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
  25 #if defined(__CYGWIN__) || defined(__WIN32__)
  26 #       include <windows.h>
  27 #       include <ntdef.h>
  28 #       include <wchar.h>
  29 #       include <sys/cygwin.h>
  30 #       include <fcntl.h>
  31 #       ifdef ERROR
  32 #               undef ERROR
  33 #       endif
  34 #       include "security.h"
  35 #endif
  36
  37 #include "wimlib_internal.h"
  38 #include "dentry.h"
  39 #include "timestamp.h"
  40 #include "lookup_table.h"
  41 #include "xml.h"
  42 #include <string.h>
  43 #include <fnmatch.h>
  44 #include <stdlib.h>
  45 #include <ctype.h>
  46 #include <sys/stat.h>
  47 #include <dirent.h>
  48 #include <errno.h>
  49 #include <unistd.h>
  50
  51 #if defined(__CYGWIN__) || defined(__WIN32__)
  52 /*#define ERROR_WIN32_SAFE(format, ...)         \*/
  53 /*{(                                            \*/
  54           /*DWORD err = GetLastError();         \*/
  55         /*ERROR(format, ##__VA_ARGS__);         \*/
  56         /*SetLastError(err);                    \*/
  57 /*)}*/
  58 #define DEBUG_WIN32_SAFE(format, ...)           \
  59 ({                                              \
  60         DWORD err = GetLastError();             \
  61         DEBUG(format, ##__VA_ARGS__);           \
  62         SetLastError(err);                      \
  63 })
  64 #endif
  65
  66 #define WIMLIB_ADD_IMAGE_FLAG_ROOT      0x80000000
  67 #define WIMLIB_ADD_IMAGE_FLAG_SOURCE    0x40000000
  68
  69 /*
  70  * Adds the dentry tree and security data for a new image to the image metadata
  71  * array of the WIMStruct.
  72  */
  73 int add_new_dentry_tree(WIMStruct *w, struct wim_dentry *root_dentry,
  74                         struct wim_security_data *sd)
  75 {
  76         struct wim_lookup_table_entry *metadata_lte;
  77         struct wim_image_metadata *imd;
  78         struct wim_image_metadata *new_imd;
  79
  80         wimlib_assert(root_dentry != NULL);
  81
  82         DEBUG("Reallocating image metadata array for image_count = %u",
  83               w->hdr.image_count + 1);
  84         imd = CALLOC((w->hdr.image_count + 1), sizeof(struct wim_image_metadata));
  85
  86         if (!imd) {
  87                 ERROR("Failed to allocate memory for new image metadata array");
  88                 goto err;
  89         }
  90
  91         memcpy(imd, w->image_metadata,
  92                w->hdr.image_count * sizeof(struct wim_image_metadata));
  93
  94         metadata_lte = new_lookup_table_entry();
  95         if (!metadata_lte)
  96                 goto err_free_imd;
  97
  98         metadata_lte->resource_entry.flags = WIM_RESHDR_FLAG_METADATA;
  99         random_hash(metadata_lte->hash);
 100         lookup_table_insert(w->lookup_table, metadata_lte);
 101
 102         new_imd = &imd[w->hdr.image_count];
 103
 104         new_imd->root_dentry    = root_dentry;
 105         new_imd->metadata_lte   = metadata_lte;
 106         new_imd->security_data  = sd;
 107         new_imd->modified       = 1;
 108
 109         FREE(w->image_metadata);
 110         w->image_metadata = imd;
 111         w->hdr.image_count++;
 112         return 0;
 113 err_free_imd:
 114         FREE(imd);
 115 err:
 116         return WIMLIB_ERR_NOMEM;
 117
 118 }
 119
 120 #if defined(__CYGWIN__) || defined(__WIN32__)
 121 static u64 FILETIME_to_u64(const FILETIME *ft)
 122 {
 123         return ((u64)ft->dwHighDateTime << 32) | (u64)ft->dwLowDateTime;
 124 }
 125
 126 #ifdef ENABLE_ERROR_MESSAGES
 127 static void win32_error(DWORD err_code)
 128 {
 129         char *buffer;
 130         DWORD nchars;
 131         nchars = FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_ALLOCATE_BUFFER,
 132                                 NULL, err_code, 0,
 133                                 (char*)&buffer, 0, NULL);
 134         if (nchars == 0) {
 135                 ERROR("Error printing error message! "
 136                       "Computer will self-destruct in 3 seconds.");
 137         } else {
 138                 ERROR("Win32 error: %s", buffer);
 139                 LocalFree(buffer);
 140         }
 141 }
 142 #else
 143 #define win32_error(err_code)
 144 #endif
 145
 146 static HANDLE win32_open_file(const wchar_t *path)
 147 {
 148         return CreateFileW(path,
 149                            GENERIC_READ | READ_CONTROL,
 150                            FILE_SHARE_READ,
 151                            NULL, /* lpSecurityAttributes */
 152                            OPEN_EXISTING,
 153                            FILE_FLAG_BACKUP_SEMANTICS |
 154                                    FILE_FLAG_OPEN_REPARSE_POINT,
 155                            NULL /* hTemplateFile */);
 156 }
 157
 158 int win32_read_file(const char *filename,
 159                     void *handle, u64 offset, size_t size, u8 *buf)
 160 {
 161         HANDLE h = handle;
 162         DWORD err;
 163         DWORD bytesRead;
 164         LARGE_INTEGER liOffset = {.QuadPart = offset};
 165
 166         wimlib_assert(size <= 0xffffffff);
 167
 168         if (SetFilePointerEx(h, liOffset, NULL, FILE_BEGIN))
 169                 if (ReadFile(h, buf, size, &bytesRead, NULL) && bytesRead == size)
 170                         return 0;
 171         err = GetLastError();
 172         ERROR("Error reading \"%s\"", filename);
 173         win32_error(err);
 174         return WIMLIB_ERR_READ;
 175 }
 176
 177 void win32_close_handle(void *handle)
 178 {
 179         CloseHandle((HANDLE)handle);
 180 }
 181
 182 void *win32_open_handle(const char *path_utf16)
 183 {
 184         return (void*)win32_open_file((const wchar_t*)path_utf16);
 185 }
 186
 187 static int build_dentry_tree(struct wim_dentry **root_ret,
 188                              const char *root_disk_path,
 189                              struct wim_lookup_table *lookup_table,
 190                              struct wim_security_data *sd,
 191                              const struct capture_config *config,
 192                              int add_image_flags,
 193                              wimlib_progress_func_t progress_func,
 194                              void *extra_arg);
 195
 196 static int win32_get_short_name(struct wim_dentry *dentry,
 197                                 const wchar_t *path_utf16)
 198 {
 199         WIN32_FIND_DATAW dat;
 200         if (FindFirstFileW(path_utf16, &dat) &&
 201             dat.cAlternateFileName[0] != L'\0')
 202         {
 203                 size_t short_name_len = wcslen(dat.cAlternateFileName) * 2;
 204                 size_t n = short_name_len + sizeof(wchar_t);
 205                 dentry->short_name = MALLOC(n);
 206                 if (!dentry->short_name)
 207                         return WIMLIB_ERR_NOMEM;
 208                 memcpy(dentry->short_name, dat.cAlternateFileName, n);
 209                 dentry->short_name_len = short_name_len;
 210         }
 211         return 0;
 212 }
 213
 214 static int win32_get_security_descriptor(struct wim_dentry *dentry,
 215                                          struct sd_set *sd_set,
 216                                          const wchar_t *path_utf16,
 217                                          const char *path)
 218 {
 219         SECURITY_INFORMATION requestedInformation;
 220         DWORD lenNeeded = 0;
 221         BOOL status;
 222         DWORD err;
 223
 224 #ifdef BACKUP_SECURITY_INFORMATION
 225         requestedInformation = BACKUP_SECURITY_INFORMATION;
 226 #else
 227         requestedInformation = DACL_SECURITY_INFORMATION |
 228                                SACL_SECURITY_INFORMATION |
 229                                OWNER_SECURITY_INFORMATION |
 230                                GROUP_SECURITY_INFORMATION;
 231 #endif
 232         /* Request length of security descriptor */
 233         status = GetFileSecurityW(path_utf16, requestedInformation,
 234                                   NULL, 0, &lenNeeded);
 235         err = GetLastError();
 236
 237         /* Error code appears to be ERROR_INSUFFICIENT_BUFFER but
 238          * GetFileSecurity is poorly documented... */
 239         if (err == ERROR_INSUFFICIENT_BUFFER || err == NO_ERROR) {
