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 #if defined(__CYGWIN__) || defined(__WIN32__)
  27 #       include <windows.h>
  28 #       include <ntdef.h>
  29 #       include <wchar.h>
  30 #       include <sys/cygwin.h>
  31 #       include <fcntl.h>
  32 #       ifdef ERROR
  33 #               undef ERROR
  34 #       endif
  35 #       include "security.h"
  36 #endif
  37
  38 #include "wimlib_internal.h"
  39 #include "dentry.h"
  40 #include "timestamp.h"
  41 #include "lookup_table.h"
  42 #include "xml.h"
  43 #include <string.h>
  44 #include <fnmatch.h>
  45 #include <stdlib.h>
  46 #include <ctype.h>
  47 #include <sys/stat.h>
  48 #include <dirent.h>
  49 #include <errno.h>
  50 #include <unistd.h>
  51
  52 #ifdef HAVE_ALLOCA_H
  53 #include <alloca.h>
  54 #else
  55 #include <stdlib.h>
  56 #endif
  57
  58 #define WIMLIB_ADD_IMAGE_FLAG_ROOT      0x80000000
  59 #define WIMLIB_ADD_IMAGE_FLAG_SOURCE    0x40000000
  60
  61 /*
  62  * Adds the dentry tree and security data for a new image to the image metadata
  63  * array of the WIMStruct.
  64  */
  65 int add_new_dentry_tree(WIMStruct *w, struct wim_dentry *root_dentry,
  66                         struct wim_security_data *sd)
  67 {
  68         struct wim_lookup_table_entry *metadata_lte;
  69         struct wim_image_metadata *imd;
  70         struct wim_image_metadata *new_imd;
  71
  72         wimlib_assert(root_dentry != NULL);
  73
  74         DEBUG("Reallocating image metadata array for image_count = %u",
  75               w->hdr.image_count + 1);
  76         imd = CALLOC((w->hdr.image_count + 1), sizeof(struct wim_image_metadata));
  77
  78         if (!imd) {
  79                 ERROR("Failed to allocate memory for new image metadata array");
  80                 goto err;
  81         }
  82
  83         memcpy(imd, w->image_metadata,
  84                w->hdr.image_count * sizeof(struct wim_image_metadata));
  85
  86         metadata_lte = new_lookup_table_entry();
  87         if (!metadata_lte)
  88                 goto err_free_imd;
  89
  90         metadata_lte->resource_entry.flags = WIM_RESHDR_FLAG_METADATA;
  91         random_hash(metadata_lte->hash);
  92         lookup_table_insert(w->lookup_table, metadata_lte);
  93
  94         new_imd = &imd[w->hdr.image_count];
  95
  96         new_imd->root_dentry    = root_dentry;
  97         new_imd->metadata_lte   = metadata_lte;
  98         new_imd->security_data  = sd;
  99         new_imd->modified       = 1;
 100
 101         FREE(w->image_metadata);
 102         w->image_metadata = imd;
 103         w->hdr.image_count++;
 104         return 0;
 105 err_free_imd:
 106         FREE(imd);
 107 err:
 108         return WIMLIB_ERR_NOMEM;
 109
 110 }
 111
 112 #if defined(__CYGWIN__) || defined(__WIN32__)
 113
 114 static u64 FILETIME_to_u64(const FILETIME *ft)
 115 {
 116         return ((u64)ft->dwHighDateTime << 32) | (u64)ft->dwLowDateTime;
 117 }
 118
 119
 120 static int build_dentry_tree(struct wim_dentry **root_ret,
 121                              const char *root_disk_path,
 122                              struct wim_lookup_table *lookup_table,
 123                              struct wim_security_data *sd,
 124                              const struct capture_config *config,
 125                              int add_image_flags,
 126                              wimlib_progress_func_t progress_func,
 127                              void *extra_arg);
 128
 129 static int win32_get_short_name(struct wim_dentry *dentry,
 130                                 const wchar_t *path_utf16)
 131 {
 132         WIN32_FIND_DATAW dat;
 133         if (FindFirstFileW(path_utf16, &dat) &&
 134             dat.cAlternateFileName[0] != L'\0')
 135         {
 136                 size_t short_name_len = wcslen(dat.cAlternateFileName) * 2;
 137                 size_t n = short_name_len + sizeof(wchar_t);
 138                 dentry->short_name = MALLOC(n);
 139                 if (!dentry->short_name)
 140                         return WIMLIB_ERR_NOMEM;
 141                 memcpy(dentry->short_name, dat.cAlternateFileName, n);
 142                 dentry->short_name_len = short_name_len;
 143         }
 144         return 0;
 145 }
 146
 147 static int win32_get_security_descriptor(struct wim_dentry *dentry,
 148                                          struct sd_set *sd_set,
 149                                          const wchar_t *path_utf16,
 150                                          const char *path)
 151 {
 152         SECURITY_INFORMATION requestedInformation;
 153         DWORD lenNeeded = 0;
 154         BOOL status;
 155         DWORD err;
 156
 157         requestedInformation = DACL_SECURITY_INFORMATION |
 158                                SACL_SECURITY_INFORMATION |
 159                                OWNER_SECURITY_INFORMATION |
 160                                GROUP_SECURITY_INFORMATION;
 161         /* Request length of security descriptor */
 162         status = GetFileSecurityW(path_utf16, requestedInformation,
 163                                   NULL, 0, &lenNeeded);
 164         err = GetLastError();
 165
 166         /* Error code appears to be ERROR_INSUFFICIENT_BUFFER but
 167          * GetFileSecurity is poorly documented... */
 168         if (err == ERROR_INSUFFICIENT_BUFFER || err == NO_ERROR) {
 169                 DWORD len = lenNeeded;
 170                 char buf[len];
 171                 if (GetFileSecurityW(path_utf16, requestedInformation,
 172                                      buf, len, &lenNeeded))
 173                 {
 174                         int security_id = sd_set_add_sd(sd_set, buf, len);
 175                         if (security_id < 0)
 176                                 return WIMLIB_ERR_NOMEM;
 177                         else {
 178                                 dentry->d_inode->i_security_id = security_id;
 179                                 return 0;
 180                         }
 181                 } else {
 182                         err = GetLastError();
 183                 }
 184         }
 185         ERROR("Win32 API: Failed to read security descriptor of \"%s\"",
 186               path);
 187         win32_error(err);
 188         return WIMLIB_ERR_READ;
 189 }
 190
 191
 192 static int win32_recurse_directory(struct wim_dentry *root,
 193                                    const char *root_disk_path,
 194                                    struct wim_lookup_table *lookup_table,
 195                                    struct wim_security_data *sd,
 196                                    const struct capture_config *config,
 197                                    int add_image_flags,
 198                                    wimlib_progress_func_t progress_func,
 199                                    struct sd_set *sd_set,
 200                                    const wchar_t *path_utf16,
 201                                    size_t path_utf16_nchars)
 202 {
 203         WIN32_FIND_DATAW dat;
 204         HANDLE hFind;
 205         DWORD err;
 206         int ret;
 207
 208         {
 209                 wchar_t pattern_buf[path_utf16_nchars + 3];
 210                 memcpy(pattern_buf, path_utf16,
 211                        path_utf16_nchars * sizeof(wchar_t));
 212                 pattern_buf[path_utf16_nchars] = L'/';
 213                 pattern_buf[path_utf16_nchars + 1] = L'*';
 214                 pattern_buf[path_utf16_nchars + 2] = L'\0';
 215                 hFind = FindFirstFileW(pattern_buf, &dat);
 216         }
 217         if (hFind == INVALID_HANDLE_VALUE) {
 218                 err = GetLastError();
 219                 if (err == ERROR_FILE_NOT_FOUND) {
 220                         return 0;
 221                 } else {
 222                         ERROR("Win32 API: Failed to read directory \"%s\"",
 223                               root_disk_path);
 224                         win32_error(err);
 225                         return WIMLIB_ERR_READ;
 226                 }
 227         }
 228         ret = 0;
 229         do {
 230                 if (!(dat.cFileName[0] == L'.' &&
 231                       (dat.cFileName[1] == L'\0' ||
 232                        (dat.cFileName[1] == L'.' && dat.cFileName[2] == L'\0'))))
 233                 {
 234                         struct wim_dentry *child;
 235
 236                         char *utf8_name;
 237                         size_t utf8_name_nbytes;
 238                         ret = utf16_to_utf8((const char*)dat.cFileName,
 239                                             wcslen(dat.cFileName) * sizeof(wchar_t),
