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

   1 /*
   2  * extract.c
   3  *
   4  * Support for extracting WIM files.
   5  *
   6  * This code does NOT contain any filesystem-specific features.  In particular,
   7  * security information (i.e. file permissions) and alternate data streams are
   8  * ignored, except possibly to read an alternate data stream that contains
   9  * symbolic link data.
  10  */
  11
  12 /*
  13  * Copyright (C) 2010 Carl Thijssen
  14  * Copyright (C) 2012 Eric Biggers
  15  *
  16  * This file is part of wimlib, a library for working with WIM files.
  17  *
  18  * wimlib is free software; you can redistribute it and/or modify it under the
  19  * terms of the GNU General Public License as published by the Free
  20  * Software Foundation; either version 3 of the License, or (at your option)
  21  * any later version.
  22  *
  23  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  24  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  25  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  26  * details.
  27  *
  28  * You should have received a copy of the GNU General Public License
  29  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  30  */
  31
  32
  33 #include <dirent.h>
  34 #include <errno.h>
  35 #include <fcntl.h>
  36 #include <string.h>
  37 #include <sys/stat.h>
  38 #include <sys/time.h>
  39 #include <unistd.h>
  40
  41 #include "config.h"
  42 #include "dentry.h"
  43 #include "lookup_table.h"
  44 #include "timestamp.h"
  45 #include "wimlib_internal.h"
  46 #include "xml.h"
  47
  48 /* Internal */
  49 #define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
  50
  51 static int extract_regular_file_linked(const struct dentry *dentry,
  52                                        const char *output_dir,
  53                                        const char *output_path,
  54                                        int extract_flags,
  55                                        struct lookup_table_entry *lte)
  56 {
  57         /* This mode overrides the normal hard-link extraction and
  58          * instead either symlinks or hardlinks *all* identical files in
  59          * the WIM, even if they are in a different image (in the case
  60          * of a multi-image extraction) */
  61         wimlib_assert(lte->extracted_file);
  62
  63         if (extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
  64                 if (link(lte->extracted_file, output_path) != 0) {
  65                         ERROR_WITH_ERRNO("Failed to hard link "
  66                                          "`%s' to `%s'",
  67                                          output_path, lte->extracted_file);
  68                         return WIMLIB_ERR_LINK;
  69                 }
  70         } else {
  71                 int num_path_components;
  72                 int num_output_dir_path_components;
  73                 size_t extracted_file_len;
  74                 char *p;
  75                 const char *p2;
  76                 size_t i;
  77
  78                 wimlib_assert(extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK);
  79
  80                 num_path_components =
  81                         get_num_path_components(dentry->full_path_utf8) - 1;
  82                 num_output_dir_path_components =
  83                         get_num_path_components(output_dir);
  84
  85                 if (extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
  86                         num_path_components++;
  87                         num_output_dir_path_components--;
  88                 }
  89                 extracted_file_len = strlen(lte->extracted_file);
  90
  91                 char buf[extracted_file_len + 3 * num_path_components + 1];
  92                 p = &buf[0];
  93
  94                 for (i = 0; i < num_path_components; i++) {
  95                         *p++ = '.';
  96                         *p++ = '.';
  97                         *p++ = '/';
  98                 }
  99                 p2 = lte->extracted_file;
 100                 while (*p2 == '/')
 101                         p2++;
 102                 while (num_output_dir_path_components--)
 103                         p2 = path_next_part(p2, NULL);
 104                 strcpy(p, p2);
 105                 if (symlink(buf, output_path) != 0) {
 106                         ERROR_WITH_ERRNO("Failed to symlink `%s' to "
 107                                          "`%s'",
 108                                          buf, lte->extracted_file);
 109                         return WIMLIB_ERR_LINK;
 110                 }
 111
 112         }
 113         return 0;
 114 }
 115
 116 static int extract_regular_file_unlinked(WIMStruct *w,
 117                                          struct dentry *dentry,
 118                                          const char *output_path,
 119                                          int extract_flags,
 120                                          struct lookup_table_entry *lte)
 121 {
 122         /* Normal mode of extraction.  Regular files and hard links are
