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                 ERROR("Win32 API: Failed to look up data streams of \"%s\"",
 432                       path);
 433                 return WIMLIB_ERR_READ;
 434         }
 435         do {
 436                 ret = win32_capture_stream(path, path_utf16,
 437                                            path_utf16_nchars,
 438                                            inode, lookup_table,
 439                                            &dat);
 440                 if (ret)
 441                         goto out_find_close;
 442         } while (FindNextStreamW(hFind, &dat));
 443         err = GetLastError();
 444         if (err != ERROR_HANDLE_EOF) {
 445                 ERROR("Win32 API: Error reading data streams from \"%s\"", path);
 446                 win32_error(err);
 447                 ret = WIMLIB_ERR_READ;
 448         }
 449 out_find_close:
 450         FindClose(hFind);
 451         return ret;
 452 }
 453 #endif
 454
 455 /*
 456  * build_dentry_tree():
 457  *      Recursively builds a tree of WIM dentries from an on-disk directory
 458  *      tree.
 459  *
 460  * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
 461  *              modified if successful.  Set to NULL if the file or directory was
 462  *              excluded from capture.
 463  *
 464  * @root_disk_path:  The path to the root of the directory tree on disk.
 465  *
 466  * @lookup_table: The lookup table for the WIM file.  For each file added to the
 467  *              dentry tree being built, an entry is added to the lookup table,
 468  *              unless an identical stream is already in the lookup table.
 469  *              These lookup table entries that are added point to the path of
 470  *              the file on disk.
 471  *
 472  * @sd:         Ignored.  (Security data only captured in NTFS mode.)
 473  *
 474  * @capture_config:
 475  *              Configuration for files to be excluded from capture.
 476  *
 477  * @add_flags:  Bitwise or of WIMLIB_ADD_IMAGE_FLAG_*
 478  *
 479  * @extra_arg:  Ignored. (Only used in NTFS mode.)
 480  *
 481  * @return:     0 on success, nonzero on failure.  It is a failure if any of
 482  *              the files cannot be `stat'ed, or if any of the needed
 483  *              directories cannot be opened or read.  Failure to add the files
 484  *              to the WIM may still occur later when trying to actually read
 485  *              the on-disk files during a call to wimlib_write() or
 486  *              wimlib_overwrite().
 487  */
 488 static int build_dentry_tree(struct wim_dentry **root_ret,
 489                              const char *root_disk_path,
 490                              struct wim_lookup_table *lookup_table,
 491                              struct wim_security_data *sd,
 492                              const struct capture_config *config,
 493                              int add_image_flags,
 494                              wimlib_progress_func_t progress_func,
 495                              void *extra_arg)
 496 {
 497         struct wim_dentry *root = NULL;
 498         int ret = 0;
 499         struct wim_inode *inode;
 500
 501         if (exclude_path(root_disk_path, config, true)) {
 502                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) {
 503                         ERROR("Cannot exclude the root directory from capture");
 504                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 505                         goto out;
 506                 }
 507                 if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 508                     && progress_func)
 509                 {
 510                         union wimlib_progress_info info;
 511                         info.scan.cur_path = root_disk_path;
 512                         info.scan.excluded = true;
 513                         progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 514                 }
 515                 goto out;
 516         }
 517
 518         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_VERBOSE)
 519             && progress_func)
 520         {
 521                 union wimlib_progress_info info;
 522                 info.scan.cur_path = root_disk_path;
 523                 info.scan.excluded = false;
 524                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_DENTRY, &info);
 525         }
 526
 527 #if !defined(__CYGWIN__) && !defined(__WIN32__)
 528         /* UNIX version of capturing a directory tree */
 529         struct stat root_stbuf;
 530         int (*stat_fn)(const char *restrict, struct stat *restrict);
 531         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE)
 532                 stat_fn = stat;
 533         else
 534                 stat_fn = lstat;
 535
 536         ret = (*stat_fn)(root_disk_path, &root_stbuf);
 537         if (ret != 0) {
 538                 ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 539                 goto out;
 540         }
 541
 542         if ((add_image_flags & WIMLIB_ADD_IMAGE_FLAG_ROOT) &&
 543               !S_ISDIR(root_stbuf.st_mode))
 544         {
 545                 /* Do a dereference-stat in case the root is a symbolic link.