 240                                             &utf8_name,
 241                                             &utf8_name_nbytes);
 242                         if (ret)
 243                                 goto out_find_close;
 244
 245                         char name[strlen(root_disk_path) + utf8_name_nbytes + 1];
 246                         sprintf(name, "%s/%s", root_disk_path, utf8_name);
 247                         FREE(utf8_name);
 248                         ret = build_dentry_tree(&child, name, lookup_table,
 249                                                 sd, config, add_image_flags,
 250                                                 progress_func, sd_set);
 251                         if (ret)
 252                                 goto out_find_close;
 253                         if (child)
 254                                 dentry_add_child(root, child);
 255                 }
 256         } while (FindNextFileW(hFind, &dat));
 257         err = GetLastError();
 258         if (err != ERROR_NO_MORE_FILES) {
 259                 ERROR("Win32 API: Failed to read directory \"%s\"", root_disk_path);
 260                 win32_error(err);
 261                 if (ret == 0)
 262                         ret = WIMLIB_ERR_READ;
 263         }
 264 out_find_close:
 265         FindClose(hFind);
 266         return ret;
 267 }
 268
 269 static int win32_capture_reparse_point(const char *path,
 270                                        HANDLE hFile,
 271                                        struct wim_inode *inode,
 272                                        struct wim_lookup_table *lookup_table)
 273 {
 274         /* "Reparse point data, including the tag and optional GUID,
 275          * cannot exceed 16 kilobytes." - MSDN  */
 276         char reparse_point_buf[16 * 1024];
 277         DWORD bytesReturned;
 278         const REPARSE_DATA_BUFFER *buf;
 279
 280         if (!DeviceIoControl(hFile, FSCTL_GET_REPARSE_POINT,
 281                              NULL, 0, reparse_point_buf,
 282                              sizeof(reparse_point_buf), &bytesReturned, NULL))
 283         {
 284                 ERROR("Win32 API: Failed to get reparse data of \"%s\"", path);
 285                 return WIMLIB_ERR_READ;
 286         }
 287         buf = (const REPARSE_DATA_BUFFER*)reparse_point_buf;
 288         inode->i_reparse_tag = buf->ReparseTag;
 289         return inode_add_ads_with_data(inode, "", (const u8*)buf + 8,
 290                                        bytesReturned - 8, lookup_table);
 291 }
 292
 293 static int win32_sha1sum(const wchar_t *path, u8 hash[SHA1_HASH_SIZE])
 294 {
 295         HANDLE hFile;
 296         SHA_CTX ctx;
 297         u8 buf[32768];
 298         DWORD bytesRead;
 299         int ret;
 300
 301         hFile = win32_open_file(path);
 302         if (hFile == INVALID_HANDLE_VALUE)
 303                 return WIMLIB_ERR_OPEN;
 304
 305         sha1_init(&ctx);
 306         for (;;) {
 307                 if (!ReadFile(hFile, buf, sizeof(buf), &bytesRead, NULL)) {
 308                         ret = WIMLIB_ERR_READ;
 309                         goto out_close_handle;
 310                 }
 311                 if (bytesRead == 0)
 312                         break;
 313                 sha1_update(&ctx, buf, bytesRead);
 314         }
 315         ret = 0;
 316         sha1_final(hash, &ctx);
 317 out_close_handle:
 318         CloseHandle(hFile);
 319         return ret;
 320 }
 321
 322 static int win32_capture_stream(const char *path,
 323                                 const wchar_t *path_utf16,
 324                                 size_t path_utf16_nchars,
 325                                 struct wim_inode *inode,
 326                                 struct wim_lookup_table *lookup_table,
 327                                 WIN32_FIND_STREAM_DATA *dat)
 328 {
 329         struct wim_ads_entry *ads_entry;
 330         u8 hash[SHA1_HASH_SIZE];
 331         struct wim_lookup_table_entry *lte;
 332         int ret;
 333         wchar_t *p, *colon;
 334         bool is_named_stream;
 335         wchar_t *spath;
 336         size_t spath_nchars;
 337         DWORD err;
 338
 339         p = dat->cStreamName;
 340         wimlib_assert(*p == L':');
 341         p += 1;
 342         colon = wcschr(p, L':');
 343         wimlib_assert(colon != NULL);
 344
 345         if (wcscmp(colon + 1, L"$DATA")) {
 346                 /* Not a DATA stream */
 347                 ret = 0;
 348                 goto out;
 349         }
 350
 351         is_named_stream = (p != colon);
 352
 353         if (is_named_stream) {
 354                 char *utf8_stream_name;
 355                 size_t utf8_stream_name_len;
 356                 ret = utf16_to_utf8((const char *)p,
 357                                     (colon - p) * sizeof(wchar_t),
 358                                     &utf8_stream_name,
 359                                     &utf8_stream_name_len);
 360                 if (ret)
 361                         goto out;
 362                 ads_entry = inode_add_ads(inode, utf8_stream_name);
 363                 FREE(utf8_stream_name);
 364                 if (!ads_entry) {
 365                         ret = WIMLIB_ERR_NOMEM;
 366                         goto out;
 367                 }
 368         }
 369
 370         *colon = '\0';
 371         spath_nchars = path_utf16_nchars;
 372         if (is_named_stream)
 373                 spath_nchars += colon - p + 1;
 374
 375         spath = MALLOC((spath_nchars + 1) * sizeof(wchar_t));
 376         memcpy(spath, path_utf16, path_utf16_nchars * sizeof(wchar_t));
 377         if (is_named_stream) {
 378                 spath[path_utf16_nchars] = L':';
 379                 memcpy(&spath[path_utf16_nchars + 1], p, (colon - p) * sizeof(wchar_t));
 380         }
 381         spath[spath_nchars] = L'\0';
 382
 383         ret = win32_sha1sum(spath, hash);
 384         if (ret) {
 385                 err = GetLastError();
 386                 ERROR("Win32 API: Failed to read \"%s\" to calculate SHA1sum", path);
 387                 win32_error(err);
 388                 goto out_free_spath;
 389         }
 390
 391         lte = __lookup_resource(lookup_table, hash);
 392         if (lte) {
 393                 lte->refcnt++;
 394         } else {
 395                 lte = new_lookup_table_entry();
 396                 if (!lte) {
 397                         ret = WIMLIB_ERR_NOMEM;
 398                         goto out_free_spath;
 399                 }
 400                 lte->file_on_disk = (char*)spath;
 401                 spath = NULL;
 402                 lte->resource_location = RESOURCE_WIN32;
 403                 lte->resource_entry.original_size = (uint64_t)dat->StreamSize.QuadPart;
 404                 lte->resource_entry.size = (uint64_t)dat->StreamSize.QuadPart;
 405                 copy_hash(lte->hash, hash);
 406                 lookup_table_insert(lookup_table, lte);
 407         }
 408         if (is_named_stream)
 409                 ads_entry->lte = lte;
 410         else
 411                 inode->i_lte = lte;
 412 out_free_spath:
 413         FREE(spath);
 414 out:
 415         return ret;
 416 }
 417
 418 static int win32_capture_streams(const char *path,
 419                                  const wchar_t *path_utf16,
 420                                  size_t path_utf16_nchars,
 421                                  struct wim_inode *inode,
 422                                  struct wim_lookup_table *lookup_table)
 423 {
 424         WIN32_FIND_STREAM_DATA dat;
 425         int ret;
 426         HANDLE hFind;
 427         DWORD err;
 428
 429         hFind = FindFirstStreamW(path_utf16, FindStreamInfoStandard, &dat, 0);
 430         if (hFind == INVALID_HANDLE_VALUE) {
 431                 err = GetLastError();
 432
 433                 /* Seems legal for this to return ERROR_HANDLE_EOF on reparse
 434                  * points and directories */
 435                 if ((inode->i_attributes &
 436                     (FILE_ATTRIBUTE_REPARSE_POINT | FILE_ATTRIBUTE_DIRECTORY))
 437                     && err == ERROR_HANDLE_EOF)
 438                 {
 439                         return 0;
 440                 } else {
 441                         ERROR("Win32 API: Failed to look up data streams of \"%s\"",
 442                               path);
 443                         win32_error(err);
 444                         return WIMLIB_ERR_READ;
 445                 }
 446         }
 447         do {
 448                 ret = win32_capture_stream(path, path_utf16,
