wimlib.net Git - wimlib/blob - src/extract.c

   1 /*
   2  * extract.c
   3  *
   4  * Support for extracting WIM files.
   5  */
   6
   7 /*
   8  * Copyright (C) 2010 Carl Thijssen
   9  * Copyright (C) 2012 Eric Biggers
  10  *
  11  * This file is part of wimlib, a library for working with WIM files.
  12  *
  13  * wimlib is free software; you can redistribute it and/or modify it under the
  14  * terms of the GNU Lesser General Public License as published by the Free
  15  * Software Foundation; either version 2.1 of the License, or (at your option)
  16  * any later version.
  17  *
  18  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  19  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  20  * A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  21  * details.
  22  *
  23  * You should have received a copy of the GNU Lesser General Public License
  24  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  25  */
  26
  27 #include "wimlib_internal.h"
  28 #include "dentry.h"
  29 #include "lookup_table.h"
  30 #include "xml.h"
  31 #include <sys/stat.h>
  32 #include <fcntl.h>
  33 #include <unistd.h>
  34 #include <dirent.h>
  35 #include <string.h>
  36 #include <errno.h>
  37
  38 #ifdef WITH_NTFS_3G
  39 #include <ntfs-3g/volume.h>
  40 #include <ntfs-3g/security.h>
  41 #endif
  42
  43 /* Sets and creates the directory to which files are to be extracted when
  44  * extracting files from the WIM. */
  45 static int set_output_dir(WIMStruct *w, const char *dir)
  46 {
  47         char *p;
  48         DEBUG("Setting output directory to `%s'", dir);
  49
  50         p = STRDUP(dir);
  51         if (!p) {
  52                 ERROR("Out of memory");
  53                 return WIMLIB_ERR_NOMEM;
  54         }
  55
  56         if (mkdir(dir, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH) != 0) {
  57                 if (errno == EEXIST) {
  58                         DEBUG("`%s' already exists", dir);
  59                         goto done;
  60                 }
  61                 ERROR_WITH_ERRNO("Cannot create directory `%s'", dir);
  62                 FREE(p);
  63                 return WIMLIB_ERR_MKDIR;
  64         } else {
  65                 DEBUG("Created directory `%s'", dir);
  66         }
  67 done:
  68         FREE(w->output_dir);
  69         w->output_dir = p;
  70         return 0;
  71 }
  72
  73 /*
  74  * Extracts a regular file from the WIM archive.
  75  *
  76  * @dentry:             The directory entry for the file, which must be a
  77  *                              regular file.
  78  * @output_path:        The path to which the file is to be extracted.
  79  * @lookup_table:       The lookup table for the WIM file.
  80  * @wim_fp:             The FILE* for the WIM, opened for reading.
  81  * @wim_ctype:          The type of compression used in the WIM.
  82  * @link_type:          One of WIM_LINK_TYPE_*; specifies what to do with
  83  *                      files that are hard-linked inside the WIM.
  84  * @is_multi_image_extraction:
  85  *                      True if the image currently being extracted is just one
  86  *                      image of a multi-image extraction.  This is needed so
  87  *                      that cross-image symbolic links can be created
  88  *                      correctly.
  89  */
  90 static int extract_regular_file(WIMStruct *w,
  91                                 const struct dentry *dentry,
  92                                 const char *output_path,
  93                                 int extract_flags)
  94 {
  95         struct lookup_table_entry *lte;
  96         int ret;
  97         int out_fd;
  98         const struct resource_entry *res_entry;
  99
 100         lte = lookup_resource(w->lookup_table, dentry_hash(dentry));
 101
 102         /* If we already extracted the same file or a hard link copy of it, we
 103          * may be able to simply create a link.  The exact action is specified
 104          * by the current @link_type. */
 105         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK)) &&
 106               lte && lte->out_refcnt != 0)
 107         {
 108                 wimlib_assert(lte->file_on_disk);
 109
 110                 if (extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
 111                         if (link(lte->file_on_disk, output_path) != 0) {
 112                                 ERROR_WITH_ERRNO("Failed to hard link "
 113                                                  "`%s' to `%s'",
 114                                                  output_path, lte->file_on_disk);
 115                                 return WIMLIB_ERR_LINK;
 116                         }
 117                 } else {
 118                         int num_path_components;
 119                         int num_output_dir_path_components;
 120                         size_t file_on_disk_len;
 121                         char *p;
 122                         const char *p2;
 123                         size_t i;
 124
 125                         num_path_components =
 126                                 get_num_path_components(dentry->full_path_utf8) - 1;
 127                         num_output_dir_path_components =