 546                  * This case is allowed, provided that the symbolic link points
 547                  * to a directory. */
 548                 ret = stat(root_disk_path, &root_stbuf);
 549                 if (ret != 0) {
 550                         ERROR_WITH_ERRNO("Failed to stat `%s'", root_disk_path);
 551                         ret = WIMLIB_ERR_STAT;
 552                         goto out;
 553                 }
 554                 if (!S_ISDIR(root_stbuf.st_mode)) {
 555                         ERROR("`%s' is not a directory", root_disk_path);
 556                         ret = WIMLIB_ERR_NOTDIR;
 557                         goto out;
 558                 }
 559         }
 560         if (!S_ISREG(root_stbuf.st_mode) && !S_ISDIR(root_stbuf.st_mode)
 561             && !S_ISLNK(root_stbuf.st_mode)) {
 562                 ERROR("`%s' is not a regular file, directory, or symbolic link.",
 563                       root_disk_path);
 564                 ret = WIMLIB_ERR_SPECIAL_FILE;
 565                 goto out;
 566         }
 567
 568         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 569         if (!root) {
 570                 if (errno == EILSEQ)
 571                         ret = WIMLIB_ERR_INVALID_UTF8_STRING;
 572                 else if (errno == ENOMEM)
 573                         ret = WIMLIB_ERR_NOMEM;
 574                 else
 575                         ret = WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 576                 goto out;
 577         }
 578
 579         inode = root->d_inode;
 580
 581 #ifdef HAVE_STAT_NANOSECOND_PRECISION
 582         inode->i_creation_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 583         inode->i_last_write_time = timespec_to_wim_timestamp(&root_stbuf.st_mtim);
 584         inode->i_last_access_time = timespec_to_wim_timestamp(&root_stbuf.st_atim);
 585 #else
 586         inode->i_creation_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 587         inode->i_last_write_time = unix_timestamp_to_wim(root_stbuf.st_mtime);
 588         inode->i_last_access_time = unix_timestamp_to_wim(root_stbuf.st_atime);
 589 #endif
 590         if (sizeof(ino_t) >= 8)
 591                 inode->i_ino = (u64)root_stbuf.st_ino;
 592         else
 593                 inode->i_ino = (u64)root_stbuf.st_ino |
 594                                    ((u64)root_stbuf.st_dev << ((sizeof(ino_t) * 8) & 63));
 595         inode->i_resolved = 1;
 596         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
 597                 ret = inode_set_unix_data(inode, root_stbuf.st_uid,
 598                                           root_stbuf.st_gid,
 599                                           root_stbuf.st_mode,
 600                                           lookup_table,
 601                                           UNIX_DATA_ALL | UNIX_DATA_CREATE);
 602                 if (ret)
 603                         goto out;
 604         }
 605         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 606         if (S_ISREG(root_stbuf.st_mode)) { /* Archiving a regular file */
 607
 608                 struct wim_lookup_table_entry *lte;
 609                 u8 hash[SHA1_HASH_SIZE];
 610
 611                 inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 612
 613                 /* Empty files do not have to have a lookup table entry. */
 614                 if (root_stbuf.st_size == 0)
 615                         goto out;
 616
 617                 /* For each regular file, we must check to see if the file is in
 618                  * the lookup table already; if it is, we increment its refcnt;
 619                  * otherwise, we create a new lookup table entry and insert it.
 620                  * */
 621
 622                 ret = sha1sum(root_disk_path, hash);
 623                 if (ret != 0)
 624                         goto out;
 625
 626                 lte = __lookup_resource(lookup_table, hash);
 627                 if (lte) {
 628                         lte->refcnt++;
 629                         DEBUG("Add lte reference %u for `%s'", lte->refcnt,
 630                               root_disk_path);
 631                 } else {
 632                         char *file_on_disk = STRDUP(root_disk_path);
 633                         if (!file_on_disk) {
 634                                 ERROR("Failed to allocate memory for file path");
 635                                 ret = WIMLIB_ERR_NOMEM;
 636                                 goto out;
 637                         }
 638                         lte = new_lookup_table_entry();
 639                         if (!lte) {
 640                                 FREE(file_on_disk);
 641                                 ret = WIMLIB_ERR_NOMEM;
 642                                 goto out;
 643                         }
 644                         lte->file_on_disk = file_on_disk;
 645                         lte->resource_location = RESOURCE_IN_FILE_ON_DISK;
 646                         lte->resource_entry.original_size = root_stbuf.st_size;
 647                         lte->resource_entry.size = root_stbuf.st_size;
 648                         copy_hash(lte->hash, hash);
 649                         lookup_table_insert(lookup_table, lte);
 650                 }
 651                 root->d_inode->i_lte = lte;
 652         } else if (S_ISDIR(root_stbuf.st_mode)) { /* Archiving a directory */
 653
 654                 inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 655
 656                 DIR *dir;
 657                 struct dirent entry, *result;
 658                 struct wim_dentry *child;
 659
 660                 dir = opendir(root_disk_path);
 661                 if (!dir) {
 662                         ERROR_WITH_ERRNO("Failed to open the directory `%s'",
 663                                          root_disk_path);
 664                         ret = WIMLIB_ERR_OPEN;
 665                         goto out;
 666                 }
 667
 668                 /* Buffer for names of files in directory. */
 669                 size_t len = strlen(root_disk_path);
 670                 char name[len + 1 + FILENAME_MAX + 1];
 671                 memcpy(name, root_disk_path, len);
 672                 name[len] = '/';
 673
 674                 /* Create a dentry for each entry in the directory on disk, and recurse
 675                  * to any subdirectories. */
 676                 while (1) {
 677                         errno = 0;
 678                         ret = readdir_r(dir, &entry, &result);
 679                         if (ret != 0) {
 680                                 ret = WIMLIB_ERR_READ;
 681                                 ERROR_WITH_ERRNO("Error reading the "
 682                                                  "directory `%s'",
 683                                                  root_disk_path);
 684                                 break;
 685                         }
 686                         if (result == NULL)
 687                                 break;
 688                         if (result->d_name[0] == '.' && (result->d_name[1] == '\0'
 689                               || (result->d_name[1] == '.' && result->d_name[2] == '\0')))
 690                                         continue;
 691                         strcpy(name + len + 1, result->d_name);
 692                         ret = build_dentry_tree(&child, name, lookup_table,
 693                                                 NULL, config, add_image_flags,
 694                                                 progress_func, NULL);
 695                         if (ret != 0)
 696                                 break;
 697                         if (child)
 698                                 dentry_add_child(root, child);
 699                 }
 700                 closedir(dir);
 701         } else { /* Archiving a symbolic link */
 702                 inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
 703                 inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
 704
 705                 /* The idea here is to call readlink() to get the UNIX target of
 706                  * the symbolic link, then turn the target into a reparse point
 707                  * data buffer that contains a relative or absolute symbolic
 708                  * link (NOT a junction point or *full* path symbolic link with
 709                  * drive letter).