 449                                            path_utf16_nchars,
 450                                            inode, lookup_table,
 451                                            &dat);
 452                 if (ret)
 453                         goto out_find_close;
 454         } while (FindNextStreamW(hFind, &dat));
 455         err = GetLastError();
 456         if (err != ERROR_HANDLE_EOF) {
 457                 ERROR("Win32 API: Error reading data streams from \"%s\"", path);
 458                 win32_error(err);
 459                 ret = WIMLIB_ERR_READ;
 460         }
 461 out_find_close:
 462         FindClose(hFind);
 463         return ret;
 464 }
 465 #endif
 466
 467 /*
 468  * build_dentry_tree():
 469  *      Recursively builds a tree of WIM dentries from an on-disk directory
 470  *      tree.
 471  *
 472  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 473  *              modified if successful.  Set to NULL if the file or directory was
 474  *              excluded from capture.
 475  *
 476  * @root_disk_path:  The path to the root of the directory tree on disk.
 477  *
 478  * @lookup_table: The lookup table for the WIM file.  For each file added to the
 479  *              dentry tree being built, an entry is added to the lookup table,
 480  *              unless an identical stream is already in the lookup table.
 481  *              These lookup table entries that are added point to the path of
 482  *              the file on disk.
 483  *
 484  * @sd:         Ignored.  (Security data only captured in NTFS mode.)
 485  *
 486  * @capture_config:
 487  *              Configuration for files to be excluded from capture.
 488  *
 489  * @add_flags:  Bitwise or of WIMLIB_ADD_IMAGE_FLAG_*
 490  *
 491  * @extra_arg:  Ignored. (Only used in NTFS mode.)
 492  *
 493  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 494  *              the files cannot be `stat'ed, or if any of the needed
 495  *              directories cannot be opened or read.  Failure to add the files
 496  *              to the WIM may still occur later when trying to actually read
 497  *              the on-disk files during a call to wimlib_write() or
 498  *              wimlib_overwrite().
 499  */
 500 static int build_dentry_tree(struct wim_dentry **root_ret,
 501                              const char *root_disk_path,
 502                              struct wim_lookup_table *lookup_table,
 503                              struct wim_security_data *sd,
 504                              const struct capture_config *config,
 505                              int add_image_flags,
 506                              wimlib_progress_func_t progress_func,
 507                              void *extra_arg)
 508 {
 509         struct wim_dentry *root = NULL;
 510         int ret = 0;
 511         struct wim_inode *inode;
 512
 513         if (exclude_path(root_disk_path, config, true)) {
 514                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) {
 515                         ERROR("Cannot exclude the root directory from capture");
 516                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 517                         goto out;
 518                 }
 519                 if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 520                     && progress_func)
 521                 {
 522                         union wimlib_progress_info info;
 523                         info.scan.cur_path = root_disk_path;
 524                         info.scan.excluded = true;
 525                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 526                 }
 527                 goto out;
 528         }
 529
 530         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 531             && progress_func)
 532         {
 533                 union wimlib_progress_info info;
 534                 info.scan.cur_path = root_disk_path;
 535                 info.scan.excluded = false;
 536                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 537         }
 538
 539 #if !defined(__CYGWIN__) && !defined(__WIN32__)
 540         /* UNIX version of capturing a directory tree */
 541         struct stat root_stbuf;
 542         int (*stat_fn)(const char *restrict, struct stat *restrict);
 543         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)
 544                 stat_fn = stat;
 545         else
 546                 stat_fn = lstat;
 547
 548         ret = (*stat_fn)(root_disk_path, &root_stbuf);
 549         if (ret != 0) {
 550                 ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 551                 goto out;
 552         }
 553
 554         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) &&
 555               !S_ISDIR(root_stbuf.st_mode))
 556         {
 557                 /* Do a dereference-stat in case the root is a symbolic link.
 558                  * This case is allowed, provided that the symbolic link points
 559                  * to a directory. */
 560                 ret = stat(root_disk_path, &root_stbuf);
 561                 if (ret != 0) {
 562                         ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 563                         ret = WIMLIB_ERR_STAT;
 564                         goto out;
 565                 }
 566                 if (!S_ISDIR(root_stbuf.st_mode)) {
 567                         ERROR("`%s' is not a directory", root_disk_path);
 568                         ret = WIMLIB_ERR_NOTDIR;
 569                         goto out;
 570                 }
 571         }
 572         if (!S_ISREG(root_stbuf.st_mode) && !S_ISDIR(root_stbuf.st_mode)
 573             && !S_ISLNK(root_stbuf.st_mode)) {
 574                 ERROR("`%s' is not a regular file, directory, or symbolic link.",
 575                       root_disk_path);
 576                 ret = WIMLIB_ERR_SPECIAL_FILE;
 577                 goto out;
 578         }
 579
 580         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 581         if (!root) {
 582                 if (errno == EILSEQ)
 583                         ret = WIMLIB_ERR_INVALID_UTF8_STRING;
 584                 else if (errno == ENOMEM)
 585                         ret = WIMLIB_ERR_NOMEM;
 586                 else
 587                         ret = WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 588                 goto out;
 589         }
 590
 591         inode = root->d_inode;
 592
 593 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 594         inode->i_creation_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 595         inode->i_last_write_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 596         inode->i_last_access_time = timespec_to_wim_timestamp(&root_stbuf.st_atim);
 597 #else
 598         inode->i_creation_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 599         inode->i_last_write_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 600         inode->i_last_access_time = unix_timestamp_to_wim(root_stbuf.st_atime);
 601 #endif
 602         if (sizeof(ino_t) >= 8)
 603                 inode->i_ino = (u64)root_stbuf.st_ino;
 604         else
 605                 inode->i_ino = (u64)root_stbuf.st_ino |
 606                                    ((u64)root_stbuf.st_dev << ((sizeof(ino_t) * 8) & 63));
 607         inode->i_resolved = 1;
 608         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 609                 ret = inode_set_unix_data(inode, root_stbuf.st_uid,
 610                                           root_stbuf.st_gid,
 611                                           root_stbuf.st_mode,
 612                                           lookup_table,
 613                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 614                 if (ret)
 615                         goto out;
 616         }
 617         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 618         if (S_ISREG(root_stbuf.st_mode)) { /* Archiving a regular file */
 619
 620                 struct wim_lookup_table_entry *lte;
 621                 u8 hash[SHA1_HASH_SIZE];
 622
 623                 inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 624
 625                 /* Empty files do not have to have a lookup table entry. */
 626                 if (root_stbuf.st_size == 0)
 627                         goto out;
 628
 629                 /* For each regular file, we must check to see if the file is in
 630                  * the lookup table already; if it is, we increment its refcnt;
 631                  * otherwise, we create a new lookup table entry and insert it.