 123          * extracted in the way that they appear in the WIM. */
 124
 125         int out_fd;
 126         int ret;
 127         struct inode *inode = dentry->inode;
 128
 129         if (!((extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE)
 130                 && (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 131                                      WIMLIB_EXTRACT_FLAG_HARDLINK))))
 132         {
 133                 /* If the dentry is one of a hard link set of at least 2
 134                  * dentries and one of the other dentries has already been
 135                  * extracted, make a hard link to the file corresponding to this
 136                  * already-extracted directory.  Otherwise, extract the file,
 137                  * and set the inode->extracted_file field so that other
 138                  * dentries in the hard link group can link to it. */
 139                 if (inode->link_count > 1) {
 140                         if (inode->extracted_file) {
 141                                 DEBUG("Extracting hard link `%s' => `%s'",
 142                                       output_path, inode->extracted_file);
 143                                 if (link(inode->extracted_file, output_path) != 0) {
 144                                         ERROR_WITH_ERRNO("Failed to hard link "
 145                                                          "`%s' to `%s'",
 146                                                          output_path,
 147                                                          inode->extracted_file);
 148                                         return WIMLIB_ERR_LINK;
 149                                 }
 150                                 return 0;
 151                         }
 152                         FREE(inode->extracted_file);
 153                         inode->extracted_file = STRDUP(output_path);
 154                         if (!inode->extracted_file) {
 155                                 ERROR("Failed to allocate memory for filename");
 156                                 return WIMLIB_ERR_NOMEM;
 157                         }
 158                 }
 159         }
 160
 161         /* Extract the contents of the file to @output_path. */
 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         if (!lte) {
 171                 /* Empty file with no lookup table entry */
 172                 DEBUG("Empty file `%s'.", output_path);
 173                 ret = 0;
 174                 goto out;
 175         }
 176
 177         ret = extract_full_wim_resource_to_fd(lte, out_fd);
 178         if (ret != 0) {
 179                 ERROR("Failed to extract resource to `%s'", output_path);
 180                 goto out;
 181         }
 182
 183 out:
 184         if (close(out_fd) != 0) {
 185                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
 186                 ret = WIMLIB_ERR_WRITE;
 187         }
 188         return ret;
 189 }
 190
 191 /*
 192  * Extracts a regular file from the WIM archive.
 193  */
 194 static int extract_regular_file(WIMStruct *w,
 195                                 struct dentry *dentry,
 196                                 const char *output_dir,
 197                                 const char *output_path,
 198                                 int extract_flags)
 199 {
 200         struct lookup_table_entry *lte;
 201         const struct inode *inode = dentry->inode;
 202
 203         lte = inode_unnamed_lte(inode, w->lookup_table);
 204
 205         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 206                               WIMLIB_EXTRACT_FLAG_HARDLINK)) && lte) {
 207                 if (++lte->out_refcnt != 1)
 208                         return extract_regular_file_linked(dentry, output_dir,
 209                                                            output_path,
 210                                                            extract_flags, lte);
 211                 FREE(lte->extracted_file);
 212                 lte->extracted_file = STRDUP(output_path);
 213                 if (!lte->extracted_file)
 214                         return WIMLIB_ERR_NOMEM;
 215         }
 216
 217         return extract_regular_file_unlinked(w, dentry, output_path,
 218                                              extract_flags, lte);
 219
 220 }
 221
 222 static int extract_symlink(const struct dentry *dentry, const char *output_path,
 223                            const WIMStruct *w)
 224 {
 225         char target[4096];
 226         ssize_t ret = inode_readlink(dentry->inode, target, sizeof(target), w);
 227         if (ret <= 0) {
 228                 ERROR("Could not read the symbolic link from dentry `%s'",
 229                       dentry->full_path_utf8);
 230                 return WIMLIB_ERR_INVALID_DENTRY;
 231         }
 232         ret = symlink(target, output_path);
 233         if (ret != 0) {
 234                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 235                                  output_path, target);
 236                 return WIMLIB_ERR_LINK;
 237         }
 238         return 0;
 239 }
 240
 241 /*
 242  * Extracts a directory from the WIM archive.