 240                 DWORD len = lenNeeded;
 241                 char buf[len];
 242                 if (GetFileSecurityW(path_utf16, requestedInformation,
 243                                      buf, len, &lenNeeded))
 244                 {
 245                         int security_id = sd_set_add_sd(sd_set, buf, len);
 246                         if (security_id < 0)
 247                                 return WIMLIB_ERR_NOMEM;
 248                         else {
 249                                 dentry->d_inode->i_security_id = security_id;
 250                                 return 0;
 251                         }
 252                 } else {
 253                         err = GetLastError();
 254                 }
 255         }
 256         ERROR("Win32 API: Failed to read security descriptor of \"%s\"",
 257               path);
 258         win32_error(err);
 259         return WIMLIB_ERR_READ;
 260 }
 261
 262
 263 static int win32_recurse_directory(struct wim_dentry *root,
 264                                    const char *root_disk_path,
 265                                    struct wim_lookup_table *lookup_table,
 266                                    struct wim_security_data *sd,
 267                                    const struct capture_config *config,
 268                                    int add_image_flags,
 269                                    wimlib_progress_func_t progress_func,
 270                                    struct sd_set *sd_set,
 271                                    const wchar_t *path_utf16,
 272                                    size_t path_utf16_nchars)
 273 {
 274         WIN32_FIND_DATAW dat;
 275         HANDLE hFind;
 276         DWORD err;
 277         int ret;
 278
 279         {
 280                 wchar_t pattern_buf[path_utf16_nchars + 3];
 281                 memcpy(pattern_buf, path_utf16,
 282                        path_utf16_nchars * sizeof(wchar_t));
 283                 pattern_buf[path_utf16_nchars] = L'/';
 284                 pattern_buf[path_utf16_nchars + 1] = L'*';
 285                 pattern_buf[path_utf16_nchars + 2] = L'\0';
 286                 hFind = FindFirstFileW(pattern_buf, &dat);
 287         }
 288         if (hFind == INVALID_HANDLE_VALUE) {
 289                 err = GetLastError();
 290                 if (err == ERROR_FILE_NOT_FOUND) {
 291                         return 0;
 292                 } else {
 293                         ERROR("Win32 API: Failed to read directory \"%s\"",
 294                               root_disk_path);
 295                         win32_error(err);
 296                         return WIMLIB_ERR_READ;
 297                 }
 298         }
 299         ret = 0;
 300         do {
 301                 if (!(dat.cFileName[0] == L'.' &&
 302                       (dat.cFileName[1] == L'\0' ||
 303                        (dat.cFileName[1] == L'.' && dat.cFileName[2] == L'\0'))))
 304                 {
 305                         struct wim_dentry *child;
 306
 307                         char *utf8_name;
 308                         size_t utf8_name_nbytes;
 309                         ret = utf16_to_utf8((const char*)dat.cFileName,
 310                                             wcslen(dat.cFileName) * sizeof(wchar_t),
 311                                             &utf8_name,
 312                                             &utf8_name_nbytes);
 313                         if (ret)
 314                                 goto out_find_close;
 315
 316                         char name[strlen(root_disk_path) + utf8_name_nbytes + 1];
 317                         sprintf(name, "%s/%s", root_disk_path, utf8_name);
 318                         FREE(utf8_name);
 319                         ret = build_dentry_tree(&child, name, lookup_table,
 320                                                 sd, config, add_image_flags,
 321                                                 progress_func, sd_set);
 322                         if (ret)
 323                                 goto out_find_close;
 324                         if (child)
 325                                 dentry_add_child(root, child);
 326                 }
 327         } while (FindNextFileW(hFind, &dat));
 328         err = GetLastError();
 329         if (err != ERROR_NO_MORE_FILES) {
 330                 ERROR("Win32 API: Failed to read directory \"%s\"", root_disk_path);
 331                 win32_error(err);
 332                 if (ret == 0)
 333                         ret = WIMLIB_ERR_READ;
 334         }
 335 out_find_close:
 336         FindClose(hFind);
 337         return ret;
 338 }
 339
 340 static int win32_capture_reparse_point(const char *path,
 341                                        HANDLE hFile,
 342                                        struct wim_inode *inode,
 343                                        struct wim_lookup_table *lookup_table)
 344 {
 345         /* "Reparse point data, including the tag and optional GUID,
 346          * cannot exceed 16 kilobytes." - MSDN  */
 347         char reparse_point_buf[16 * 1024];
 348         DWORD bytesReturned;
 349         const REPARSE_DATA_BUFFER *buf;
 350
 351         if (!DeviceIoControl(hFile, FSCTL_GET_REPARSE_POINT,
 352                              NULL, 0, reparse_point_buf,
 353                              sizeof(reparse_point_buf), &bytesReturned, NULL))
 354         {
 355                 ERROR("Win32 API: Failed to get reparse data of \"%s\"", path);
 356                 return WIMLIB_ERR_READ;
 357         }
 358         buf = (const REPARSE_DATA_BUFFER*)reparse_point_buf;
 359         inode->i_reparse_tag = buf->ReparseTag;
 360         return inode_add_ads_with_data(inode, "", (const u8*)buf + 8,
 361                                        bytesReturned - 8, lookup_table);
 362 }
 363
 364 static int win32_sha1sum(const wchar_t *path, u8 hash[SHA1_HASH_SIZE])
 365 {
 366         HANDLE hFile;
 367         SHA_CTX ctx;
 368         u8 buf[32768];
 369         DWORD bytesRead;
 370         int ret;
 371
 372         hFile = win32_open_file(path);
 373         if (hFile == INVALID_HANDLE_VALUE)
 374                 return WIMLIB_ERR_OPEN;
 375
 376         sha1_init(&ctx);
 377         for (;;) {
 378                 if (!ReadFile(hFile, buf, sizeof(buf), &bytesRead, NULL)) {
 379                         ret = WIMLIB_ERR_READ;
 380                         goto out_close_handle;
 381                 }
 382                 if (bytesRead == 0)
 383                         break;
 384                 sha1_update(&ctx, buf, bytesRead);
 385         }
 386         ret = 0;
 387         sha1_final(hash, &ctx);
 388 out_close_handle:
 389         CloseHandle(hFile);
 390         return ret;
 391 }
 392
 393 static int win32_capture_stream(const char *path,
 394                                 const wchar_t *path_utf16,
 395                                 size_t path_utf16_nchars,
 396                                 struct wim_inode *inode,
 397                                 struct wim_lookup_table *lookup_table,
 398                                 WIN32_FIND_STREAM_DATA *dat)
 399 {
 400         struct wim_ads_entry *ads_entry;
 401         u8 hash[SHA1_HASH_SIZE];
 402         struct wim_lookup_table_entry *lte;
 403         int ret;
 404         wchar_t *p, *colon;
 405         bool is_named_stream;
 406         wchar_t *spath;
 407         size_t spath_nchars;
 408         DWORD err;
 409
 410         p = dat->cStreamName;
 411         wimlib_assert(*p == L':');
 412         p += 1;
 413         colon = wcschr(p, L':');
 414         wimlib_assert(colon != NULL);
 415
 416         if (wcscmp(colon + 1, L"$DATA")) {
 417                 /* Not a DATA stream */
 418                 ret = 0;
 419                 goto out;
 420         }
 421
 422         is_named_stream = (p != colon);
 423
 424         if (is_named_stream) {
 425                 char *utf8_stream_name;
 426                 size_t utf8_stream_name_len;
 427                 ret = utf16_to_utf8((const char *)p,
 428                                     colon - p,