 632                  * */
 633
 634                 ret = sha1sum(root_disk_path, hash);
 635                 if (ret != 0)
 636                         goto out;
 637
 638                 lte = __lookup_resource(lookup_table, hash);
 639                 if (lte) {
 640                         lte->refcnt++;
 641                         DEBUG("Add lte reference %u for `%s'", lte->refcnt,
 642                               root_disk_path);
 643                 } else {
 644                         char *file_on_disk = STRDUP(root_disk_path);
 645                         if (!file_on_disk) {
 646                                 ERROR("Failed to allocate memory for file path");
 647                                 ret = WIMLIB_ERR_NOMEM;
 648                                 goto out;
 649                         }
 650                         lte = new_lookup_table_entry();
 651                         if (!lte) {
 652                                 FREE(file_on_disk);
 653                                 ret = WIMLIB_ERR_NOMEM;
 654                                 goto out;
 655                         }
 656                         lte->file_on_disk = file_on_disk;
 657                         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 658                         lte->resource_entry.original_size = root_stbuf.st_size;
 659                         lte->resource_entry.size = root_stbuf.st_size;
 660                         copy_hash(lte->hash, hash);
 661                         lookup_table_insert(lookup_table, lte);
 662                 }
 663                 root->d_inode->i_lte = lte;
 664         } else if (S_ISDIR(root_stbuf.st_mode)) { /* Archiving a directory */
 665
 666                 inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 667
 668                 DIR *dir;
 669                 struct dirent entry, *result;
 670                 struct wim_dentry *child;
 671
 672                 dir = opendir(root_disk_path);
 673                 if (!dir) {
 674                         ERROR_WITH_ERRNO("Failed to open the directory `%s'",
 675                                          root_disk_path);
 676                         ret = WIMLIB_ERR_OPEN;
 677                         goto out;
 678                 }
 679
 680                 /* Buffer for names of files in directory. */
 681                 size_t len = strlen(root_disk_path);
 682                 char name[len + 1 + FILENAME_MAX + 1];
 683                 memcpy(name, root_disk_path, len);
 684                 name[len] = '/';
 685
 686                 /* Create a dentry for each entry in the directory on disk, and recurse
 687                  * to any subdirectories. */
 688                 while (1) {
 689                         errno = 0;
 690                         ret = readdir_r(dir, &entry, &result);
 691                         if (ret != 0) {
 692                                 ret = WIMLIB_ERR_READ;
 693                                 ERROR_WITH_ERRNO("Error reading the "
 694                                                  "directory `%s'",
 695                                                  root_disk_path);
 696                                 break;
 697                         }
 698                         if (result == NULL)
 699                                 break;
 700                         if (result->d_name[0] == '.' && (result->d_name[1] == '\0'
 701                               || (result->d_name[1] == '.' && result->d_name[2] == '\0')))
 702                                         continue;
 703                         strcpy(name + len + 1, result->d_name);
 704                         ret = build_dentry_tree(&child, name, lookup_table,
 705                                                 NULL, config, add_image_flags,
 706                                                 progress_func, NULL);
 707                         if (ret != 0)
 708                                 break;
 709                         if (child)
 710                                 dentry_add_child(root, child);
 711                 }
 712                 closedir(dir);
 713         } else { /* Archiving a symbolic link */
 714                 inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 715                 inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 716
 717                 /* The idea here is to call readlink() to get the UNIX target of
 718                  * the symbolic link, then turn the target into a reparse point
 719                  * data buffer that contains a relative or absolute symbolic
 720                  * link (NOT a junction point or *full* path symbolic link with
 721                  * drive letter).
 722                  */
 723
 724                 char deref_name_buf[4096];
 725                 ssize_t deref_name_len;
 726
 727                 deref_name_len = readlink(root_disk_path, deref_name_buf,
 728                                           sizeof(deref_name_buf) - 1);
 729                 if (deref_name_len >= 0) {
 730                         deref_name_buf[deref_name_len] = '\0';
 731                         DEBUG("Read symlink `%s'", deref_name_buf);
 732                         ret = inode_set_symlink(root->d_inode, deref_name_buf,
 733                                                 lookup_table, NULL);
 734                         if (ret == 0) {
 735                                 /*
 736                                  * Unfortunately, Windows seems to have the
 737                                  * concept of "file" symbolic links as being
 738                                  * different from "directory" symbolic links...
 739                                  * so FILE_ATTRIBUTE_DIRECTORY needs to be set
 740                                  * on the symbolic link if the *target* of the
 741                                  * symbolic link is a directory.