 128                                 get_num_path_components(w->output_dir);
 129
 130                         if (w->is_multi_image_extraction) {
 131                                 num_path_components++;
 132                                 num_output_dir_path_components--;
 133                         }
 134                         file_on_disk_len = strlen(lte->file_on_disk);
 135
 136                         char buf[file_on_disk_len + 3 * num_path_components + 1];
 137                         p = &buf[0];
 138
 139                         for (i = 0; i < num_path_components; i++) {
 140                                 *p++ = '.';
 141                                 *p++ = '.';
 142                                 *p++ = '/';
 143                         }
 144                         p2 = lte->file_on_disk;
 145                         while (*p2 == '/')
 146                                 p2++;
 147                         while (num_output_dir_path_components--)
 148                                 p2 = path_next_part(p2, NULL);
 149                         strcpy(p, p2);
 150                         if (symlink(buf, output_path) != 0) {
 151                                 ERROR_WITH_ERRNO("Failed to symlink `%s' to "
 152                                                  "`%s'",
 153                                                  buf, lte->file_on_disk);
 154                                 return WIMLIB_ERR_LINK;
 155                         }
 156
 157                 }
 158                 return 0;
 159         }
 160
 161         /* Otherwise, we must actually extract the file contents. */
 162
 163         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
 164         if (out_fd == -1) {
 165                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
 166                                  output_path);
 167                 return WIMLIB_ERR_OPEN;
 168         }
 169
 170         /* Extract empty file, with no lookup table entry... */
 171         if (!lte) {
 172                 DEBUG("Empty file `%s'.", output_path);
 173                 ret = 0;
 174                 goto done;
 175         }
 176
 177         res_entry = &lte->resource_entry;
 178
 179         ret = extract_resource_to_fd(w, res_entry, out_fd,
 180                                      res_entry->original_size);
 181
 182         if (ret != 0) {
 183                 ERROR("Failed to extract resource to `%s'", output_path);
 184                 goto done;
 185         }
 186
 187         /* Mark the lookup table entry to indicate this file has been extracted. */
 188         lte->out_refcnt++;
 189         FREE(lte->file_on_disk);
 190         lte->file_on_disk = STRDUP(output_path);
 191         if (lte->file_on_disk)
 192                 ret = 0;
 193         else
 194                 ret = WIMLIB_ERR_NOMEM;
 195 done:
 196         close(out_fd);
 197         return ret;
 198 }
 199
 200 static int extract_symlink(const struct dentry *dentry, const char *output_path,
 201                            const WIMStruct *w)
 202 {
 203         char target[4096];
 204         ssize_t ret = dentry_readlink(dentry, target, sizeof(target), w);
 205         if (ret <= 0) {
 206                 ERROR("Could not read the symbolic link from dentry `%s'",
 207                       dentry->full_path_utf8);
 208                 return WIMLIB_ERR_INVALID_DENTRY;
 209         }
 210         ret = symlink(target, output_path);
 211         if (ret != 0) {
 212                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 213                                  output_path, target);
 214                 return WIMLIB_ERR_LINK;
 215         }
 216         return 0;
 217 }
 218
 219 /*
 220  * Extracts a directory from the WIM archive.
 221  *
 222  * @dentry:             The directory entry for the directory.
 223  * @output_path:        The path to which the directory is to be extracted to.
 224  * @return:             True on success, false on failure.
 225  */
 226 static int extract_directory(struct dentry *dentry, const char *output_path)
 227 {
 228         /* Compute the output path directory to the directory. */
 229         if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH) != 0)
 230         {
 231                 switch (errno) {
 232                 case EEXIST: /* Already existing directory is OK */
 233                 case EACCES: /* We may have permissions to extract files inside
 234                                  the directory, but not for the directory
 235                                  itself. */
 236                         return 0;
 237                 default:
 238                         ERROR_WITH_ERRNO("Cannot create directory `%s'",
 239                                          output_path);
 240                         return WIMLIB_ERR_MKDIR;
 241                 }
 242         }
 243         return 0;
 244 }
 245
 246 struct extract_args {
 247         WIMStruct *w;
 248         int extract_flags;
 249 #ifdef WITH_NTFS_3G
 250         struct SECURITY_API *scapi;
 251 #endif
 252 };
 253
 254 /*
 255  * Extracts a file or directory from the WIM archive.  For use in
 256  * for_dentry_in_tree().
 257  *
 258  * @dentry:     The dentry to extract.
 259  * @arg:        A pointer to the WIMStruct for the WIM file.