 429                                     &utf8_stream_name,
 430                                     &utf8_stream_name_len);
 431                 if (ret)
 432                         goto out;
 433                 DEBUG_WIN32_SAFE("Add alternate data stream %s:%s", path, utf8_stream_name);
 434                 ads_entry = inode_add_ads(inode, utf8_stream_name);
 435                 FREE(utf8_stream_name);
 436                 if (!ads_entry) {
 437                         ret = WIMLIB_ERR_NOMEM;
 438                         goto out;
 439                 }
 440         }
 441
 442         *colon = '\0';
 443         spath_nchars = path_utf16_nchars;
 444         if (is_named_stream)
 445                 spath_nchars += colon - p + 1;
 446
 447         spath = MALLOC((spath_nchars + 1) * sizeof(wchar_t));
 448         memcpy(spath, path_utf16, path_utf16_nchars * sizeof(wchar_t));
 449         if (is_named_stream) {
 450                 spath[path_utf16_nchars] = L':';
 451                 memcpy(&spath[path_utf16_nchars + 1], p, (colon - p) * sizeof(wchar_t));
 452         }
 453         spath[spath_nchars] = L'\0';
 454
 455         ret = win32_sha1sum(spath, hash);
 456         if (ret) {
 457                 err = GetLastError();
 458                 ERROR("Win32 API: Failed to read \"%s\" to calculate SHA1sum", path);
 459                 win32_error(err);
 460                 goto out_free_spath;
 461         }
 462
 463         lte = __lookup_resource(lookup_table, hash);
 464         if (lte) {
 465                 lte->refcnt++;
 466         } else {
 467                 lte = new_lookup_table_entry();
 468                 if (!lte) {
 469                         ret = WIMLIB_ERR_NOMEM;
 470                         goto out_free_spath;
 471                 }
 472                 lte->file_on_disk = (char*)spath;
 473                 spath = NULL;
 474                 lte->resource_location = RESOURCE_WIN32;
 475                 lte->resource_entry.original_size = (uint64_t)dat->StreamSize.QuadPart;
 476                 lte->resource_entry.size = (uint64_t)dat->StreamSize.QuadPart;
 477                 copy_hash(lte->hash, hash);
 478                 lookup_table_insert(lookup_table, lte);
 479         }
 480         if (is_named_stream)
 481                 ads_entry->lte = lte;
 482         else
 483                 inode->i_lte = lte;
 484 out_free_spath:
 485         FREE(spath);
 486 out:
 487         return ret;
 488 }
 489
 490 static int win32_capture_streams(const char *path,
 491                                  const wchar_t *path_utf16,
 492                                  size_t path_utf16_nchars,
 493                                  struct wim_inode *inode,
 494                                  struct wim_lookup_table *lookup_table)
 495 {
 496         WIN32_FIND_STREAM_DATA dat;
 497         int ret;
 498         HANDLE hFind;
 499         DWORD err;
 500
 501         hFind = FindFirstStreamW(path_utf16, FindStreamInfoStandard, &dat, 0);
 502         if (hFind == INVALID_HANDLE_VALUE) {
 503                 ERROR("Win32 API: Failed to look up data streams of \"%s\"",
 504                       path);
 505                 return WIMLIB_ERR_READ;
 506         }
 507         do {
 508                 ret = win32_capture_stream(path, path_utf16,
 509                                            path_utf16_nchars,
 510                                            inode, lookup_table,
 511                                            &dat);
 512                 if (ret)
 513                         goto out_find_close;
 514         } while (FindNextStreamW(hFind, &dat));
 515         err = GetLastError();
 516         if (err != ERROR_HANDLE_EOF) {
 517                 ERROR("Win32 API: Error reading data streams from \"%s\"", path);
 518                 win32_error(err);
 519                 ret = WIMLIB_ERR_READ;
 520         }
 521 out_find_close:
 522         FindClose(hFind);
 523         return ret;
 524 }
 525 #endif
 526
 527 /*
 528  * build_dentry_tree():
 529  *      Recursively builds a tree of WIM dentries from an on-disk directory
 530  *      tree.
 531  *
 532  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 533  *              modified if successful.  Set to NULL if the file or directory was
 534  *              excluded from capture.
 535  *
 536  * @root_disk_path:  The path to the root of the directory tree on disk.
 537  *
 538  * @lookup_table: The lookup table for the WIM file.  For each file added to the
 539  *              dentry tree being built, an entry is added to the lookup table,
 540  *              unless an identical stream is already in the lookup table.
 541  *              These lookup table entries that are added point to the path of
 542  *              the file on disk.
 543  *
 544  * @sd:         Ignored.  (Security data only captured in NTFS mode.)
 545  *
 546  * @capture_config:
 547  *              Configuration for files to be excluded from capture.
 548  *
 549  * @add_flags:  Bitwise or of WIMLIB_ADD_IMAGE_FLAG_*
 550  *
 551  * @extra_arg:  Ignored. (Only used in NTFS mode.)
 552  *
 553  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 554  *              the files cannot be `stat'ed, or if any of the needed
 555  *              directories cannot be opened or read.  Failure to add the files
 556  *              to the WIM may still occur later when trying to actually read
 557  *              the on-disk files during a call to wimlib_write() or
 558  *              wimlib_overwrite().
 559  */
 560 static int build_dentry_tree(struct wim_dentry **root_ret,
 561                              const char *root_disk_path,
 562                              struct wim_lookup_table *lookup_table,
 563                              struct wim_security_data *sd,
 564                              const struct capture_config *config,
 565                              int add_image_flags,
 566                              wimlib_progress_func_t progress_func,
 567                              void *extra_arg)
 568 {
 569         struct wim_dentry *root = NULL;
 570         int ret = 0;
 571         struct wim_inode *inode;
 572
 573         if (exclude_path(root_disk_path, config, true)) {
 574                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) {
 575                         ERROR("Cannot exclude the root directory from capture");
 576                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 577                         goto out;
 578                 }
 579                 if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 580                     && progress_func)
 581                 {
 582                         union wimlib_progress_info info;
 583                         info.scan.cur_path = root_disk_path;
 584                         info.scan.excluded = true;
 585                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 586                 }
 587                 goto out;
 588         }
 589
 590         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 591             && progress_func)
 592         {
 593                 union wimlib_progress_info info;
 594                 info.scan.cur_path = root_disk_path;
 595                 info.scan.excluded = false;
 596                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 597         }
 598
 599 #if !defined(__CYGWIN__) && !defined(__WIN32__)
 600         /* UNIX version of capturing a directory tree */
 601         struct stat root_stbuf;
 602         int (*stat_fn)(const char *restrict, struct stat *restrict);
 603         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)
 604                 stat_fn = stat;
 605         else
 606                 stat_fn = lstat;
 607
 608         ret = (*stat_fn)(root_disk_path, &root_stbuf);
 609         if (ret != 0) {
 610                 ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 611                 goto out;
 612         }
 613
 614         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) &&
 615               !S_ISDIR(root_stbuf.st_mode))
 616         {
 617                 /* Do a dereference-stat in case the root is a symbolic link.
