         /* Revert the change to avoid leaking the directory tree
 807                          * rooted at @child */
 808                         dentry_add_child(branch, child);
 809                         ERROR("Overlay error: file `%s' already exists "
 810                               "as a child of `%s'",
 811                               child->file_name_utf8, target->file_name_utf8);
 812                         return WIMLIB_ERR_INVALID_OVERLAY;
 813                 }
 814         }
 815         free_dentry(branch);
 816         return 0;
 817
 818 }
 819
 820 /* Attach or overlay a branch onto the WIM image.
 821  *
 822  * @root_p:
 823  *      Pointer to the root of the WIM image, or pointer to NULL if it has not
 824  *      been created yet.
 825  * @branch
 826  *      Branch to add.
 827  * @target_path:
 828  *      Path in the WIM image to add the branch, with leading and trailing
 829  *      slashes stripped.
 830  */
 831 static int attach_branch(struct wim_dentry **root_p,
 832                          struct wim_dentry *branch,
 833                          char *target_path)
 834 {
 835         char *slash;
 836         struct wim_dentry *dentry, *parent, *target;
 837
 838         DEBUG("Attaching branch \"%s\" => \"%s\"",
 839               branch->file_name_utf8, target_path);
 840
 841         if (*target_path == '\0') {
 842                 /* Target: root directory */
 843                 if (*root_p) {
 844                         /* Overlay on existing root */
 845                         return do_overlay(*root_p, branch);
 846                 } else  {
 847                         /* Set as root */
 848                         *root_p = branch;
 849                         return 0;
 850                 }
 851         }
 852
 853         /* Adding a non-root branch.  Create root if it hasn't been created
 854          * already. */
 855         if (!*root_p) {
 856                 *root_p = new_filler_directory("");
 857                 if (!*root_p)
 858                         return WIMLIB_ERR_NOMEM;
 859         }
 860
 861         /* Walk the path to the branch, creating filler directories as needed.
 862          * */
 863         parent = *root_p;
 864         while ((slash = strchr(target_path, '/'))) {
 865                 *slash = '\0';
 866                 dentry = get_dentry_child_with_name(parent, target_path);
 867                 if (!dentry) {
 868                         dentry = new_filler_directory(target_path);
 869                         if (!dentry)
 870                                 return WIMLIB_ERR_NOMEM;
 871                         dentry_add_child(parent, dentry);
 872                 }
 873                 parent = dentry;
 874                 target_path = slash;
 875                 /* Skip over slashes.  Note: this cannot overrun the length of
 876                  * the string because the last character cannot be a slash, as
 877                  * trailing slashes were tripped.  */
 878                 do {
 879                         ++target_path;
 880                 } while (*target_path == '/');
 881         }
 882
 883         /* If the target path already existed, overlay the branch onto it.
 884          * Otherwise, set the branch as the target path. */
 885         target = get_dentry_child_with_name(parent, branch->file_name_utf8);
 886         if (target) {
 887                 return do_overlay(target, branch);
 888         } else {
 889                 dentry_add_child(parent, branch);
 890                 return 0;
 891         }
 892 }
 893
 894 WIMLIBAPI int wimlib_add_image_multisource(WIMStruct *w,
 895                                            struct wimlib_capture_source *sources,
 896                                            size_t num_sources,
 897                                            const char *name,
 898                                            const char *config_str,
 899                                            size_t config_len,
 900                                            int add_image_flags,
 901                                            wimlib_progress_func_t progress_func)
 902 {
 903         int (*capture_tree)(struct wim_dentry **, const char *,
 904                             struct wim_lookup_table *,
 905                             struct wim_security_data *,
 906                             const struct capture_config *,
 907                             int, wimlib_progress_func_t, void *);
 908         void *extra_arg;
 909         struct wim_dentry *root_dentry;
 910         struct wim_dentry *branch;
 911         struct wim_security_data *sd;
 912         struct capture_config config;
 913         struct wim_image_metadata *imd;
 914         int ret;
 915
 916         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
 917 #ifdef WITH_NTFS_3G