 260  */
 261 static int extract_dentry(struct dentry *dentry, void *arg)
 262 {
 263         struct extract_args *args = arg;
 264         WIMStruct *w = args->w;
 265         int extract_flags = args->extract_flags;
 266         size_t len = strlen(w->output_dir);
 267         char output_path[len + dentry->full_path_utf8_len + 1];
 268         int ret = 0;
 269
 270         if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE)
 271                 puts(dentry->full_path_utf8);
 272
 273         memcpy(output_path, w->output_dir, len);
 274         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 275         output_path[len + dentry->full_path_utf8_len] = '\0';
 276
 277         if (dentry_is_symlink(dentry)) {
 278                 ret = extract_symlink(dentry, output_path, w);
 279         } else if (dentry_is_directory(dentry)) {
 280                 if (!dentry_is_root(dentry)) /* Root doesn't need to be extracted. */
 281                         ret = extract_directory(dentry, output_path);
 282         } else {
 283                 ret = extract_regular_file(w, dentry, output_path, extract_flags);
 284         }
 285 }
 286
 287
 288 static int extract_single_image(WIMStruct *w, int image, int extract_flags)
 289 {
 290         DEBUG("Extracting image %d", image);
 291
 292         int ret;
 293         ret = wimlib_select_image(w, image);
 294         if (ret != 0)
 295                 return ret;
 296
 297         struct extract_args args = {
 298                 .w = w,
 299                 .extract_flags = extract_flags,
 300         #ifdef WITH_NTFS_3G
 301                 .scapi = NULL
 302         #endif
 303         };
 304
 305         return for_dentry_in_tree(wim_root_dentry(w), extract_dentry, &args);
 306 }
 307
 308
 309 /* Extracts all images from the WIM to w->output_dir, with the images placed in
 310  * subdirectories named by their image names. */
 311 static int extract_all_images(WIMStruct *w, int extract_flags)
 312 {
 313         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
 314         size_t output_path_len = strlen(w->output_dir);
 315         char buf[output_path_len + 1 + image_name_max_len + 1];
 316         int ret;
 317         int image;
 318         const char *image_name;
 319
 320         DEBUG("Attempting to extract all images from `%s'", w->filename);
 321
 322         memcpy(buf, w->output_dir, output_path_len);
 323         buf[output_path_len] = '/';
 324         for (image = 1; image <= w->hdr.image_count; image++) {
 325
 326                 image_name = wimlib_get_image_name(w, image);
 327                 if (*image_name) {
 328                         strcpy(buf + output_path_len + 1, image_name);
 329                 } else {
 330                         /* Image name is empty. Use image number instead */
 331                         sprintf(buf + output_path_len + 1, "%d", image);
 332                 }
 333                 ret = set_output_dir(w, buf);
 334                 if (ret != 0)
 335                         goto done;
 336                 ret = extract_single_image(w, image, extract_flags);
 337                 if (ret != 0)
 338                         goto done;
 339         }
 340 done:
 341         /* Restore original output directory */
 342         buf[output_path_len + 1] = '\0';
 343         return 0;
 344 }
 345
 346 /* Extracts a single image or all images from a WIM file. */
 347 WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
 348                                    const char *output_dir, int flags)
 349 {
 350         int ret;
 351         if (!output_dir)
 352                 return WIMLIB_ERR_INVALID_PARAM;
 353         if ((flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 354                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 355                 return WIMLIB_ERR_INVALID_PARAM;
 356
 357         ret = set_output_dir(w, output_dir);
 358         if (ret != 0)
 359                 return ret;
 360
 361         if ((flags & WIMLIB_EXTRACT_FLAG_NTFS)) {
 362         #ifdef WITH_NTFS_3G
 363                 unsigned long mnt_flags;
 364                 ret = ntfs_check_if_mounted(output_dir, &mnt_flags);
 365                 if (ret != 0) {
 366                         ERROR_WITH_ERRNO("NTFS-3g: Cannot determine if `%s' "
 367                                          "is mounted", output_dir);
 368                         return WIMLIB_ERR_NTFS_3G;
 369                 }
 370                 if (!(mnt_flags & NTFS_MF_MOUNTED)) {
 371                         ERROR("NTFS-3g: Filesystem on `%s' is not mounted ",
 372                               output_dir);
 373                 }
 374                 if (mnt_flags & NTFS_MF_READONLY) {
 375                         ERROR("NTFS-3g: Filesystem on `%s' is mounted "
 376                               "read-only", output_dir);
 377                         return WIMLIB_ERR_NTFS_3G;
 378                 }
 379         #else
 380                 ERROR("wimlib was compiled without support for NTFS-3g, so");
 381                 ERROR("we cannot extract a WIM image while preserving NTFS-");
 382                 ERROR("specific information");
 383                 return WIMLIB_ERR_UNSUPPORTED;
 384         #endif
 385         }
 386         if (image == WIM_ALL_IMAGES) {
 387                 w->is_multi_image_extraction = true;
 388                 ret = extract_all_images(w, flags);
 389         } else {
 390                 w->is_multi_image_extraction = false;
 391                 ret = extract_single_image(w, image, flags);
 392         }
 393         return ret;
 394
 395 }