 243  *
 244  * @dentry:             The directory entry for the directory.
 245  * @output_path:        The path to which the directory is to be extracted to.
 246  * @return:             True on success, false on failure.
 247  */
 248 static int extract_directory(const char *output_path, bool is_root)
 249 {
 250         int ret;
 251         struct stat stbuf;
 252         ret = stat(output_path, &stbuf);
 253         if (ret == 0) {
 254                 if (S_ISDIR(stbuf.st_mode)) {
 255                         if (!is_root)
 256                                 WARNING("`%s' already exists", output_path);
 257                         return 0;
 258                 } else {
 259                         ERROR("`%s' is not a directory", output_path);
 260                         return WIMLIB_ERR_MKDIR;
 261                 }
 262         } else {
 263                 if (errno != ENOENT) {
 264                         ERROR_WITH_ERRNO("Failed to stat `%s'", output_path);
 265                         return WIMLIB_ERR_STAT;
 266                 }
 267         }
 268         /* Compute the output path directory to the directory. */
 269         if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP |
 270                                S_IROTH | S_IXOTH) != 0) {
 271                 ERROR_WITH_ERRNO("Cannot create directory `%s'",
 272                                  output_path);
 273                 return WIMLIB_ERR_MKDIR;
 274         }
 275         return 0;
 276 }
 277
 278 struct extract_args {
 279         WIMStruct *w;
 280         int extract_flags;
 281         const char *output_dir;
 282 };
 283
 284 /*
 285  * Extracts a file, directory, or symbolic link from the WIM archive.  For use
 286  * in for_dentry_in_tree().
 287  */
 288 static int extract_dentry(struct dentry *dentry, void *arg)
 289 {
 290         struct extract_args *args = arg;
 291         WIMStruct *w = args->w;
 292         int extract_flags = args->extract_flags;
 293         size_t len = strlen(args->output_dir);
 294         char output_path[len + dentry->full_path_utf8_len + 1];
 295
 296         if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) {
 297                 wimlib_assert(dentry->full_path_utf8);
 298                 puts(dentry->full_path_utf8);
 299         }
 300
 301         memcpy(output_path, args->output_dir, len);
 302         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 303         output_path[len + dentry->full_path_utf8_len] = '\0';
 304
 305         if (dentry_is_symlink(dentry))
 306                 return extract_symlink(dentry, output_path, w);
 307         else if (dentry_is_directory(dentry))
 308                 return extract_directory(output_path, dentry_is_root(dentry));
 309         else
 310                 return extract_regular_file(w, dentry, args->output_dir,
 311                                             output_path, extract_flags);
 312 }
 313
 314 /* Apply timestamp to extracted file */
 315 static int apply_dentry_timestamps(struct dentry *dentry, void *arg)
 316 {
 317         struct extract_args *args = arg;
 318         size_t len = strlen(args->output_dir);
 319         char output_path[len + dentry->full_path_utf8_len + 1];
 320
 321         memcpy(output_path, args->output_dir, len);
 322         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 323         output_path[len + dentry->full_path_utf8_len] = '\0';
 324
 325         struct timeval tv[2];
 326         wim_timestamp_to_timeval(dentry->inode->last_access_time, &tv[0]);
 327         wim_timestamp_to_timeval(dentry->inode->last_write_time, &tv[1]);
 328         if (lutimes(output_path, tv) != 0) {
 329                 WARNING("Failed to set timestamp on file `%s': %s",
 330                         output_path, strerror(errno));
 331         }
 332         return 0;
 333 }
 334
 335
 336 static int extract_single_image(WIMStruct *w, int image,
 337                                 const char *output_dir, int extract_flags)
 338 {
 339         DEBUG("Extracting image %d", image);
 340
 341         int ret;
 342         ret = wimlib_select_image(w, image);
 343         if (ret != 0)
 344                 return ret;
 345