 918                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
 919                         ERROR("Cannot dereference files when capturing directly from NTFS");
 920                         return WIMLIB_ERR_INVALID_PARAM;
 921                 }
 922                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 923                         ERROR("Capturing UNIX owner and mode not supported "
 924                               "when capturing directly from NTFS");
 925                         return WIMLIB_ERR_INVALID_PARAM;
 926                 }
 927                 capture_tree = build_dentry_tree_ntfs;
 928                 extra_arg = &w->ntfs_vol;
 929 #else
 930                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
 931                       "        cannot capture a WIM image directly from a NTFS volume!");
 932                 return WIMLIB_ERR_UNSUPPORTED;
 933 #endif
 934         } else {
 935         #ifdef __WIN32__
 936                 capture_tree = win32_build_dentry_tree;
 937         #else
 938                 capture_tree = unix_build_dentry_tree;
 939         #endif
 940                 extra_arg = NULL;
 941         }
 942
 943 #ifdef __WIN32__
 944         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 945                 ERROR("Capturing UNIX-specific data is not supported on Windows");
 946                 return WIMLIB_ERR_INVALID_PARAM;
 947         }
 948         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
 949                 ERROR("Dereferencing symbolic links is not supported on Windows");
 950                 return WIMLIB_ERR_INVALID_PARAM;
 951         }
 952 #endif
 953
 954         if (!name || !*name) {
 955                 ERROR("Must specify a non-empty string for the image name");
 956                 return WIMLIB_ERR_INVALID_PARAM;
 957         }
 958
 959         if (w->hdr.total_parts != 1) {
 960                 ERROR("Cannot add an image to a split WIM");
 961                 return WIMLIB_ERR_SPLIT_UNSUPPORTED;
 962         }
 963
 964         if (wimlib_image_name_in_use(w, name)) {
 965                 ERROR("There is already an image named \"%s\" in `%s'",
 966                       name, w->filename);
 967                 return WIMLIB_ERR_IMAGE_NAME_COLLISION;
 968         }
 969
 970         if (!config_str) {
 971                 DEBUG("Using default capture configuration");
 972                 config_str = default_config;
 973                 config_len = strlen(default_config);
 974         }
 975         ret = init_capture_config(&config, config_str, config_len);
 976         if (ret)
 977                 goto out;
 978
 979         DEBUG("Allocating security data");
 980         sd = CALLOC(1, sizeof(struct wim_security_data));
 981         if (!sd) {
 982                 ret = WIMLIB_ERR_NOMEM;
 983                 goto out_destroy_capture_config;
 984         }
 985         sd->total_length = 8;
 986         sd->refcnt = 1;
 987
 988         DEBUG("Using %zu capture sources", num_sources);
 989         canonicalize_targets(sources, num_sources);
 990         sort_sources(sources, num_sources);
 991         ret = check_sorted_sources(sources, num_sources, add_image_flags);
 992         if (ret) {
 993                 ret = WIMLIB_ERR_INVALID_PARAM;
 994                 goto out_free_security_data;
 995         }
 996
 997         DEBUG("Building dentry tree.");
 998         if (num_sources == 0) {
 999                 root_dentry = new_filler_directory("");
1000                 if (!root_dentry) {
1001                         ret = WIMLIB_ERR_NOMEM;
1002                         goto out_free_security_data;
1003                 }
1004         } else {
1005                 size_t i;
1006
1007         #ifdef __WIN32__
1008                 win32_acquire_capture_privileges();
1009         #endif
1010
1011                 root_dentry = NULL;
1012                 i = 0;
1013                 do {
1014                         int flags;
1015                         union wimlib_progress_info progress;
1016
1017                         DEBUG("Building dentry tree for source %zu of %zu "
1018                               "(\"%s\" => \"%s\")", i + 1, num_sources,
1019                               sources[i].fs_source_path,
1020                               sources[i].wim_target_path);
1021                         if (progress_func) {
1022                                 memset(&progress, 0, sizeof(progress));
1023                                 progress.scan.source = sources[i].fs_source_path;
1024                                 progress.scan.wim_target_path = sources[i].wim_target_path;
1025                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_BEGIN, &progress);
1026                         }
1027                         ret = capture_config_set_prefix(&config,