 742                                  */
 743                                 struct stat stbuf;
 744                                 if (stat(root_disk_path, &stbuf) == 0 &&
 745                                     S_ISDIR(stbuf.st_mode))
 746                                 {
 747                                         inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 748                                 }
 749                         }
 750                 } else {
 751                         ERROR_WITH_ERRNO("Failed to read target of "
 752                                          "symbolic link `%s'", root_disk_path);
 753                         ret = WIMLIB_ERR_READLINK;
 754                 }
 755         }
 756 #else
 757         /* Win32 version of capturing a directory tree */
 758
 759         wchar_t *path_utf16;
 760         size_t path_utf16_nchars;
 761         struct sd_set *sd_set;
 762         DWORD err;
 763
 764         if (extra_arg == NULL) {
 765                 sd_set = alloca(sizeof(struct sd_set));
 766                 sd_set->rb_root.rb_node = NULL,
 767                 sd_set->sd = sd;
 768         } else {
 769                 sd_set = extra_arg;
 770         }
 771
 772         ret = utf8_to_utf16(root_disk_path, strlen(root_disk_path),
 773                             (char**)&path_utf16, &path_utf16_nchars);
 774         if (ret)
 775                 goto out_destroy_sd_set;
 776         path_utf16_nchars /= sizeof(wchar_t);
 777
 778         HANDLE hFile = win32_open_file(path_utf16);
 779         if (hFile == INVALID_HANDLE_VALUE) {
 780                 err = GetLastError();
 781                 ERROR("Win32 API: Failed to open \"%s\"", root_disk_path);
 782                 win32_error(err);
 783                 ret = WIMLIB_ERR_OPEN;
 784                 goto out_free_path_utf16;
 785         }
 786
 787         BY_HANDLE_FILE_INFORMATION file_info;
 788         if (!GetFileInformationByHandle(hFile, &file_info)) {
 789                 err = GetLastError();
 790                 ERROR("Win32 API: Failed to get file information for \"%s\"",
 791                       root_disk_path);
 792                 win32_error(err);
 793                 ret = WIMLIB_ERR_STAT;
 794                 goto out_close_handle;
 795         }
 796
 797         /* Create a WIM dentry */
 798         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 799         if (!root) {
 800                 if (errno == EILSEQ)
 801                         ret = WIMLIB_ERR_INVALID_UTF8_STRING;
 802                 else if (errno == ENOMEM)
 803                         ret = WIMLIB_ERR_NOMEM;
 804                 else
 805                         ret = WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 806                 goto out_free_path_utf16;
 807         }
 808
 809         /* Start preparing the associated WIM inode */
 810         inode = root->d_inode;
 811
 812         inode->i_attributes = file_info.dwFileAttributes;
 813         inode->i_creation_time = FILETIME_to_u64(&file_info.ftCreationTime);
 814         inode->i_last_write_time = FILETIME_to_u64(&file_info.ftLastWriteTime);
 815         inode->i_last_access_time = FILETIME_to_u64(&file_info.ftLastAccessTime);
 816         inode->i_ino = ((u64)file_info.nFileIndexHigh << 32) |
 817                         (u64)file_info.nFileIndexLow;
 818
 819         inode->i_resolved = 1;
 820         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 821
 822         /* Get DOS name and security descriptor (if any). */
 823         ret = win32_get_short_name(root, path_utf16);
 824         if (ret)
 825                 goto out_close_handle;
 826         ret = win32_get_security_descriptor(root, sd_set, path_utf16,
 827                                             root_disk_path);
 828         if (ret)
 829                 goto out_close_handle;
 830
 831         if (inode_is_directory(inode)) {
 832                 /* Directory (not a reparse point) --- recurse to children */
 833
 834                 /* But first... directories may have alternate data streams that
 835                  * need to be captured (maybe?) */
 836                 ret = win32_capture_streams(root_disk_path,
 837                                             path_utf16,
 838                                             path_utf16_nchars,
 839                                             inode,
 840                                             lookup_table);
 841                 if (ret)
 842                         goto out_close_handle;
 843                 ret = win32_recurse_directory(root,
 844                                               root_disk_path,
 845                                               lookup_table,
 846                                               sd,
 847                                               config,
 848                                               add_image_flags,
 849                                               progress_func,
 850                                               sd_set,
 851                                               path_utf16,
 852                                               path_utf16_nchars);
 853         } else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 854                 /* Reparse point: save the reparse tag and data */
 855
 856                 ret = win32_capture_reparse_point(root_disk_path,
 857                                                   hFile,
 858                                                   inode,
 859                                                   lookup_table);
 860
 861         } else {
 862                 /* Not a directory, not a reparse point */
 863                 ret = win32_capture_streams(root_disk_path,
 864                                             path_utf16,
 865                                             path_utf16_nchars,
 866                                             inode,
 867                                             lookup_table);
 868         }
 869 out_close_handle:
 870         CloseHandle(hFile);
 871 out_destroy_sd_set:
 872         if (extra_arg == NULL)
 873                 destroy_sd_set(sd_set);
 874 out_free_path_utf16:
 875         FREE(path_utf16);
 876 #endif
 877 out:
 878         if (ret == 0)
 879                 *root_ret = root;
 880         else
 881                 free_dentry_tree(root, lookup_table);
 882         return ret;
 883 }
 884
 885 enum pattern_type {
 886         NONE = 0,
 887         EXCLUSION_LIST,
 888         EXCLUSION_EXCEPTION,
 889         COMPRESSION_EXCLUSION_LIST,
 890         ALIGNMENT_LIST,
 891 };
 892
 893 #define COMPAT_DEFAULT_CONFIG
 894
 895 /* Default capture configuration file when none is specified. */
 896 static const char *default_config =
 897 #ifdef COMPAT_DEFAULT_CONFIG /* XXX: This policy is being moved to library
 898                                 users.  The next ABI-incompatible library
 899                                 version will default to the empty string here. */
 900 "[ExclusionList]\n"
 901 "\\$ntfs.log\n"
 902 "\\hiberfil.sys\n"
 903 "\\pagefile.sys\n"
 904 "\\System Volume Information\n"
 905 "\\RECYCLER\n"
 906 "\\Windows\\CSC\n"
 907 "\n"
 908 "[CompressionExclusionList]\n"
 909 "*.mp3\n"
 910 "*.zip\n"
 911 "*.cab\n"
 912 "\\WINDOWS\\inf\\*.pnf\n";
 913 #else
 914 "";
 915 #endif
 916
 917 static void destroy_pattern_list(struct pattern_list *list)
 918 {
 919         FREE(list->pats);
 920 }
 921
 922 static void destroy_capture_config(struct capture_config *config)
 923 {
 924         destroy_pattern_list(&config->exclusion_list);
 925         destroy_pattern_list(&config->exclusion_exception);
 926         destroy_pattern_list(&config->compression_exclusion_list);
 927         destroy_pattern_list(&config->alignment_list);
 928         FREE(config->config_str);
 929         FREE(config->prefix);
 930         memset(config, 0, sizeof(*config));
 931 }
 932
 933 static int pattern_list_add_pattern(struct pattern_list *list,
 934                                     const char *pattern)
 935 {
 936         const char **pats;
 937         if (list->num_pats >= list->num_allocated_pats) {
 938                 pats = REALLOC(list->pats,
 939                                sizeof(list->pats[0]) * (list->num_allocated_pats + 8));
 940                 if (!pats)
 941                         return WIMLIB_ERR_NOMEM;
 942                 list->num_allocated_pats += 8;
 943                 list->pats = pats;
 944         }
 945         list->pats[list->num_pats++] = pattern;
 946         return 0;
 947 }
 948
 949 /* Parses the contents of the image capture configuration file and fills in a
 950  * `struct capture_config'. */
 951 static int init_capture_config(struct capture_config *config,
 952                                const char *_config_str, size_t config_len)
 953 {
 954         char *config_str;
 955         char *p;
 956         char *eol;
 957         char *next_p;