 346         struct extract_args args = {
 347                 .w = w,
 348                 .extract_flags = extract_flags,
 349                 .output_dir = output_dir,
 350         };
 351
 352         ret = for_dentry_in_tree(wim_root_dentry(w), extract_dentry, &args);
 353         if (ret != 0)
 354                 return ret;
 355         return for_dentry_in_tree_depth(wim_root_dentry(w),
 356                                         apply_dentry_timestamps, &args);
 357
 358 }
 359
 360
 361 /* Extracts all images from the WIM to @output_dir, with the images placed in
 362  * subdirectories named by their image names. */
 363 static int extract_all_images(WIMStruct *w, const char *output_dir,
 364                               int extract_flags)
 365 {
 366         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
 367         size_t output_path_len = strlen(output_dir);
 368         char buf[output_path_len + 1 + image_name_max_len + 1];
 369         int ret;
 370         int image;
 371         const char *image_name;
 372
 373         DEBUG("Attempting to extract all images from `%s' to `%s'",
 374               w->filename, output_dir);
 375
 376         ret = extract_directory(output_dir, true);
 377         if (ret != 0)
 378                 return ret;
 379
 380         memcpy(buf, output_dir, output_path_len);
 381         buf[output_path_len] = '/';
 382         for (image = 1; image <= w->hdr.image_count; image++) {
 383
 384                 image_name = wimlib_get_image_name(w, image);
 385                 if (*image_name) {
 386                         strcpy(buf + output_path_len + 1, image_name);
 387                 } else {
 388                         /* Image name is empty. Use image number instead */
 389                         sprintf(buf + output_path_len + 1, "%d", image);
 390                 }
 391                 ret = extract_single_image(w, image, buf, extract_flags);
 392                 if (ret != 0)
 393                         return ret;
 394         }
 395         return 0;
 396 }
 397
 398
 399 /* Extracts a single image or all images from a WIM file. */
 400 WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
 401                                    const char *output_dir, int flags,
 402                                    WIMStruct **additional_swms,
 403                                    unsigned num_additional_swms)
 404 {
 405         struct lookup_table *joined_tab, *w_tab_save;
 406         int ret;
 407
 408         DEBUG("w->filename = %s, image = %d, output_dir = %s, flags = 0x%x, "
 409               "num_additional_swms = %u",
 410               w->filename, image, output_dir, flags, num_additional_swms);
 411
 412         if (!w || !output_dir)
 413                 return WIMLIB_ERR_INVALID_PARAM;
 414
 415         if ((flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 416                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 417                 return WIMLIB_ERR_INVALID_PARAM;
 418
 419         ret = verify_swm_set(w, additional_swms, num_additional_swms);
 420         if (ret != 0)
 421                 return ret;
 422
 423         if (num_additional_swms) {
 424                 ret = new_joined_lookup_table(w, additional_swms,
 425                                               num_additional_swms, &joined_tab);
 426                 if (ret != 0)
 427                         return ret;
 428                 w_tab_save = w->lookup_table;
 429                 w->lookup_table = joined_tab;
 430         }
 431
 432
 433         for_lookup_table_entry(w->lookup_table, zero_out_refcnts, NULL);
 434
 435         if (image == WIM_ALL_IMAGES) {
 436                 flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 437                 ret = extract_all_images(w, output_dir, flags);
 438         } else {
 439                 flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 440                 ret = extract_single_image(w, image, output_dir, flags);
 441         }
 442         if (num_additional_swms) {
 443                 free_lookup_table(w->lookup_table);
 444                 w->lookup_table = w_tab_save;
 445         }
 446         return ret;
 447
 448 }