 618                  * This case is allowed, provided that the symbolic link points
 619                  * to a directory. */
 620                 ret = stat(root_disk_path, &root_stbuf);
 621                 if (ret != 0) {
 622                         ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 623                         ret = WIMLIB_ERR_STAT;
 624                         goto out;
 625                 }
 626                 if (!S_ISDIR(root_stbuf.st_mode)) {
 627                         ERROR("`%s' is not a directory", root_disk_path);
 628                         ret = WIMLIB_ERR_NOTDIR;
 629                         goto out;
 630                 }
 631         }
 632         if (!S_ISREG(root_stbuf.st_mode) && !S_ISDIR(root_stbuf.st_mode)
 633             && !S_ISLNK(root_stbuf.st_mode)) {
 634                 ERROR("`%s' is not a regular file, directory, or symbolic link.",
 635                       root_disk_path);
 636                 ret = WIMLIB_ERR_SPECIAL_FILE;
 637                 goto out;
 638         }
 639
 640         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 641         if (!root) {
 642                 if (errno == EILSEQ)
 643                         ret = WIMLIB_ERR_INVALID_UTF8_STRING;
 644                 else if (errno == ENOMEM)
 645                         ret = WIMLIB_ERR_NOMEM;
 646                 else
 647                         ret = WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 648                 goto out;
 649         }
 650
 651         inode = root->d_inode;
 652
 653 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 654         inode->i_creation_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 655         inode->i_last_write_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 656         inode->i_last_access_time = timespec_to_wim_timestamp(&root_stbuf.st_atim);
 657 #else
 658         inode->i_creation_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 659         inode->i_last_write_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 660         inode->i_last_access_time = unix_timestamp_to_wim(root_stbuf.st_atime);
 661 #endif
 662         if (sizeof(ino_t) >= 8)
 663                 inode->i_ino = (u64)root_stbuf.st_ino;
 664         else
 665                 inode->i_ino = (u64)root_stbuf.st_ino |
 666                                    ((u64)root_stbuf.st_dev << ((sizeof(ino_t) * 8) & 63));
 667         inode->i_resolved = 1;
 668         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 669                 ret = inode_set_unix_data(inode, root_stbuf.st_uid,
 670                                           root_stbuf.st_gid,
 671                                           root_stbuf.st_mode,
 672                                           lookup_table,
 673                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 674                 if (ret)
 675                         goto out;
 676         }
 677         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 678         if (S_ISREG(root_stbuf.st_mode)) { /* Archiving a regular file */
 679
 680                 struct wim_lookup_table_entry *lte;
 681                 u8 hash[SHA1_HASH_SIZE];
 682
 683                 inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 684
 685                 /* Empty files do not have to have a lookup table entry. */
 686                 if (root_stbuf.st_size == 0)
 687                         goto out;
 688
 689                 /* For each regular file, we must check to see if the file is in
 690                  * the lookup table already; if it is, we increment its refcnt;
 691                  * otherwise, we create a new lookup table entry and insert it.
 692                  * */
 693
 694                 ret = sha1sum(root_disk_path, hash);
 695                 if (ret != 0)
 696                         goto out;
 697
 698                 lte = __lookup_resource(lookup_table, hash);
 699                 if (lte) {
 700                         lte->refcnt++;
 701                         DEBUG("Add lte reference %u for `%s'", lte->refcnt,
 702                               root_disk_path);
 703                 } else {
 704                         char *file_on_disk = STRDUP(root_disk_path);
 705                         if (!file_on_disk) {
 706                                 ERROR("Failed to allocate memory for file path");
 707                                 ret = WIMLIB_ERR_NOMEM;
 708                                 goto out;
 709                         }
 710                         lte = new_lookup_table_entry();
 711                         if (!lte) {
 712                                 FREE(file_on_disk);
 713                                 ret = WIMLIB_ERR_NOMEM;
 714                                 goto out;
 715                         }
 716                         lte->file_on_disk = file_on_disk;
 717                         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 718                         lte->resource_entry.original_size = root_stbuf.st_size;
 719                         lte->resource_entry.size = root_stbuf.st_size;
 720                         copy_hash(lte->hash, hash);
 721                         lookup_table_insert(lookup_table, lte);
 722                 }
 723                 root->d_inode->i_lte = lte;
 724         } else if (S_ISDIR(root_stbuf.st_mode)) { /* Archiving a directory */
 725
 726                 inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 727
 728                 DIR *dir;
 729                 struct dirent entry, *result;
 730                 struct wim_dentry *child;
 731
 732                 dir = opendir(root_disk_path);
 733                 if (!dir) {
 734                         ERROR_WITH_ERRNO("Failed to open the directory `%s'",
 735                                          root_disk_path);
 736                         ret = WIMLIB_ERR_OPEN;
 737                         goto out;
 738                 }
 739
 740                 /* Buffer for names of files in directory. */
 741                 size_t len = strlen(root_disk_path);
 742                 char name[len + 1 + FILENAME_MAX + 1];
 743                 memcpy(name, root_disk_path, len);
 744                 name[len] = '/';
 745
 746                 /* Create a dentry for each entry in the directory on disk, and recurse
 747                  * to any subdirectories. */
 748                 while (1) {
 749                         errno = 0;
 750                         ret = readdir_r(dir, &entry, &result);
 751                         if (ret != 0) {
 752                                 ret = WIMLIB_ERR_READ;
 753                                 ERROR_WITH_ERRNO("Error reading the "
 754                                                  "directory `%s'",
 755                                                  root_disk_path);
 756                                 break;
 757                         }
 758                         if (result == NULL)
 759                                 break;
 760                         if (result->d_name[0] == '.' && (result->d_name[1] == '\0'
 761                               || (result->d_name[1] == '.' && result->d_name[2] == '\0')))
 762                                         continue;
 763                         strcpy(name + len + 1, result->d_name);
 764                         ret = build_dentry_tree(&child, name, lookup_table,
 765                                                 NULL, config, add_image_flags,
 766                                                 progress_func, NULL);
 767                         if (ret != 0)
 768                                 break;
 769                         if (child)
 770                                 dentry_add_child(root, child);
 771                 }
 772                 closedir(dir);
 773         } else { /* Archiving a symbolic link */
 774                 inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 775                 inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 776
 777                 /* The idea here is to call readlink() to get the UNIX target of
 778                  * the symbolic link, then turn the target into a reparse point
 779                  * data buffer that contains a relative or absolute symbolic
 780                  * link (NOT a junction point or *full* path symbolic link with
 781                  * drive letter).
 782                  */
 783
 784                 char deref_name_buf[4096];
 785                 ssize_t deref_name_len;
 786
 787                 deref_name_len = readlink(root_disk_path, deref_name_buf,
 788                                           sizeof(deref_name_buf) - 1);
 789                 if (deref_name_len >= 0) {
 790                         deref_name_buf[deref_name_len] = '\0';
 791                         DEBUG("Read symlink `%s'", deref_name_buf);
 792                         ret = inode_set_symlink(root->d_inode, deref_name_buf,
 793                                                 lookup_table, NULL);
 794                         if (ret == 0) {
 795                                 /*
 796                                  * Unfortunately, Windows seems to have the
 797                                  * concept of "file" symbolic links as being
 798                                  * different from "directory" symbolic links...
 799                                  * so FILE_ATTRIBUTE_DIRECTORY needs to be set
 800                                  * on the symbolic link if the *target* of the
 801                                  * symbolic link is a directory.