 958         size_t bytes_remaining;
 959         enum pattern_type type = NONE;
 960         int ret;
 961         unsigned long line_no = 0;
 962
 963         DEBUG("config_len = %zu", config_len);
 964         bytes_remaining = config_len;
 965         memset(config, 0, sizeof(*config));
 966         config_str = MALLOC(config_len);
 967         if (!config_str) {
 968                 ERROR("Could not duplicate capture config string");
 969                 return WIMLIB_ERR_NOMEM;
 970         }
 971
 972         memcpy(config_str, _config_str, config_len);
 973         next_p = config_str;
 974         config->config_str = config_str;
 975         while (bytes_remaining) {
 976                 line_no++;
 977                 p = next_p;
 978                 eol = memchr(p, '\n', bytes_remaining);
 979                 if (!eol) {
 980                         ERROR("Expected end-of-line in capture config file on "
 981                               "line %lu", line_no);
 982                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 983                         goto out_destroy;
 984                 }
 985
 986                 next_p = eol + 1;
 987                 bytes_remaining -= (next_p - p);
 988                 if (eol == p)
 989                         continue;
 990
 991                 if (*(eol - 1) == '\r')
 992                         eol--;
 993                 *eol = '\0';
 994
 995                 /* Translate backslash to forward slash */
 996                 for (char *pp = p; pp != eol; pp++)
 997                         if (*pp == '\\')
 998                                 *pp = '/';
 999
1000                 /* Remove drive letter */
1001                 if (eol - p > 2 && isalpha(*p) && *(p + 1) == ':')
1002                         p += 2;
1003
1004                 ret = 0;
1005                 if (strcmp(p, "[ExclusionList]") == 0)
1006                         type = EXCLUSION_LIST;
1007                 else if (strcmp(p, "[ExclusionException]") == 0)
1008                         type = EXCLUSION_EXCEPTION;
1009                 else if (strcmp(p, "[CompressionExclusionList]") == 0)
1010                         type = COMPRESSION_EXCLUSION_LIST;
1011                 else if (strcmp(p, "[AlignmentList]") == 0)
1012                         type = ALIGNMENT_LIST;
1013                 else if (p[0] == '[' && strrchr(p, ']')) {
1014                         ERROR("Unknown capture configuration section `%s'", p);
1015                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
1016                 } else switch (type) {
1017                 case EXCLUSION_LIST:
1018                         DEBUG("Adding pattern \"%s\" to exclusion list", p);
1019                         ret = pattern_list_add_pattern(&config->exclusion_list, p);
1020                         break;
1021                 case EXCLUSION_EXCEPTION:
1022                         DEBUG("Adding pattern \"%s\" to exclusion exception list", p);
1023                         ret = pattern_list_add_pattern(&config->exclusion_exception, p);
1024                         break;
1025                 case COMPRESSION_EXCLUSION_LIST:
1026                         DEBUG("Adding pattern \"%s\" to compression exclusion list", p);
1027                         ret = pattern_list_add_pattern(&config->compression_exclusion_list, p);
1028                         break;
1029                 case ALIGNMENT_LIST:
1030                         DEBUG("Adding pattern \"%s\" to alignment list", p);
1031                         ret = pattern_list_add_pattern(&config->alignment_list, p);
1032                         break;
1033                 default:
1034                         ERROR("Line %lu of capture configuration is not "
1035                               "in a block (such as [ExclusionList])",
1036                               line_no);
1037                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
1038                         break;
1039                 }
1040                 if (ret != 0)
1041                         goto out_destroy;
1042         }
1043         return 0;
1044 out_destroy:
1045         destroy_capture_config(config);
1046         return ret;
1047 }
1048
1049 static int capture_config_set_prefix(struct capture_config *config,
1050                                      const char *_prefix)
1051 {
1052         char *prefix = STRDUP(_prefix);
1053
1054         if (!prefix)
1055                 return WIMLIB_ERR_NOMEM;
1056         FREE(config->prefix);
1057         config->prefix = prefix;
1058         config->prefix_len = strlen(prefix);
1059         return 0;
1060 }
1061
1062 static bool match_pattern(const char *path, const char *path_basename,
1063                           const struct pattern_list *list)
1064 {
1065         for (size_t i = 0; i < list->num_pats; i++) {
1066                 const char *pat = list->pats[i];
1067                 const char *string;
1068                 if (pat[0] == '/')
1069                         /* Absolute path from root of capture */
1070                         string = path;
1071                 else {
1072                         if (strchr(pat, '/'))
1073                                 /* Relative path from root of capture */
1074                                 string = path + 1;
1075                         else
1076                                 /* A file name pattern */
1077                                 string = path_basename;
1078                 }
1079                 if (fnmatch(pat, string, FNM_PATHNAME
1080                         #ifdef FNM_CASEFOLD
1081                                         | FNM_CASEFOLD
1082                         #endif
1083                         ) == 0)
1084                 {
1085                         DEBUG("`%s' matches the pattern \"%s\"",
1086                               string, pat);
1087                         return true;
1088                 }
1089         }
1090         return false;
1091 }
1092
1093 /* Return true if the image capture configuration file indicates we should
1094  * exclude the filename @path from capture.
1095  *
1096  * If @exclude_prefix is %true, the part of the path up and including the name
1097  * of the directory being captured is not included in the path for matching
1098  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
1099  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
1100  * directory.
1101  */
1102 bool exclude_path(const char *path, const struct capture_config *config,
1103                   bool exclude_prefix)
1104 {
1105         const char *basename = path_basename(path);
1106         if (exclude_prefix) {
1107                 wimlib_assert(strlen(path) >= config->prefix_len);
1108                 if (memcmp(config->prefix, path, config->prefix_len) == 0
1109                      && path[config->prefix_len] == '/')
1110                         path += config->prefix_len;
1111         }
1112         return match_pattern(path, basename, &config->exclusion_list) &&
1113                 !match_pattern(path, basename, &config->exclusion_exception);
1114
1115 }
1116
1117 /* Strip leading and trailing forward slashes from a string.  Modifies it in
1118  * place and returns the stripped string. */
1119 static const char *canonicalize_target_path(char *target_path)
1120 {
1121         char *p;
1122         if (target_path == NULL)
1123                 target_path = "";
1124         for (;;) {
1125                 if (*target_path == '\0')
1126                         return target_path;
1127                 else if (*target_path == '/')
1128                         target_path++;
1129                 else
1130                         break;
1131         }
1132
1133         p = target_path + strlen(target_path) - 1;
1134         while (*p == '/')
1135                 *p-- = '\0';
1136         return target_path;
1137 }
1138
1139 /* Strip leading and trailing slashes from the target paths */
1140 static void canonicalize_targets(struct wimlib_capture_source *sources,
1141                                  size_t num_sources)
1142 {
1143         while (num_sources--) {
1144                 DEBUG("Canonicalizing { source: \"%s\", target=\"%s\"}",
1145                       sources->fs_source_path,
1146                       sources->wim_target_path);
1147                 sources->wim_target_path =
1148                         (char*)canonicalize_target_path(sources->wim_target_path);
1149                 DEBUG("Canonical target: \"%s\"", sources->wim_target_path);
1150                 sources++;
1151         }
1152 }
1153
1154 static int capture_source_cmp(const void *p1, const void *p2)
1155 {
1156         const struct wimlib_capture_source *s1 = p1, *s2 = p2;
1157         return strcmp(s1->wim_target_path, s2->wim_target_path);
1158 }
1159
1160 /* Sorts the capture sources lexicographically by target path.  This occurs
1161  * after leading and trailing forward slashes are stripped.