1028                                                         sources[i].fs_source_path);
1029                         if (ret)
1030                                 goto out_free_dentry_tree;
1031                         flags = add_image_flags | WIMLIB_ADD_IMAGE_FLAG_SOURCE;
1032                         if (!*sources[i].wim_target_path)
1033                                 flags |= WIMLIB_ADD_IMAGE_FLAG_ROOT;
1034                         ret = (*capture_tree)(&branch, sources[i].fs_source_path,
1035                                               w->lookup_table, sd,
1036                                               &config,
1037                                               flags,
1038                                               progress_func, extra_arg);
1039                         if (ret) {
1040                                 ERROR("Failed to build dentry tree for `%s'",
1041                                       sources[i].fs_source_path);
1042                                 goto out_free_dentry_tree;
1043                         }
1044                         if (branch) {
1045                                 /* Use the target name, not the source name, for
1046                                  * the root of each branch from a capture
1047                                  * source.  (This will also set the root dentry
1048                                  * of the entire image to be unnamed.) */
1049                                 ret = set_dentry_name(branch,
1050                                                       path_basename(sources[i].wim_target_path));
1051                                 if (ret)
1052                                         goto out_free_branch;
1053
1054                                 ret = attach_branch(&root_dentry, branch,
1055                                                     sources[i].wim_target_path);
1056                                 if (ret)
1057                                         goto out_free_branch;
1058                         }
1059                         if (progress_func)
1060                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_END, &progress);
1061                 } while (++i != num_sources);
1062         }
1063
1064         DEBUG("Calculating full paths of dentries.");
1065         ret = for_dentry_in_tree(root_dentry, calculate_dentry_full_path, NULL);
1066         if (ret != 0)
1067                 goto out_free_dentry_tree;
1068
1069         ret = add_new_dentry_tree(w, root_dentry, sd);
1070         if (ret != 0)
1071                 goto out_free_dentry_tree;
1072
1073         imd = &w->image_metadata[w->hdr.image_count - 1];
1074
1075         ret = dentry_tree_fix_inodes(root_dentry, &imd->inode_list);
1076         if (ret != 0)
1077                 goto out_destroy_imd;
1078
1079         DEBUG("Assigning hard link group IDs");
1080         assign_inode_numbers(&imd->inode_list);
1081
1082         ret = xml_add_image(w, name);
1083         if (ret != 0)
1084                 goto out_destroy_imd;
1085
1086         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_BOOT)
1087                 wimlib_set_boot_idx(w, w->hdr.image_count);
1088         ret = 0;
1089         goto out_destroy_capture_config;
1090 out_destroy_imd:
1091         destroy_image_metadata(&w->image_metadata[w->hdr.image_count - 1],
1092                                w->lookup_table);
1093         w->hdr.image_count--;
1094         goto out;
1095 out_free_branch:
1096         free_dentry_tree(branch, w->lookup_table);
1097 out_free_dentry_tree:
1098         free_dentry_tree(root_dentry, w->lookup_table);
1099 out_free_security_data:
1100         free_security_data(sd);
1101 out_destroy_capture_config:
1102         destroy_capture_config(&config);
1103 out:
1104 #ifdef __WIN32__
1105         win32_release_capture_privileges();
1106 #endif
1107         return ret;
1108 }
1109
1110 WIMLIBAPI int wimlib_add_image(WIMStruct *w, const char *source,
1111                                const char *name, const char *config_str,
1112                                size_t config_len, int add_image_flags,
1113                                wimlib_progress_func_t progress_func)
1114 {
1115         if (!source || !*source)
1116                 return WIMLIB_ERR_INVALID_PARAM;
1117
1118         char *fs_source_path = STRDUP(source);
1119         int ret;
1120         struct wimlib_capture_source capture_src = {
1121                 .fs_source_path = fs_source_path,
1122                 .wim_target_path = NULL,
1123                 .reserved = 0,
1124         };
1125         ret = wimlib_add_image_multisource(w, &capture_src, 1, name,
1126                                            config_str, config_len,
1127                                            add_image_flags, progress_func);
1128         FREE(fs_source_path);
1129         return ret;
1130 }