 710                  */
 711
 712                 char deref_name_buf[4096];
 713                 ssize_t deref_name_len;
 714
 715                 deref_name_len = readlink(root_disk_path, deref_name_buf,
 716                                           sizeof(deref_name_buf) - 1);
 717                 if (deref_name_len >= 0) {
 718                         deref_name_buf[deref_name_len] = '\0';
 719                         DEBUG("Read symlink `%s'", deref_name_buf);
 720                         ret = inode_set_symlink(root->d_inode, deref_name_buf,
 721                                                 lookup_table, NULL);
 722                         if (ret == 0) {
 723                                 /*
 724                                  * Unfortunately, Windows seems to have the
 725                                  * concept of "file" symbolic links as being
 726                                  * different from "directory" symbolic links...
 727                                  * so FILE_ATTRIBUTE_DIRECTORY needs to be set
 728                                  * on the symbolic link if the *target* of the
 729                                  * symbolic link is a directory.
 730                                  */
 731                                 struct stat stbuf;
 732                                 if (stat(root_disk_path, &stbuf) == 0 &&
 733                                     S_ISDIR(stbuf.st_mode))
 734                                 {
 735                                         inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 736                                 }
 737                         }
 738                 } else {
 739                         ERROR_WITH_ERRNO("Failed to read target of "
 740                                          "symbolic link `%s'", root_disk_path);
 741                         ret = WIMLIB_ERR_READLINK;
 742                 }
 743         }
 744 #else
 745         /* Win32 version of capturing a directory tree */
 746
 747         wchar_t *path_utf16;
 748         size_t path_utf16_nchars;
 749         struct sd_set *sd_set;
 750         DWORD err;
 751
 752         if (extra_arg == NULL) {
 753                 sd_set = alloca(sizeof(struct sd_set));
 754                 sd_set->rb_root.rb_node = NULL,
 755                 sd_set->sd = sd;
 756         } else {
 757                 sd_set = extra_arg;
 758         }
 759
 760         ret = utf8_to_utf16(root_disk_path, strlen(root_disk_path),
 761                             (char**)&path_utf16, &path_utf16_nchars);
 762         if (ret)
 763                 goto out_destroy_sd_set;
 764         path_utf16_nchars /= sizeof(wchar_t);
 765
 766         HANDLE hFile = win32_open_file(path_utf16);
 767         if (hFile == INVALID_HANDLE_VALUE) {
 768                 err = GetLastError();
 769                 ERROR("Win32 API: Failed to open \"%s\"", root_disk_path);
 770                 win32_error(err);
 771                 ret = WIMLIB_ERR_OPEN;
 772                 goto out_free_path_utf16;
 773         }
 774
 775         BY_HANDLE_FILE_INFORMATION file_info;
 776         if (!GetFileInformationByHandle(hFile, &file_info)) {
 777                 err = GetLastError();
 778                 ERROR("Win32 API: Failed to get file information for \"%s\"",
 779                       root_disk_path);
 780                 win32_error(err);
 781                 ret = WIMLIB_ERR_STAT;
 782                 goto out_close_handle;
 783         }
 784
 785         /* Create a WIM dentry */
 786         root = new_dentry_with_timeless_inode(path_basename(root_disk_path));
 787         if (!root) {
 788                 if (errno == EILSEQ)
 789                         ret = WIMLIB_ERR_INVALID_UTF8_STRING;
 790                 else if (errno == ENOMEM)
 791                         ret = WIMLIB_ERR_NOMEM;
 792                 else
 793                         ret = WIMLIB_ERR_ICONV_NOT_AVAILABLE;
 794                 goto out_free_path_utf16;
 795         }
 796
 797         /* Start preparing the associated WIM inode */
 798         inode = root->d_inode;
 799
 800         inode->i_attributes = file_info.dwFileAttributes;
 801         inode->i_creation_time = FILETIME_to_u64(&file_info.ftCreationTime);
 802         inode->i_last_write_time = FILETIME_to_u64(&file_info.ftLastWriteTime);
 803         inode->i_last_access_time = FILETIME_to_u64(&file_info.ftLastAccessTime);
 804         inode->i_ino = ((u64)file_info.nFileIndexHigh << 32) |
 805                         (u64)file_info.nFileIndexLow;
 806
 807         inode->i_resolved = 1;
 808         add_image_flags &= ~(WIMLIB_ADD_IMAGE_FLAG_ROOT | WIMLIB_ADD_IMAGE_FLAG_SOURCE);
 809
 810         /* Get DOS name and security descriptor (if any). */
 811         ret = win32_get_short_name(root, path_utf16);
 812         if (ret)
 813                 goto out_close_handle;
 814         ret = win32_get_security_descriptor(root, sd_set, path_utf16,
 815                                             root_disk_path);
 816         if (ret)
 817                 goto out_close_handle;
 818
 819         if (inode_is_directory(inode)) {
 820                 /* Directory (not a reparse point) --- recurse to children */
 821
 822                 /* But first... directories may have alternate data streams that
 823                  * need to be captured */
 824                 ret = win32_capture_streams(root_disk_path,
 825                                             path_utf16,
 826                                             path_utf16_nchars,
 827                                             inode,
 828                                             lookup_table);
 829                 if (ret)
 830                         goto out_close_handle;
 831                 ret = win32_recurse_directory(root,
 832                                               root_disk_path,
 833                                               lookup_table,
 834                                               sd,
 835                                               config,
 836                                               add_image_flags,
 837                                               progress_func,
 838                                               sd_set,
 839                                               path_utf16,
 840                                               path_utf16_nchars);