 802                                  */
 803                                 struct stat stbuf;
 804                                 if (stat(root_disk_path, &stbuf) == 0 &&
 805                                     S_ISDIR(stbuf.st_mode))
 806                                 {
 807                                         inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 808                                 }
 809                         }
 810                 } else {
 811                         ERROR_WITH_ERRNO("Failed to read target of "
 812                                          "symbolic link `%s'", root_disk_path);
 813                         ret = WIMLIB_ERR_READLINK;
 814                 }
 815         }
 816 #else
 817         /* Win32 version of capturing a directory tree */
 818
 819         wchar_t *path_utf16;
 820         size_t path_utf16_nchars;
 821         struct sd_set *sd_set;
 822         DWORD err;
 823
 824         if (extra_arg == NULL) {
 825                 sd_set = alloca(sizeof(struct sd_set));
 826                 sd_set->rb_root.rb_node = NULL,
 827                 sd_set->sd = sd;
 828         } else {
 829                 sd_set = extra_arg;
 830         }
 831
 832         DEBUG_WIN32_SAFE("root_disk_path=\"%s\"", root_disk_path);
 833         ret = utf8_to_utf16(root_disk_path, strlen(root_disk_path),
 834                             (char**)&path_utf16, &path_utf16_nchars);
 835         if (ret)
 836                 goto out_destroy_sd_set;
 837         path_utf16_nchars /= sizeof(wchar_t);
 838
 839         DEBUG_WIN32_SAFE("Win32: Opening file `%s'", root_disk_path);
 840         HANDLE hFile = win32_open_file(path_utf16);
 841         if (hFile == INVALID_HANDLE_VALUE) {
 842                 err = GetLastError();
 843                 ERROR("Win32 API: Failed to open \"%s\"", root_disk_path);
 844                 win32_error(err);
 845                 ret = WIMLIB_ERR_OPEN;
 846                 goto out_free_path_utf16;
 847         }
 848
 849         BY_HANDLE_FILE_INFORMATION file_info;
 850         if (!GetFileInformationByHandle(hFile, &file_info)) {
 851                 err = GetLastError();
 852                 ERROR("Win32 API: Failed to get file information for \"%s\"",
 853                       root_disk_path);
 854                 win32_error(err);
 855                 ret = WIMLIB_ERR_STAT;
 856                 goto out_close_handle;
 857         }
 858
 859         /* Create a WIM dentry */
 860         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 861         if (!root) {
 862                 if (errno == EILSEQ)
 863                         ret = WIMLIB_ERR_INVALID_UTF8_STRING;
 864                 else if (errno == ENOMEM)
 865                         ret = WIMLIB_ERR_NOMEM;
 866                 else
 867                         ret = WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 868                 goto out_free_path_utf16;
 869         }
 870
 871         /* Start preparing the associated WIM inode */
 872         inode = root->d_inode;
 873
 874         inode->i_attributes = file_info.dwFileAttributes;
 875         inode->i_creation_time = FILETIME_to_u64(&file_info.ftCreationTime);
 876         inode->i_last_write_time = FILETIME_to_u64(&file_info.ftLastWriteTime);
 877         inode->i_last_access_time = FILETIME_to_u64(&file_info.ftLastAccessTime);
 878         inode->i_ino = ((u64)file_info.nFileIndexHigh << 32) |
 879                         (u64)file_info.nFileIndexLow;
 880
 881         inode->i_resolved = 1;
 882         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 883
 884         /* Get DOS name and security descriptor (if any). */
 885         ret = win32_get_short_name(root, path_utf16);
 886         if (ret)
 887                 goto out_close_handle;
 888         ret = win32_get_security_descriptor(root, sd_set, path_utf16,
 889                                             root_disk_path);
 890         if (ret)
 891                 goto out_close_handle;
 892
 893         if (inode_is_directory(inode)) {
 894                 /* Directory (not a reparse point) --- recurse to children */
 895                 DEBUG_WIN32_SAFE("Recursing to directory \"%s\"", root_disk_path);
 896                 ret = win32_recurse_directory(root,
 897                                               root_disk_path,
 898                                               lookup_table,
 899                                               sd,
 900                                               config,
 901                                               add_image_flags,
 902                                               progress_func,
 903                                               sd_set,
 904                                               path_utf16,
 905                                               path_utf16_nchars);
 906         } else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 907                 /* Reparse point: save the reparse tag and data */
 908
 909                 DEBUG_WIN32_SAFE("Capturing reparse point `%s'", root_disk_path);
 910                 ret = win32_capture_reparse_point(root_disk_path,
 911                                                   hFile,
 912                                                   inode,
 913                                                   lookup_table);
 914
 915         } else {
 916                 DEBUG_WIN32_SAFE("Capturing streams of \"%s\"", root_disk_path);
 917                 /* Not a directory, not a reparse point */
 918                 ret = win32_capture_streams(root_disk_path,
 919                                             path_utf16,
 920                                             path_utf16_nchars,
 921                                             inode,
 922                                             lookup_table);
 923         }
 924 out_close_handle:
 925         CloseHandle(hFile);
 926 out_destroy_sd_set:
 927         if (extra_arg == NULL)
 928                 destroy_sd_set(sd_set);
 929 out_free_path_utf16:
 930         FREE(path_utf16);
 931 #endif
 932 out:
 933         if (ret == 0)
 934                 *root_ret = root;
 935         else
 936                 free_dentry_tree(root, lookup_table);
 937         return ret;
 938 }
 939
 940 enum pattern_type {
 941         NONE = 0,
 942         EXCLUSION_LIST,
 943         EXCLUSION_EXCEPTION,
 944         COMPRESSION_EXCLUSION_LIST,
 945         ALIGNMENT_LIST,
 946 };
 947
 948 #define COMPAT_DEFAULT_CONFIG
 949
 950 /* Default capture configuration file when none is specified. */
 951 static const char *default_config =
 952 #ifdef COMPAT_DEFAULT_CONFIG /* XXX: This policy is being moved to library
 953                                 users.  The next ABI-incompatible library
 954                                 version will default to the empty string here. */
 955 "[ExclusionList]\n"
 956 "\\$ntfs.log\n"
 957 "\\hiberfil.sys\n"
 958 "\\pagefile.sys\n"
 959 "\\System Volume Information\n"
 960 "\\RECYCLER\n"
 961 "\\Windows\\CSC\n"
 962 "\n"
 963 "[CompressionExclusionList]\n"
 964 "*.mp3\n"
 965 "*.zip\n"
 966 "*.cab\n"
 967 "\\WINDOWS\\inf\\*.pnf\n";
 968 #else
 969 "";
 970 #endif
 971
 972 static void destroy_pattern_list(struct pattern_list *list)
 973 {
 974         FREE(list->pats);
 975 }
 976
 977 static void destroy_capture_config(struct capture_config *config)
 978 {
 979         destroy_pattern_list(&config->exclusion_list);
 980         destroy_pattern_list(&config->exclusion_exception);
 981         destroy_pattern_list(&config->compression_exclusion_list);
 982         destroy_pattern_list(&config->alignment_list);
 983         FREE(config->config_str);
 984         FREE(config->prefix);
 985         memset(config, 0, sizeof(*config));
 986 }
 987
 988 static int pattern_list_add_pattern(struct pattern_list *list,
 989                                     const char *pattern)
 990 {
 991         const char **pats;
 992         if (list->num_pats >= list->num_allocated_pats) {
 993                 pats = REALLOC(list->pats,
 994                                sizeof(list->pats[0]) * (list->num_allocated_pats + 8));
 995                 if (!pats)
 996                         return WIMLIB_ERR_NOMEM;
 997                 list->num_allocated_pats += 8;
 998                 list->pats = pats;
 999         }
1000         list->pats[list->num_pats++] = pattern;
1001         return 0;
1002 }
1003
1004 /* Parses the contents of the image capture configuration file and fills in a
1005  * `struct capture_config'. */
1006 static int init_capture_config(struct capture_config *config,
1007                                const char *_config_str, size_t config_len)
1008 {
1009         char *config_str;
1010         char *p;
1011         char *eol;
1012         char *next_p;
1013         size_t bytes_remaining;
1014         enum pattern_type type = NONE;