1162  *
1163  * One purpose of this is to make sure that target paths that are inside other
1164  * target paths are extracted after the containing target paths. */
1165 static void sort_sources(struct wimlib_capture_source *sources,
1166                          size_t num_sources)
1167 {
1168         qsort(sources, num_sources, sizeof(sources[0]), capture_source_cmp);
1169 }
1170
1171 static int check_sorted_sources(struct wimlib_capture_source *sources,
1172                                 size_t num_sources, int add_image_flags)
1173 {
1174         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
1175                 if (num_sources != 1) {
1176                         ERROR("Must specify exactly 1 capture source "
1177                               "(the NTFS volume) in NTFS mode!");
1178                         return WIMLIB_ERR_INVALID_PARAM;
1179                 }
1180                 if (sources[0].wim_target_path[0] != '\0') {
1181                         ERROR("In NTFS capture mode the target path inside "
1182                               "the image must be the root directory!");
1183                         return WIMLIB_ERR_INVALID_PARAM;
1184                 }
1185         } else if (num_sources != 0) {
1186                 /* This code is disabled because the current code
1187                  * unconditionally attempts to do overlays.  So, duplicate
1188                  * target paths are OK. */
1189         #if 0
1190                 if (num_sources > 1 && sources[0].wim_target_path[0] == '\0') {
1191                         ERROR("Cannot specify root target when using multiple "
1192                               "capture sources!");
1193                         return WIMLIB_ERR_INVALID_PARAM;
1194                 }
1195                 for (size_t i = 0; i < num_sources - 1; i++) {
1196                         size_t len = strlen(sources[i].wim_target_path);
1197                         size_t j = i + 1;
1198                         const char *target1 = sources[i].wim_target_path;
1199                         do {
1200                                 const char *target2 = sources[j].wim_target_path;
1201                                 DEBUG("target1=%s, target2=%s",
1202                                       target1,target2);
1203                                 if (strncmp(target1, target2, len) ||
1204                                     target2[len] > '/')
1205                                         break;
1206                                 if (target2[len] == '/') {
1207                                         ERROR("Invalid target `%s': is a prefix of `%s'",
1208                                               target1, target2);
1209                                         return WIMLIB_ERR_INVALID_PARAM;
1210                                 }
1211                                 if (target2[len] == '\0') {
1212                                         ERROR("Invalid target `%s': is a duplicate of `%s'",
1213                                               target1, target2);
1214                                         return WIMLIB_ERR_INVALID_PARAM;
1215                                 }
1216                         } while (++j != num_sources);
1217                 }
1218         #endif
1219         }
1220         return 0;
1221
1222 }
1223
1224 /* Creates a new directory to place in the WIM image.  This is to create parent
1225  * directories that are not part of any target as needed.  */
1226 static struct wim_dentry *
1227 new_filler_directory(const char *name)
1228 {
1229         struct wim_dentry *dentry;
1230         DEBUG("Creating filler directory \"%s\"", name);
1231         dentry = new_dentry_with_inode(name);
1232         if (dentry) {
1233                 /* Set the inode number to 0 for now.  The final inode number
1234                  * will be assigned later by assign_inode_numbers(). */
1235                 dentry->d_inode->i_ino = 0;
1236                 dentry->d_inode->i_resolved = 1;
1237                 dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
1238         }
1239         return dentry;
1240 }
1241
1242 /* Transfers the children of @branch to @target.  It is an error if @target is
1243  * not a directory or if both @branch and @target contain a child dentry with
1244  * the same name. */
1245 static int do_overlay(struct wim_dentry *target, struct wim_dentry *branch)
1246 {
1247         struct rb_root *rb_root;
1248
1249         if (!dentry_is_directory(target)) {
1250                 ERROR("Cannot overlay directory `%s' over non-directory",
1251                       branch->file_name_utf8);
1252                 return WIMLIB_ERR_INVALID_OVERLAY;
1253         }
1254
1255         rb_root = &branch->d_inode->i_children;
1256         while (rb_root->rb_node) { /* While @branch has children... */
1257                 struct wim_dentry *child = rbnode_dentry(rb_root->rb_node);
1258                 /* Move @child to the directory @target */
1259                 unlink_dentry(child);
1260                 if (!dentry_add_child(target, child)) {
1261                         /* Revert the change to avoid leaking the directory tree
1262                          * rooted at @child */
1263                         dentry_add_child(branch, child);
1264                         ERROR("Overlay error: file `%s' already exists "
1265                               "as a child of `%s'",
1266                               child->file_name_utf8, target->file_name_utf8);
1267                         return WIMLIB_ERR_INVALID_OVERLAY;
1268                 }
1269         }
1270         return 0;
1271
1272 }
1273
1274 /* Attach or overlay a branch onto the WIM image.
1275  *
1276  * @root_p:
1277  *      Pointer to the root of the WIM image, or pointer to NULL if it has not
1278  *      been created yet.
1279  * @branch
1280  *      Branch to add.
1281  * @target_path:
1282  *      Path in the WIM image to add the branch, with leading and trailing
1283  *      slashes stripped.
1284  */
1285 static int attach_branch(struct wim_dentry **root_p,
1286                          struct wim_dentry *branch,
1287                          char *target_path)
1288 {
1289         char *slash;
1290         struct wim_dentry *dentry, *parent, *target;
1291
1292         if (*target_path == '\0') {
1293                 /* Target: root directory */
1294                 if (*root_p) {
1295                         /* Overlay on existing root */
1296                         return do_overlay(*root_p, branch);
1297                 } else  {
1298                         /* Set as root */
1299                         *root_p = branch;
1300                         return 0;
1301                 }
1302         }
1303
1304         /* Adding a non-root branch.  Create root if it hasn't been created
1305          * already. */
1306         if (!*root_p) {
1307                 *root_p = new_filler_directory("");
1308                 if (!*root_p)
1309                         return WIMLIB_ERR_NOMEM;
1310         }
1311
1312         /* Walk the path to the branch, creating filler directories as needed.
1313          * */
1314         parent = *root_p;
1315         while ((slash = strchr(target_path, '/'))) {
1316                 *slash = '\0';
1317                 dentry = get_dentry_child_with_name(parent, target_path);
1318                 if (!dentry) {
1319                         dentry = new_filler_directory(target_path);
1320                         if (!dentry)
1321                                 return WIMLIB_ERR_NOMEM;
1322                         dentry_add_child(parent, dentry);
1323                 }
1324                 parent = dentry;
1325                 target_path = slash;
1326                 /* Skip over slashes.  Note: this cannot overrun the length of
1327                  * the string because the last character cannot be a slash, as
1328                  * trailing slashes were tripped.  */
1329                 do {
1330                         ++target_path;
1331                 } while (*target_path == '/');
1332         }
1333
1334         /* If the target path already existed, overlay the branch onto it.