 841         } else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
 842                 /* Reparse point: save the reparse tag and data */
 843
 844                 ret = win32_capture_reparse_point(root_disk_path,
 845                                                   hFile,
 846                                                   inode,
 847                                                   lookup_table);
 848
 849         } else {
 850                 /* Not a directory, not a reparse point */
 851                 ret = win32_capture_streams(root_disk_path,
 852                                             path_utf16,
 853                                             path_utf16_nchars,
 854                                             inode,
 855                                             lookup_table);
 856         }
 857 out_close_handle:
 858         CloseHandle(hFile);
 859 out_destroy_sd_set:
 860         if (extra_arg == NULL)
 861                 destroy_sd_set(sd_set);
 862 out_free_path_utf16:
 863         FREE(path_utf16);
 864 #endif
 865 out:
 866         if (ret == 0)
 867                 *root_ret = root;
 868         else
 869                 free_dentry_tree(root, lookup_table);
 870         return ret;
 871 }
 872
 873 enum pattern_type {
 874         NONE = 0,
 875         EXCLUSION_LIST,
 876         EXCLUSION_EXCEPTION,
 877         COMPRESSION_EXCLUSION_LIST,
 878         ALIGNMENT_LIST,
 879 };
 880
 881 #define COMPAT_DEFAULT_CONFIG
 882
 883 /* Default capture configuration file when none is specified. */
 884 static const char *default_config =
 885 #ifdef COMPAT_DEFAULT_CONFIG /* XXX: This policy is being moved to library
 886                                 users.  The next ABI-incompatible library
 887                                 version will default to the empty string here. */
 888 "[ExclusionList]\n"
 889 "\\$ntfs.log\n"
 890 "\\hiberfil.sys\n"
 891 "\\pagefile.sys\n"
 892 "\\System Volume Information\n"
 893 "\\RECYCLER\n"
 894 "\\Windows\\CSC\n"
 895 "\n"
 896 "[CompressionExclusionList]\n"
 897 "*.mp3\n"
 898 "*.zip\n"
 899 "*.cab\n"
 900 "\\WINDOWS\\inf\\*.pnf\n";
 901 #else
 902 "";
 903 #endif
 904
 905 static void destroy_pattern_list(struct pattern_list *list)
 906 {
 907         FREE(list->pats);
 908 }
 909
 910 static void destroy_capture_config(struct capture_config *config)
 911 {
 912         destroy_pattern_list(&config->exclusion_list);
 913         destroy_pattern_list(&config->exclusion_exception);
 914         destroy_pattern_list(&config->compression_exclusion_list);
 915         destroy_pattern_list(&config->alignment_list);
 916         FREE(config->config_str);
 917         FREE(config->prefix);
 918         memset(config, 0, sizeof(*config));
 919 }
 920
 921 static int pattern_list_add_pattern(struct pattern_list *list,
 922                                     const char *pattern)
 923 {
 924         const char **pats;
 925         if (list->num_pats >= list->num_allocated_pats) {
 926                 pats = REALLOC(list->pats,
 927                                sizeof(list->pats[0]) * (list->num_allocated_pats + 8));
 928                 if (!pats)
 929                         return WIMLIB_ERR_NOMEM;
 930                 list->num_allocated_pats += 8;
 931                 list->pats = pats;
 932         }
 933         list->pats[list->num_pats++] = pattern;
 934         return 0;
 935 }
 936
 937 /* Parses the contents of the image capture configuration file and fills in a
 938  * `struct capture_config'. */
 939 static int init_capture_config(struct capture_config *config,
 940                                const char *_config_str, size_t config_len)
 941 {
 942         char *config_str;
 943         char *p;
 944         char *eol;
 945         char *next_p;
 946         size_t bytes_remaining;
 947         enum pattern_type type = NONE;
 948         int ret;
 949         unsigned long line_no = 0;
 950
 951         DEBUG("config_len = %zu", config_len);
 952         bytes_remaining = config_len;
 953         memset(config, 0, sizeof(*config));
 954         config_str = MALLOC(config_len);
 955         if (!config_str) {
 956                 ERROR("Could not duplicate capture config string");
 957                 return WIMLIB_ERR_NOMEM;
 958         }
 959
 960         memcpy(config_str, _config_str, config_len);
 961         next_p = config_str;
 962         config->config_str = config_str;
 963         while (bytes_remaining) {
 964                 line_no++;
 965                 p = next_p;
 966                 eol = memchr(p, '\n', bytes_remaining);
 967                 if (!eol) {
 968                         ERROR("Expected end-of-line in capture config file on "
 969                               "line %lu", line_no);
 970                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
 971                         goto out_destroy;
 972                 }
 973
 974                 next_p = eol + 1;
 975                 bytes_remaining -= (next_p - p);
 976                 if (eol == p)
 977                         continue;
 978
 979                 if (*(eol - 1) == '\r')
 980                         eol--;
 981                 *eol = '\0';
 982
 983                 /* Translate backslash to forward slash */
 984                 for (char *pp = p; pp != eol; pp++)
 985                         if (*pp == '\\')
 986                                 *pp = '/';
 987
 988                 /* Remove drive letter */
 989                 if (eol - p > 2 && isalpha(*p) && *(p + 1) == ':')
 990                         p += 2;
 991
 992                 ret = 0;
 993                 if (strcmp(p, "[ExclusionList]") == 0)
 994                         type = EXCLUSION_LIST;
 995                 else if (strcmp(p, "[ExclusionException]") == 0)
 996                         type = EXCLUSION_EXCEPTION;
 997                 else if (strcmp(p, "[CompressionExclusionList]") == 0)
 998                         type = COMPRESSION_EXCLUSION_LIST;
 999                 else if (strcmp(p, "[AlignmentList]") == 0)
1000                         type = ALIGNMENT_LIST;
1001                 else if (p[0] == '[' && strrchr(p, ']')) {
1002                         ERROR("Unknown capture configuration section `%s'", p);
1003                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
1004                 } else switch (type) {