1015         int ret;
1016         unsigned long line_no = 0;
1017
1018         DEBUG("config_len = %zu", config_len);
1019         bytes_remaining = config_len;
1020         memset(config, 0, sizeof(*config));
1021         config_str = MALLOC(config_len);
1022         if (!config_str) {
1023                 ERROR("Could not duplicate capture config string");
1024                 return WIMLIB_ERR_NOMEM;
1025         }
1026
1027         memcpy(config_str, _config_str, config_len);
1028         next_p = config_str;
1029         config->config_str = config_str;
1030         while (bytes_remaining) {
1031                 line_no++;
1032                 p = next_p;
1033                 eol = memchr(p, '\n', bytes_remaining);
1034                 if (!eol) {
1035                         ERROR("Expected end-of-line in capture config file on "
1036                               "line %lu", line_no);
1037                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
1038                         goto out_destroy;
1039                 }
1040
1041                 next_p = eol + 1;
1042                 bytes_remaining -= (next_p - p);
1043                 if (eol == p)
1044                         continue;
1045
1046                 if (*(eol - 1) == '\r')
1047                         eol--;
1048                 *eol = '\0';
1049
1050                 /* Translate backslash to forward slash */
1051                 for (char *pp = p; pp != eol; pp++)
1052                         if (*pp == '\\')
1053                                 *pp = '/';
1054
1055                 /* Remove drive letter */
1056                 if (eol - p > 2 && isalpha(*p) && *(p + 1) == ':')
1057                         p += 2;
1058
1059                 ret = 0;
1060                 if (strcmp(p, "[ExclusionList]") == 0)
1061                         type = EXCLUSION_LIST;
1062                 else if (strcmp(p, "[ExclusionException]") == 0)
1063                         type = EXCLUSION_EXCEPTION;
1064                 else if (strcmp(p, "[CompressionExclusionList]") == 0)
1065                         type = COMPRESSION_EXCLUSION_LIST;
1066                 else if (strcmp(p, "[AlignmentList]") == 0)
1067                         type = ALIGNMENT_LIST;
1068                 else if (p[0] == '[' && strrchr(p, ']')) {
1069                         ERROR("Unknown capture configuration section `%s'", p);
1070                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
1071                 } else switch (type) {
1072                 case EXCLUSION_LIST:
1073                         DEBUG("Adding pattern \"%s\" to exclusion list", p);
1074                         ret = pattern_list_add_pattern(&config->exclusion_list, p);
1075                         break;
1076                 case EXCLUSION_EXCEPTION:
1077                         DEBUG("Adding pattern \"%s\" to exclusion exception list", p);
1078                         ret = pattern_list_add_pattern(&config->exclusion_exception, p);
1079                         break;
1080                 case COMPRESSION_EXCLUSION_LIST:
1081                         DEBUG("Adding pattern \"%s\" to compression exclusion list", p);
1082                         ret = pattern_list_add_pattern(&config->compression_exclusion_list, p);
1083                         break;
1084                 case ALIGNMENT_LIST:
1085                         DEBUG("Adding pattern \"%s\" to alignment list", p);
1086                         ret = pattern_list_add_pattern(&config->alignment_list, p);
1087                         break;
1088                 default:
1089                         ERROR("Line %lu of capture configuration is not "
1090                               "in a block (such as [ExclusionList])",
1091                               line_no);
1092                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
1093                         break;
1094                 }
1095                 if (ret != 0)
1096                         goto out_destroy;
1097         }
1098         return 0;
1099 out_destroy:
1100         destroy_capture_config(config);
1101         return ret;
1102 }
1103
1104 static int capture_config_set_prefix(struct capture_config *config,
1105                                      const char *_prefix)
1106 {
1107         char *prefix = STRDUP(_prefix);
1108
1109         if (!prefix)
1110                 return WIMLIB_ERR_NOMEM;
1111         FREE(config->prefix);
1112         config->prefix = prefix;
1113         config->prefix_len = strlen(prefix);
1114         return 0;
1115 }
1116
1117 static bool match_pattern(const char *path, const char *path_basename,
1118                           const struct pattern_list *list)
1119 {
1120         for (size_t i = 0; i < list->num_pats; i++) {
1121                 const char *pat = list->pats[i];
1122                 const char *string;
1123                 if (pat[0] == '/')
1124                         /* Absolute path from root of capture */
1125                         string = path;
1126                 else {
1127                         if (strchr(pat, '/'))
1128                                 /* Relative path from root of capture */
1129                                 string = path + 1;
1130                         else
1131                                 /* A file name pattern */
1132                                 string = path_basename;
1133                 }
1134                 if (fnmatch(pat, string, FNM_PATHNAME
1135                         #ifdef FNM_CASEFOLD
1136                                         | FNM_CASEFOLD
1137                         #endif
1138                         ) == 0)
1139                 {
1140                         DEBUG("`%s' matches the pattern \"%s\"",
1141                               string, pat);
1142                         return true;
1143                 }
1144         }
1145         return false;
1146 }
1147
1148 /* Return true if the image capture configuration file indicates we should
1149  * exclude the filename @path from capture.
1150  *
1151  * If @exclude_prefix is %true, the part of the path up and including the name
1152  * of the directory being captured is not included in the path for matching
1153  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
1154  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
1155  * directory.
1156  */
1157 bool exclude_path(const char *path, const struct capture_config *config,
1158                   bool exclude_prefix)
1159 {
1160         const char *basename = path_basename(path);
1161         if (exclude_prefix) {
1162                 wimlib_assert(strlen(path) >= config->prefix_len);
1163                 if (memcmp(config->prefix, path, config->prefix_len) == 0
1164                      && path[config->prefix_len] == '/')
1165                         path += config->prefix_len;
1166         }
1167         return match_pattern(path, basename, &config->exclusion_list) &&
1168                 !match_pattern(path, basename, &config->exclusion_exception);
1169
1170 }
1171
1172 /* Strip leading and trailing forward slashes from a string.  Modifies it in
1173  * place and returns the stripped string. */
1174 static const char *canonicalize_target_path(char *target_path)
1175 {
1176         char *p;
1177         if (target_path == NULL)
1178                 target_path = "";
1179         for (;;) {
1180                 if (*target_path == '\0')
1181                         return target_path;
1182                 else if (*target_path == '/')
1183                         target_path++;
1184                 else
1185                         break;
1186         }
1187
1188         p = target_path + strlen(target_path) - 1;
1189         while (*p == '/')
1190                 *p-- = '\0';
1191         return target_path;
1192 }
1193
1194 /* Strip leading and trailing slashes from the target paths */
1195 static void canonicalize_targets(struct wimlib_capture_source *sources,
1196                                  size_t num_sources)
1197 {
1198         while (num_sources--) {
1199                 DEBUG("Canonicalizing { source: \"%s\", target=\"%s\"}",
1200                       sources->fs_source_path,
1201                       sources->wim_target_path);
1202                 sources->wim_target_path =
1203                         (char*)canonicalize_target_path(sources->wim_target_path);
1204                 DEBUG("Canonical target: \"%s\"", sources->wim_target_path);
1205                 sources++;
1206         }
1207 }
1208
1209 static int capture_source_cmp(const void *p1, const void *p2)
1210 {
1211         const struct wimlib_capture_source *s1 = p1, *s2 = p2;
1212         return strcmp(s1->wim_target_path, s2->wim_target_path);
1213 }
1214
1215 /* Sorts the capture sources lexicographically by target path.  This occurs
1216  * after leading and trailing forward slashes are stripped.