1335          * Otherwise, set the branch as the target path. */
1336         target = get_dentry_child_with_name(parent, branch->file_name_utf8);
1337         if (target) {
1338                 return do_overlay(target, branch);
1339         } else {
1340                 dentry_add_child(parent, branch);
1341                 return 0;
1342         }
1343 }
1344
1345 WIMLIBAPI int wimlib_add_image_multisource(WIMStruct *w,
1346                                            struct wimlib_capture_source *sources,
1347                                            size_t num_sources,
1348                                            const char *name,
1349                                            const char *config_str,
1350                                            size_t config_len,
1351                                            int add_image_flags,
1352                                            wimlib_progress_func_t progress_func)
1353 {
1354         int (*capture_tree)(struct wim_dentry **, const char *,
1355                             struct wim_lookup_table *,
1356                             struct wim_security_data *,
1357                             const struct capture_config *,
1358                             int, wimlib_progress_func_t, void *);
1359         void *extra_arg;
1360         struct wim_dentry *root_dentry;
1361         struct wim_dentry *branch;
1362         struct wim_security_data *sd;
1363         struct capture_config config;
1364         struct wim_image_metadata *imd;
1365         int ret;
1366
1367         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
1368 #ifdef WITH_NTFS_3G
1369                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
1370                         ERROR("Cannot dereference files when capturing directly from NTFS");
1371                         return WIMLIB_ERR_INVALID_PARAM;
1372                 }
1373                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
1374                         ERROR("Capturing UNIX owner and mode not supported "
1375                               "when capturing directly from NTFS");
1376                         return WIMLIB_ERR_INVALID_PARAM;
1377                 }
1378                 capture_tree = build_dentry_tree_ntfs;
1379                 extra_arg = &w->ntfs_vol;
1380 #else
1381                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1382                       "        cannot capture a WIM image directly from a NTFS volume!");
1383                 return WIMLIB_ERR_UNSUPPORTED;
1384 #endif
1385         } else {
1386                 capture_tree = build_dentry_tree;
1387                 extra_arg = NULL;
1388         }
1389
1390 #if defined(__CYGWIN__) || defined(__WIN32__)
1391         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
1392                 ERROR("Capturing UNIX-specific data is not supported on Windows");
1393                 return WIMLIB_ERR_INVALID_PARAM;
1394         }
1395         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
1396                 ERROR("Dereferencing symbolic links is not supported on Windows");
1397                 return WIMLIB_ERR_INVALID_PARAM;
1398         }
1399 #endif
1400
1401         if (!name || !*name) {
1402                 ERROR("Must specify a non-empty string for the image name");
1403                 return WIMLIB_ERR_INVALID_PARAM;
1404         }
1405
1406         if (w->hdr.total_parts != 1) {
1407                 ERROR("Cannot add an image to a split WIM");
1408                 return WIMLIB_ERR_SPLIT_UNSUPPORTED;
1409         }
1410
1411         if (wimlib_image_name_in_use(w, name)) {
1412                 ERROR("There is already an image named \"%s\" in `%s'",
1413                       name, w->filename);
1414                 return WIMLIB_ERR_IMAGE_NAME_COLLISION;
1415         }
1416
1417         if (!config_str) {
1418                 DEBUG("Using default capture configuration");
1419                 config_str = default_config;
1420                 config_len = strlen(default_config);
1421         }
1422         ret = init_capture_config(&config, config_str, config_len);
1423         if (ret)
1424                 goto out;
1425
1426         DEBUG("Allocating security data");
1427         sd = CALLOC(1, sizeof(struct wim_security_data));
1428         if (!sd) {
1429                 ret = WIMLIB_ERR_NOMEM;
1430                 goto out_destroy_capture_config;
1431         }
1432         sd->total_length = 8;
1433         sd->refcnt = 1;
1434
1435         DEBUG("Using %zu capture sources", num_sources);
1436         canonicalize_targets(sources, num_sources);
1437         sort_sources(sources, num_sources);
1438         ret = check_sorted_sources(sources, num_sources, add_image_flags);
1439         if (ret) {
1440                 ret = WIMLIB_ERR_INVALID_PARAM;
1441                 goto out_free_security_data;
1442         }
1443
1444         DEBUG("Building dentry tree.");
1445         if (num_sources == 0) {
1446                 root_dentry = new_filler_directory("");
1447                 if (!root_dentry)
1448                         goto out_free_security_data;
1449         } else {
1450                 size_t i;
1451
1452                 root_dentry = NULL;
1453                 i = 0;
1454                 do {
1455                         int flags;
1456                         union wimlib_progress_info progress;
1457
1458                         DEBUG("Building dentry tree for source %zu of %zu "
1459                               "(\"%s\" => \"%s\")", i + 1, num_sources,
1460                               sources[i].fs_source_path,
1461                               sources[i].wim_target_path);
1462                         if (progress_func) {
1463                                 memset(&progress, 0, sizeof(progress));
1464                                 progress.scan.source = sources[i].fs_source_path;
1465                                 progress.scan.wim_target_path = sources[i].wim_target_path;
1466                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_BEGIN, &progress);
1467                         }
1468                         ret = capture_config_set_prefix(&config,
1469                                                         sources[i].fs_source_path);
1470                         if (ret)
1471                                 goto out_free_dentry_tree;
1472                         flags = add_image_flags | WIMLIB_ADD_IMAGE_FLAG_SOURCE;
1473                         if (!*sources[i].wim_target_path)
1474                                 flags |= WIMLIB_ADD_IMAGE_FLAG_ROOT;
1475                         ret = (*capture_tree)(&branch, sources[i].fs_source_path,
1476                                               w->lookup_table, sd,
1477                                               &config,
1478                                               flags,
1479                                               progress_func, extra_arg);
1480                         if (ret) {
1481                                 ERROR("Failed to build dentry tree for `%s'",
1482                                       sources[i].fs_source_path);
1483                                 goto out_free_dentry_tree;
1484                         }
1485                         if (branch) {
1486                                 /* Use the target name, not the source name, for
1487                                  * the root of each branch from a capture
1488                                  * source.  (This will also set the root dentry
1489                                  * of the entire image to be unnamed.) */
1490                                 ret = set_dentry_name(branch,
1491                                                       path_basename(sources[i].wim_target_path));
1492                                 if (ret)
1493                                         goto out_free_branch;
1494
1495                                 ret = attach_branch(&root_dentry, branch,
1496                                                     sources[i].wim_target_path);
1497                                 if (ret)
1498                                         goto out_free_branch;
1499                         }
1500                         if (progress_func)
1501                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_END, &progress);
1502                 } while (++i != num_sources);
1503         }
1504
1505         DEBUG("Calculating full paths of dentries.");
1506         ret = for_dentry_in_tree(root_dentry, calculate_dentry_full_path, NULL);
1507         if (ret != 0)
1508                 goto out_free_dentry_tree;
1509
1510         ret = add_new_dentry_tree(w, root_dentry, sd);
1511         if (ret != 0)
1512                 goto out_free_dentry_tree;
1513
1514         imd = &w->image_metadata[w->hdr.image_count - 1];
1515
1516         ret = dentry_tree_fix_inodes(root_dentry, &imd->inode_list);
1517         if (ret != 0)
1518                 goto out_destroy_imd;
1519
1520         DEBUG("Assigning hard link group IDs");
1521         assign_inode_numbers(&imd->inode_list);
1522
1523         ret = xml_add_image(w, name);
1524         if (ret != 0)
1525                 goto out_destroy_imd;
1526
1527         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_BOOT)
1528                 wimlib_set_boot_idx(w, w->hdr.image_count);
1529         ret = 0;
1530         goto out;
1531 out_destroy_imd:
1532         destroy_image_metadata(&w->image_metadata[w->hdr.image_count - 1],
1533                                w->lookup_table);
1534         w->hdr.image_count--;
1535         goto out;
1536 out_free_branch:
1537         free_dentry_tree(branch, w->lookup_table);
1538 out_free_dentry_tree:
1539         free_dentry_tree(root_dentry, w->lookup_table);
1540 out_free_security_data:
1541         free_security_data(sd);
1542 out_destroy_capture_config:
1543         destroy_capture_config(&config);
1544 out:
1545         return ret;
1546 }
1547
1548 WIMLIBAPI int wimlib_add_image(WIMStruct *w, const char *source,
1549                                const char *name, const char *config_str,
1550                                size_t config_len, int add_image_flags,
1551                                wimlib_progress_func_t progress_func)
1552 {
1553         if (!source || !*source)
1554                 return WIMLIB_ERR_INVALID_PARAM;
1555
1556         char *fs_source_path = STRDUP(source);
1557         int ret;
1558         struct wimlib_capture_source capture_src = {
1559                 .fs_source_path = fs_source_path,
1560                 .wim_target_path = NULL,
1561                 .reserved = 0,
1562         };
1563         ret = wimlib_add_image_multisource(w, &capture_src, 1, name,
1564                                            config_str, config_len,
1565                                            add_image_flags, progress_func);
1566         FREE(fs_source_path);
1567         return ret;
1568 }