1005                 case EXCLUSION_LIST:
1006                         DEBUG("Adding pattern \"%s\" to exclusion list", p);
1007                         ret = pattern_list_add_pattern(&config->exclusion_list, p);
1008                         break;
1009                 case EXCLUSION_EXCEPTION:
1010                         DEBUG("Adding pattern \"%s\" to exclusion exception list", p);
1011                         ret = pattern_list_add_pattern(&config->exclusion_exception, p);
1012                         break;
1013                 case COMPRESSION_EXCLUSION_LIST:
1014                         DEBUG("Adding pattern \"%s\" to compression exclusion list", p);
1015                         ret = pattern_list_add_pattern(&config->compression_exclusion_list, p);
1016                         break;
1017                 case ALIGNMENT_LIST:
1018                         DEBUG("Adding pattern \"%s\" to alignment list", p);
1019                         ret = pattern_list_add_pattern(&config->alignment_list, p);
1020                         break;
1021                 default:
1022                         ERROR("Line %lu of capture configuration is not "
1023                               "in a block (such as [ExclusionList])",
1024                               line_no);
1025                         ret = WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
1026                         break;
1027                 }
1028                 if (ret != 0)
1029                         goto out_destroy;
1030         }
1031         return 0;
1032 out_destroy:
1033         destroy_capture_config(config);
1034         return ret;
1035 }
1036
1037 static int capture_config_set_prefix(struct capture_config *config,
1038                                      const char *_prefix)
1039 {
1040         char *prefix = STRDUP(_prefix);
1041
1042         if (!prefix)
1043                 return WIMLIB_ERR_NOMEM;
1044         FREE(config->prefix);
1045         config->prefix = prefix;
1046         config->prefix_len = strlen(prefix);
1047         return 0;
1048 }
1049
1050 static bool match_pattern(const char *path, const char *path_basename,
1051                           const struct pattern_list *list)
1052 {
1053         for (size_t i = 0; i < list->num_pats; i++) {
1054                 const char *pat = list->pats[i];
1055                 const char *string;
1056                 if (pat[0] == '/')
1057                         /* Absolute path from root of capture */
1058                         string = path;
1059                 else {
1060                         if (strchr(pat, '/'))
1061                                 /* Relative path from root of capture */
1062                                 string = path + 1;
1063                         else
1064                                 /* A file name pattern */
1065                                 string = path_basename;
1066                 }
1067                 if (fnmatch(pat, string, FNM_PATHNAME
1068                         #ifdef FNM_CASEFOLD
1069                                         | FNM_CASEFOLD
1070                         #endif
1071                         ) == 0)
1072                 {
1073                         DEBUG("`%s' matches the pattern \"%s\"",
1074                               string, pat);
1075                         return true;
1076                 }
1077         }
1078         return false;
1079 }
1080
1081 /* Return true if the image capture configuration file indicates we should
1082  * exclude the filename @path from capture.
1083  *
1084  * If @exclude_prefix is %true, the part of the path up and including the name
1085  * of the directory being captured is not included in the path for matching
1086  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
1087  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
1088  * directory.
1089  */
1090 bool exclude_path(const char *path, const struct capture_config *config,
1091                   bool exclude_prefix)
1092 {
1093         const char *basename = path_basename(path);
1094         if (exclude_prefix) {
1095                 wimlib_assert(strlen(path) >= config->prefix_len);
1096                 if (memcmp(config->prefix, path, config->prefix_len) == 0
1097                      && path[config->prefix_len] == '/')
1098                         path += config->prefix_len;
1099         }
1100         return match_pattern(path, basename, &config->exclusion_list) &&
1101                 !match_pattern(path, basename, &config->exclusion_exception);
1102
1103 }
1104
1105 /* Strip leading and trailing forward slashes from a string.  Modifies it in
1106  * place and returns the stripped string. */
1107 static const char *canonicalize_target_path(char *target_path)
1108 {
1109         char *p;
1110         if (target_path == NULL)
1111                 target_path = "";
1112         for (;;) {
1113                 if (*target_path == '\0')
1114                         return target_path;
1115                 else if (*target_path == '/')
1116                         target_path++;
1117                 else
1118                         break;
1119         }
1120
1121         p = target_path + strlen(target_path) - 1;
1122         while (*p == '/')
1123                 *p-- = '\0';
1124         return target_path;
1125 }
1126
1127 /* Strip leading and trailing slashes from the target paths */
1128 static void canonicalize_targets(struct wimlib_capture_source *sources,
1129                                  size_t num_sources)
1130 {
1131         while (num_sources--) {
1132                 DEBUG("Canonicalizing { source: \"%s\", target=\"%s\"}",
1133                       sources->fs_source_path,
1134                       sources->wim_target_path);
1135                 sources->wim_target_path =
1136                         (char*)canonicalize_target_path(sources->wim_target_path);
1137                 DEBUG("Canonical target: \"%s\"", sources->wim_target_path);
1138                 sources++;
1139         }
1140 }
1141
1142 static int capture_source_cmp(const void *p1, const void *p2)
1143 {
1144         const struct wimlib_capture_source *s1 = p1, *s2 = p2;
1145         return strcmp(s1->wim_target_path, s2->wim_target_path);
1146 }
1147
1148 /* Sorts the capture sources lexicographically by target path.  This occurs
1149  * after leading and trailing forward slashes are stripped.