1217  *
1218  * One purpose of this is to make sure that target paths that are inside other
1219  * target paths are extracted after the containing target paths. */
1220 static void sort_sources(struct wimlib_capture_source *sources,
1221                          size_t num_sources)
1222 {
1223         qsort(sources, num_sources, sizeof(sources[0]), capture_source_cmp);
1224 }
1225
1226 static int check_sorted_sources(struct wimlib_capture_source *sources,
1227                                 size_t num_sources, int add_image_flags)
1228 {
1229         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
1230                 if (num_sources != 1) {
1231                         ERROR("Must specify exactly 1 capture source "
1232                               "(the NTFS volume) in NTFS mode!");
1233                         return WIMLIB_ERR_INVALID_PARAM;
1234                 }
1235                 if (sources[0].wim_target_path[0] != '\0') {
1236                         ERROR("In NTFS capture mode the target path inside "
1237                               "the image must be the root directory!");
1238                         return WIMLIB_ERR_INVALID_PARAM;
1239                 }
1240         } else if (num_sources != 0) {
1241                 /* This code is disabled because the current code
1242                  * unconditionally attempts to do overlays.  So, duplicate
1243                  * target paths are OK. */
1244         #if 0
1245                 if (num_sources > 1 && sources[0].wim_target_path[0] == '\0') {
1246                         ERROR("Cannot specify root target when using multiple "
1247                               "capture sources!");
1248                         return WIMLIB_ERR_INVALID_PARAM;
1249                 }
1250                 for (size_t i = 0; i < num_sources - 1; i++) {
1251                         size_t len = strlen(sources[i].wim_target_path);
1252                         size_t j = i + 1;
1253                         const char *target1 = sources[i].wim_target_path;
1254                         do {
1255                                 const char *target2 = sources[j].wim_target_path;
1256                                 DEBUG("target1=%s, target2=%s",
1257                                       target1,target2);
1258                                 if (strncmp(target1, target2, len) ||
1259                                     target2[len] > '/')
1260                                         break;
1261                                 if (target2[len] == '/') {
1262                                         ERROR("Invalid target `%s': is a prefix of `%s'",
1263                                               target1, target2);
1264                                         return WIMLIB_ERR_INVALID_PARAM;
1265                                 }
1266                                 if (target2[len] == '\0') {
1267                                         ERROR("Invalid target `%s': is a duplicate of `%s'",
1268                                               target1, target2);
1269                                         return WIMLIB_ERR_INVALID_PARAM;
1270                                 }
1271                         } while (++j != num_sources);
1272                 }
1273         #endif
1274         }
1275         return 0;
1276
1277 }
1278
1279 /* Creates a new directory to place in the WIM image.  This is to create parent
1280  * directories that are not part of any target as needed.  */
1281 static struct wim_dentry *
1282 new_filler_directory(const char *name)
1283 {
1284         struct wim_dentry *dentry;
1285         DEBUG("Creating filler directory \"%s\"", name);
1286         dentry = new_dentry_with_inode(name);
1287         if (dentry) {
1288                 /* Set the inode number to 0 for now.  The final inode number
1289                  * will be assigned later by assign_inode_numbers(). */
1290                 dentry->d_inode->i_ino = 0;
1291                 dentry->d_inode->i_resolved = 1;
1292                 dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
1293         }
1294         return dentry;
1295 }
1296
1297 /* Transfers the children of @branch to @target.  It is an error if @target is
1298  * not a directory or if both @branch and @target contain a child dentry with
1299  * the same name. */
1300 static int do_overlay(struct wim_dentry *target, struct wim_dentry *branch)
1301 {
1302         struct rb_root *rb_root;
1303
1304         if (!dentry_is_directory(target)) {
1305                 ERROR("Cannot overlay directory `%s' over non-directory",
1306                       branch->file_name_utf8);
1307                 return WIMLIB_ERR_INVALID_OVERLAY;
1308         }
1309
1310         rb_root = &branch->d_inode->i_children;
1311         while (rb_root->rb_node) { /* While @branch has children... */
1312                 struct wim_dentry *child = rbnode_dentry(rb_root->rb_node);
1313                 /* Move @child to the directory @target */
1314                 unlink_dentry(child);
1315                 if (!dentry_add_child(target, child)) {
1316                         /* Revert the change to avoid leaking the directory tree
1317                          * rooted at @child */
1318                         dentry_add_child(branch, child);
1319                         ERROR("Overlay error: file `%s' already exists "
1320                               "as a child of `%s'",
1321                               child->file_name_utf8, target->file_name_utf8);
1322                         return WIMLIB_ERR_INVALID_OVERLAY;
1323                 }
1324         }
1325         return 0;
1326
1327 }
1328
1329 /* Attach or overlay a branch onto the WIM image.
1330  *
1331  * @root_p:
1332  *      Pointer to the root of the WIM image, or pointer to NULL if it has not
1333  *      been created yet.
1334  * @branch
1335  *      Branch to add.
1336  * @target_path:
1337  *      Path in the WIM image to add the branch, with leading and trailing
1338  *      slashes stripped.
1339  */
1340 static int attach_branch(struct wim_dentry **root_p,
1341                          struct wim_dentry *branch,
1342                          char *target_path)
1343 {
1344         char *slash;
1345         struct wim_dentry *dentry, *parent, *target;
1346
1347         if (*target_path == '\0') {
1348                 /* Target: root directory */
1349                 if (*root_p) {
1350                         /* Overlay on existing root */
1351                         return do_overlay(*root_p, branch);
1352                 } else  {
1353                         /* Set as root */
1354                         *root_p = branch;
1355                         return 0;
1356                 }
1357         }
1358
1359         /* Adding a non-root branch.  Create root if it hasn't been created
1360          * already. */
1361         if (!*root_p) {
1362                 *root_p = new_filler_directory("");
1363                 if (!*root_p)
1364                         return WIMLIB_ERR_NOMEM;
1365         }
1366
1367         /* Walk the path to the branch, creating filler directories as needed.
1368          * */
1369         parent = *root_p;
1370         while ((slash = strchr(target_path, '/'))) {
1371                 *slash = '\0';
1372                 dentry = get_dentry_child_with_name(parent, target_path);
1373                 if (!dentry) {
1374                         dentry = new_filler_directory(target_path);
1375                         if (!dentry)
1376                                 return WIMLIB_ERR_NOMEM;
1377                         dentry_add_child(parent, dentry);
1378                 }
1379                 parent = dentry;
1380                 target_path = slash;
1381                 /* Skip over slashes.  Note: this cannot overrun the length of
1382                  * the string because the last character cannot be a slash, as
1383                  * trailing slashes were tripped.  */
1384                 do {
1385                         ++target_path;
1386                 } while (*target_path == '/');
1387         }
1388
1389         /* If the target path already existed, overlay the branch onto it.