1150  *
1151  * One purpose of this is to make sure that target paths that are inside other
1152  * target paths are extracted after the containing target paths. */
1153 static void sort_sources(struct wimlib_capture_source *sources,
1154                          size_t num_sources)
1155 {
1156         qsort(sources, num_sources, sizeof(sources[0]), capture_source_cmp);
1157 }
1158
1159 static int check_sorted_sources(struct wimlib_capture_source *sources,
1160                                 size_t num_sources, int add_image_flags)
1161 {
1162         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
1163                 if (num_sources != 1) {
1164                         ERROR("Must specify exactly 1 capture source "
1165                               "(the NTFS volume) in NTFS mode!");
1166                         return WIMLIB_ERR_INVALID_PARAM;
1167                 }
1168                 if (sources[0].wim_target_path[0] != '\0') {
1169                         ERROR("In NTFS capture mode the target path inside "
1170                               "the image must be the root directory!");
1171                         return WIMLIB_ERR_INVALID_PARAM;
1172                 }
1173         } else if (num_sources != 0) {
1174                 /* This code is disabled because the current code
1175                  * unconditionally attempts to do overlays.  So, duplicate
1176                  * target paths are OK. */
1177         #if 0
1178                 if (num_sources > 1 && sources[0].wim_target_path[0] == '\0') {
1179                         ERROR("Cannot specify root target when using multiple "
1180                               "capture sources!");
1181                         return WIMLIB_ERR_INVALID_PARAM;
1182                 }
1183                 for (size_t i = 0; i < num_sources - 1; i++) {
1184                         size_t len = strlen(sources[i].wim_target_path);
1185                         size_t j = i + 1;
1186                         const char *target1 = sources[i].wim_target_path;
1187                         do {
1188                                 const char *target2 = sources[j].wim_target_path;
1189                                 DEBUG("target1=%s, target2=%s",
1190                                       target1,target2);
1191                                 if (strncmp(target1, target2, len) ||
1192                                     target2[len] > '/')
1193                                         break;
1194                                 if (target2[len] == '/') {
1195                                         ERROR("Invalid target `%s': is a prefix of `%s'",
1196                                               target1, target2);
1197                                         return WIMLIB_ERR_INVALID_PARAM;
1198                                 }
1199                                 if (target2[len] == '\0') {
1200                                         ERROR("Invalid target `%s': is a duplicate of `%s'",
1201                                               target1, target2);
1202                                         return WIMLIB_ERR_INVALID_PARAM;
1203                                 }
1204                         } while (++j != num_sources);
1205                 }
1206         #endif
1207         }
1208         return 0;
1209
1210 }
1211
1212 /* Creates a new directory to place in the WIM image.  This is to create parent
1213  * directories that are not part of any target as needed.  */
1214 static struct wim_dentry *
1215 new_filler_directory(const char *name)
1216 {
1217         struct wim_dentry *dentry;
1218         DEBUG("Creating filler directory \"%s\"", name);
1219         dentry = new_dentry_with_inode(name);
1220         if (dentry) {
1221                 /* Set the inode number to 0 for now.  The final inode number
1222                  * will be assigned later by assign_inode_numbers(). */
1223                 dentry->d_inode->i_ino = 0;
1224                 dentry->d_inode->i_resolved = 1;
1225                 dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
1226         }
1227         return dentry;
1228 }
1229
1230 /* Transfers the children of @branch to @target.  It is an error if @target is
1231  * not a directory or if both @branch and @target contain a child dentry with
1232  * the same name. */
1233 static int do_overlay(struct wim_dentry *target, struct wim_dentry *branch)
1234 {
1235         struct rb_root *rb_root;
1236
1237         if (!dentry_is_directory(target)) {
1238                 ERROR("Cannot overlay directory `%s' over non-directory",
1239                       branch->file_name_utf8);
1240                 return WIMLIB_ERR_INVALID_OVERLAY;
1241         }
1242
1243         rb_root = &branch->d_inode->i_children;
1244         while (rb_root->rb_node) { /* While @branch has children... */
1245                 struct wim_dentry *child = rbnode_dentry(rb_root->rb_node);
1246                 /* Move @child to the directory @target */
1247                 unlink_dentry(child);
1248                 if (!dentry_add_child(target, child)) {
1249                         /* Revert the change to avoid leaking the directory tree
1250                          * rooted at @child */
1251                         dentry_add_child(branch, child);
1252                         ERROR("Overlay error: file `%s' already exists "
1253                               "as a child of `%s'",
1254                               child->file_name_utf8, target->file_name_utf8);
1255                         return WIMLIB_ERR_INVALID_OVERLAY;
1256                 }
1257         }
1258         return 0;
1259
1260 }
1261
1262 /* Attach or overlay a branch onto the WIM image.
1263  *
1264  * @root_p:
1265  *      Pointer to the root of the WIM image, or pointer to NULL if it has not
1266  *      been created yet.
1267  * @branch
1268  *      Branch to add.
1269  * @target_path:
1270  *      Path in the WIM image to add the branch, with leading and trailing
1271  *      slashes stripped.
1272  */
1273 static int attach_branch(struct wim_dentry **root_p,
1274                          struct wim_dentry *branch,
1275                          char *target_path)
1276 {
1277         char *slash;
1278         struct wim_dentry *dentry, *parent, *target;
1279
1280         if (*target_path == '\0') {
1281                 /* Target: root directory */
1282                 if (*root_p) {
1283                         /* Overlay on existing root */
1284                         return do_overlay(*root_p, branch);
1285                 } else  {
1286                         /* Set as root */
1287                         *root_p = branch;
1288                         return 0;
1289                 }
1290         }
1291
1292         /* Adding a non-root branch.  Create root if it hasn't been created
1293          * already. */
1294         if (!*root_p) {
1295                 *root_p = new_filler_directory("");
1296                 if (!*root_p)
1297                         return WIMLIB_ERR_NOMEM;
1298         }
1299
1300         /* Walk the path to the branch, creating filler directories as needed.
1301          * */
1302         parent = *root_p;
1303         while ((slash = strchr(target_path, '/'))) {
1304                 *slash = '\0';
1305                 dentry = get_dentry_child_with_name(parent, target_path);
1306                 if (!dentry) {
1307                         dentry = new_filler_directory(target_path);
1308                         if (!dentry)
1309                                 return WIMLIB_ERR_NOMEM;
1310                         dentry_add_child(parent, dentry);
1311                 }
1312                 parent = dentry;
1313                 target_path = slash;
1314                 /* Skip over slashes.  Note: this cannot overrun the length of
1315                  * the string because the last character cannot be a slash, as
1316                  * trailing slashes were tripped.  */
1317                 do {
1318                         ++target_path;
1319                 } while (*target_path == '/');
1320         }
1321
1322         /* If the target path already existed, overlay the branch onto it.
1323          * Otherwise, set the branch as the target path. */
1324         target = get_dentry_child_with_name(parent, branch->file_name_utf8);
1325         if (target) {
1326                 return do_overlay(target, branch);
1327         } else {
1328                 dentry_add_child(parent, branch);
1329                 return 0;
1330         }
1331 }
1332
1333 WIMLIBAPI int wimlib_add_image_multisource(WIMStruct *w,
1334                                            struct wimlib_capture_source *sources,
1335                                            size_t num_sources,
1336                                            const char *name,
1337                                            const char *config_str,
1338                                            size_t config_len,
1339                                            int add_image_flags,
1340                                            wimlib_progress_func_t progress_func)
1341 {
1342         int (*capture_tree)(struct wim_dentry **, const char *,
1343                             struct wim_lookup_table *,
1344                             struct wim_security_data *,
1345                             const struct capture_config *,
1346                             int, wimlib_progress_func_t, void *);
1347         void *extra_arg;
1348         struct wim_dentry *root_dentry;
1349         struct wim_dentry *branch;
1350         struct wim_security_data *sd;
1351         struct capture_config config;
1352         struct wim_image_metadata *imd;
1353         int ret;
1354
1355         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_NTFS) {
1356 #ifdef WITH_NTFS_3G
1357                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_DEREFERENCE) {
1358                         ERROR("Cannot dereference files when capturing directly from NTFS");
1359                         return WIMLIB_ERR_INVALID_PARAM;
1360                 }
1361                 if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_UNIX_DATA) {
1362                         ERROR("Capturing UNIX owner and mode not supported "
1363                               "when capturing directly from NTFS");
1364                         return WIMLIB_ERR_INVALID_PARAM;
1365                 }
1366                 capture_tree = build_dentry_tree_ntfs;
1367                 extra_arg = &w->ntfs_vol;
1368 #else
1369                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1370                       "        cannot capture a WIM image directly from a NTFS volume!");
1371                 return WIMLIB_ERR_UNSUPPORTED;
1372 #endif
1373         } else {
1374                 capture_tree = build_dentry_tree;
1375                 extra_arg = NULL;
1376         }
1377
1378         if (!name || !*name) {
1379                 ERROR("Must specify a non-empty string for the image name");
1380                 return WIMLIB_ERR_INVALID_PARAM;
1381         }
1382
1383         if (w->hdr.total_parts != 1) {
1384                 ERROR("Cannot add an image to a split WIM");
1385                 return WIMLIB_ERR_SPLIT_UNSUPPORTED;
1386         }
1387
1388         if (wimlib_image_name_in_use(w, name)) {
1389                 ERROR("There is already an image named \"%s\" in `%s'",
1390                       name, w->filename);
1391                 return WIMLIB_ERR_IMAGE_NAME_COLLISION;
1392         }
1393
1394         if (!config_str) {
1395                 DEBUG("Using default capture configuration");
1396                 config_str = default_config;
1397                 config_len = strlen(default_config);
1398         }
1399         ret = init_capture_config(&config, config_str, config_len);
1400         if (ret)
1401                 goto out;
1402
1403         DEBUG("Allocating security data");
1404         sd = CALLOC(1, sizeof(struct wim_security_data));
1405         if (!sd) {
1406                 ret = WIMLIB_ERR_NOMEM;
1407                 goto out_destroy_capture_config;
1408         }
1409         sd->total_length = 8;