1390          * Otherwise, set the branch as the target path. */
1391         target = get_dentry_child_with_name(parent, branch->file_name_utf8);
1392         if (target) {
1393                 return do_overlay(target, branch);
1394         } else {
1395                 dentry_add_child(parent, branch);
1396                 return 0;
1397         }
1398 }
1399
1400 WIMLIBAPI int wimlib_add_image_multisource(WIMStruct *w,
1401                                            struct wimlib_capture_source *sources,
1402                                            size_t num_sources,
1403                                            const char *name,
1404                                            const char *config_str,
1405                                            size_t config_len,
1406                                            int add_image_flags,
1407                                            wimlib_progress_func_t progress_func)
1408 {
1409         int (*capture_tree)(struct wim_dentry **, const char *,
1410                             struct wim_lookup_table *,
1411                             struct wim_security_data *,
1412                             const struct capture_config *,
1413                             int, wimlib_progress_func_t, void *);
1414         void *extra_arg;
1415         struct wim_dentry *root_dentry;
1416         struct wim_dentry *branch;
1417         struct wim_security_data *sd;
1418         struct capture_config config;
1419         struct wim_image_metadata *imd;
1420         int ret;
1421
1422         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
1423 #ifdef WITH_NTFS_3G
1424                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
1425                         ERROR("Cannot dereference files when capturing directly from NTFS");
1426                         return WIMLIB_ERR_INVALID_PARAM;
1427                 }
1428                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
1429                         ERROR("Capturing UNIX owner and mode not supported "
1430                               "when capturing directly from NTFS");
1431                         return WIMLIB_ERR_INVALID_PARAM;
1432                 }
1433                 capture_tree = build_dentry_tree_ntfs;
1434                 extra_arg = &w->ntfs_vol;
1435 #else
1436                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1437                       "        cannot capture a WIM image directly from a NTFS volume!");
1438                 return WIMLIB_ERR_UNSUPPORTED;
1439 #endif
1440         } else {
1441                 capture_tree = build_dentry_tree;
1442                 extra_arg = NULL;
1443         }
1444
1445         if (!name || !*name) {
1446                 ERROR("Must specify a non-empty string for the image name");
1447                 return WIMLIB_ERR_INVALID_PARAM;
1448         }
1449
1450         if (w->hdr.total_parts != 1) {
1451                 ERROR("Cannot add an image to a split WIM");
1452                 return WIMLIB_ERR_SPLIT_UNSUPPORTED;
1453         }
1454
1455         if (wimlib_image_name_in_use(w, name)) {
1456                 ERROR("There is already an image named \"%s\" in `%s'",
1457                       name, w->filename);
1458                 return WIMLIB_ERR_IMAGE_NAME_COLLISION;
1459         }
1460
1461         if (!config_str) {
1462                 DEBUG("Using default capture configuration");
1463                 config_str = default_config;
1464                 config_len = strlen(default_config);
1465         }
1466         ret = init_capture_config(&config, config_str, config_len);
1467         if (ret)
1468                 goto out;
1469
1470         DEBUG("Allocating security data");
1471         sd = CALLOC(1, sizeof(struct wim_security_data));
1472         if (!sd) {
1473                 ret = WIMLIB_ERR_NOMEM;
1474                 goto out_destroy_capture_config;
1475         }
1476         sd->total_length = 8;
1477         sd->refcnt = 1;
1478
1479         DEBUG("Using %zu capture sources", num_sources);
1480         canonicalize_targets(sources, num_sources);
1481         sort_sources(sources, num_sources);
1482         ret = check_sorted_sources(sources, num_sources, add_image_flags);
1483         if (ret) {
1484                 ret = WIMLIB_ERR_INVALID_PARAM;
1485                 goto out_free_security_data;
1486         }
1487
1488         DEBUG("Building dentry tree.");
1489         if (num_sources == 0) {
1490                 root_dentry = new_filler_directory("");
1491                 if (!root_dentry)
1492                         goto out_free_security_data;
1493         } else {
1494                 size_t i;
1495
1496                 root_dentry = NULL;
1497                 i = 0;
1498                 do {
1499                         int flags;
1500                         union wimlib_progress_info progress;
1501
1502                         DEBUG("Building dentry tree for source %zu of %zu "
1503                               "(\"%s\" => \"%s\")", i + 1, num_sources,
1504                               sources[i].fs_source_path,
1505                               sources[i].wim_target_path);
1506                         if (progress_func) {
1507                                 memset(&progress, 0, sizeof(progress));
1508                                 progress.scan.source = sources[i].fs_source_path;
1509                                 progress.scan.wim_target_path = sources[i].wim_target_path;
1510                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_BEGIN, &progress);
1511                         }
1512                         ret = capture_config_set_prefix(&config,
1513                                                         sources[i].fs_source_path);
1514                         if (ret)
1515                                 goto out_free_dentry_tree;
1516                         flags = add_image_flags | WIMLIB_ADD_IMAGE_FLAG_SOURCE;
1517                         if (!*sources[i].wim_target_path)
1518                                 flags |= WIMLIB_ADD_IMAGE_FLAG_ROOT;
1519                         ret = (*capture_tree)(&branch, sources[i].fs_source_path,
1520                                               w->lookup_table, sd,
1521                                               &config,
1522                                               flags,
1523                                               progress_func, extra_arg);
1524                         if (ret) {
1525                                 ERROR("Failed to build dentry tree for `%s'",
1526                                       sources[i].fs_source_path);
1527                                 goto out_free_dentry_tree;
1528                         }
1529                         if (branch) {
1530                                 /* Use the target name, not the source name, for
1531                                  * the root of each branch from a capture
1532                                  * source.  (This will also set the root dentry
1533                                  * of the entire image to be unnamed.) */
1534                                 ret = set_dentry_name(branch,
1535                                                       path_basename(sources[i].wim_target_path));
1536                                 if (ret)
1537                                         goto out_free_branch;
1538
1539                                 ret = attach_branch(&root_dentry, branch,
1540                                                     sources[i].wim_target_path);
1541                                 if (ret)
1542                                         goto out_free_branch;
1543                         }
1544                         if (progress_func)
1545                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_END, &progress);
1546                 } while (++i != num_sources);
1547         }
1548
1549         DEBUG("Calculating full paths of dentries.");
1550         ret = for_dentry_in_tree(root_dentry, calculate_dentry_full_path, NULL);
1551         if (ret != 0)
1552                 goto out_free_dentry_tree;
1553
1554         ret = add_new_dentry_tree(w, root_dentry, sd);
1555         if (ret != 0)
1556                 goto out_free_dentry_tree;
1557
1558         imd = &w->image_metadata[w->hdr.image_count - 1];
1559
1560         ret = dentry_tree_fix_inodes(root_dentry, &imd->inode_list);
1561         if (ret != 0)
1562                 goto out_destroy_imd;
1563
1564         DEBUG("Assigning hard link group IDs");
1565         assign_inode_numbers(&imd->inode_list);
1566
1567         ret = xml_add_image(w, name);
1568         if (ret != 0)
1569                 goto out_destroy_imd;
1570
1571         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_BOOT)
1572                 wimlib_set_boot_idx(w, w->hdr.image_count);
1573         ret = 0;
1574         goto out;
1575 out_destroy_imd:
1576         destroy_image_metadata(&w->image_metadata[w->hdr.image_count - 1],
1577                                w->lookup_table);
1578         w->hdr.image_count--;
1579         goto out;
1580 out_free_branch:
1581         free_dentry_tree(branch, w->lookup_table);
1582 out_free_dentry_tree:
1583         free_dentry_tree(root_dentry, w->lookup_table);
1584 out_free_security_data:
1585         free_security_data(sd);
1586 out_destroy_capture_config:
1587         destroy_capture_config(&config);
1588 out:
1589         return ret;
1590 }
1591
1592 WIMLIBAPI int wimlib_add_image(WIMStruct *w, const char *source,
1593                                const char *name, const char *config_str,
1594                                size_t config_len, int add_image_flags,
1595                                wimlib_progress_func_t progress_func)
1596 {
1597         if (!source || !*source)
1598                 return WIMLIB_ERR_INVALID_PARAM;
1599
1600         char *fs_source_path = STRDUP(source);
1601         int ret;
1602         struct wimlib_capture_source capture_src = {
1603                 .fs_source_path = fs_source_path,
1604                 .wim_target_path = NULL,
1605                 .reserved = 0,
1606         };
1607         ret = wimlib_add_image_multisource(w, &capture_src, 1, name,
1608                                            config_str, config_len,
1609                                            add_image_flags, progress_func);
1610         FREE(fs_source_path);
1611         return ret;
1612 }