1410         sd->refcnt = 1;
1411
1412         DEBUG("Using %zu capture sources", num_sources);
1413         canonicalize_targets(sources, num_sources);
1414         sort_sources(sources, num_sources);
1415         ret = check_sorted_sources(sources, num_sources, add_image_flags);
1416         if (ret) {
1417                 ret = WIMLIB_ERR_INVALID_PARAM;
1418                 goto out_free_security_data;
1419         }
1420
1421         DEBUG("Building dentry tree.");
1422         if (num_sources == 0) {
1423                 root_dentry = new_filler_directory("");
1424                 if (!root_dentry)
1425                         goto out_free_security_data;
1426         } else {
1427                 size_t i;
1428
1429                 root_dentry = NULL;
1430                 i = 0;
1431                 do {
1432                         int flags;
1433                         union wimlib_progress_info progress;
1434
1435                         DEBUG("Building dentry tree for source %zu of %zu "
1436                               "(\"%s\" => \"%s\")", i + 1, num_sources,
1437                               sources[i].fs_source_path,
1438                               sources[i].wim_target_path);
1439                         if (progress_func) {
1440                                 memset(&progress, 0, sizeof(progress));
1441                                 progress.scan.source = sources[i].fs_source_path;
1442                                 progress.scan.wim_target_path = sources[i].wim_target_path;
1443                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_BEGIN, &progress);
1444                         }
1445                         ret = capture_config_set_prefix(&config,
1446                                                         sources[i].fs_source_path);
1447                         if (ret)
1448                                 goto out_free_dentry_tree;
1449                         flags = add_image_flags | WIMLIB_ADD_IMAGE_FLAG_SOURCE;
1450                         if (!*sources[i].wim_target_path)
1451                                 flags |= WIMLIB_ADD_IMAGE_FLAG_ROOT;
1452                         ret = (*capture_tree)(&branch, sources[i].fs_source_path,
1453                                               w->lookup_table, sd,
1454                                               &config,
1455                                               flags,
1456                                               progress_func, extra_arg);
1457                         if (ret) {
1458                                 ERROR("Failed to build dentry tree for `%s'",
1459                                       sources[i].fs_source_path);
1460                                 goto out_free_dentry_tree;
1461                         }
1462                         if (branch) {
1463                                 /* Use the target name, not the source name, for
1464                                  * the root of each branch from a capture
1465                                  * source.  (This will also set the root dentry
1466                                  * of the entire image to be unnamed.) */
1467                                 ret = set_dentry_name(branch,
1468                                                       path_basename(sources[i].wim_target_path));
1469                                 if (ret)
1470                                         goto out_free_branch;
1471
1472                                 ret = attach_branch(&root_dentry, branch,
1473                                                     sources[i].wim_target_path);
1474                                 if (ret)
1475                                         goto out_free_branch;
1476                         }
1477                         if (progress_func)
1478                                 progress_func(WIMLIB_PROGRESS_MSG_SCAN_END, &progress);
1479                 } while (++i != num_sources);
1480         }
1481
1482         DEBUG("Calculating full paths of dentries.");
1483         ret = for_dentry_in_tree(root_dentry, calculate_dentry_full_path, NULL);
1484         if (ret != 0)
1485                 goto out_free_dentry_tree;
1486
1487         ret = add_new_dentry_tree(w, root_dentry, sd);
1488         if (ret != 0)
1489                 goto out_free_dentry_tree;
1490
1491         imd = &w->image_metadata[w->hdr.image_count - 1];
1492
1493         ret = dentry_tree_fix_inodes(root_dentry, &imd->inode_list);
1494         if (ret != 0)
1495                 goto out_destroy_imd;
1496
1497         DEBUG("Assigning hard link group IDs");
1498         assign_inode_numbers(&imd->inode_list);
1499
1500         ret = xml_add_image(w, name);
1501         if (ret != 0)
1502                 goto out_destroy_imd;
1503
1504         if (add_image_flags & WIMLIB_ADD_IMAGE_FLAG_BOOT)
1505                 wimlib_set_boot_idx(w, w->hdr.image_count);
1506         ret = 0;
1507         goto out;
1508 out_destroy_imd:
1509         destroy_image_metadata(&w->image_metadata[w->hdr.image_count - 1],
1510                                w->lookup_table);
1511         w->hdr.image_count--;
1512         goto out;
1513 out_free_branch:
1514         free_dentry_tree(branch, w->lookup_table);
1515 out_free_dentry_tree:
1516         free_dentry_tree(root_dentry, w->lookup_table);
1517 out_free_security_data:
1518         free_security_data(sd);
1519 out_destroy_capture_config:
1520         destroy_capture_config(&config);
1521 out:
1522         return ret;
1523 }
1524
1525 WIMLIBAPI int wimlib_add_image(WIMStruct *w, const char *source,
1526                                const char *name, const char *config_str,
1527                                size_t config_len, int add_image_flags,
1528                                wimlib_progress_func_t progress_func)
1529 {
1530         if (!source || !*source)
1531                 return WIMLIB_ERR_INVALID_PARAM;
1532
1533         char *fs_source_path = STRDUP(source);
1534         int ret;
1535         struct wimlib_capture_source capture_src = {
1536                 .fs_source_path = fs_source_path,
1537                 .wim_target_path = NULL,
1538                 .reserved = 0,
1539         };
1540         ret = wimlib_add_image_multisource(w, &capture_src, 1, name,
1541                                            config_str, config_len,
1542                                            add_image_flags, progress_func);
1543         FREE(fs_source_path);
1544         return ret;
1545 }