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
  28 #include <dirent.h>
  29 #include <errno.h>
  30 #include <fcntl.h>
  31 #include <string.h>
  32 #include <sys/stat.h>
  33 #include <sys/time.h>
  34 #include <unistd.h>
  35
  36 #include "config.h"
  37 #include "dentry.h"
  38 #include "lookup_table.h"
  39 #include "timestamp.h"
  40 #include "wimlib_internal.h"
  41 #include "xml.h"
  42
  43 /* Internal */
  44 #define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
  45
  46 static int extract_regular_file_linked(const struct dentry *dentry,
  47                                        const char *output_dir,
  48                                        const char *output_path,
  49                                        int extract_flags,
  50                                        struct lookup_table_entry *lte)
  51 {
  52         /* This mode overrides the normal hard-link extraction and
  53          * instead either symlinks or hardlinks *all* identical files in
  54          * the WIM, even if they are in a different image (in the case
  55          * of a multi-image extraction) */
  56
  57         wimlib_assert(lte->file_on_disk);
  58
  59
  60         if (extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
  61                 if (link(lte->file_on_disk, output_path) != 0) {
  62                         ERROR_WITH_ERRNO("Failed to hard link "
  63                                          "`%s' to `%s'",
  64                                          output_path, lte->file_on_disk);
  65                         return WIMLIB_ERR_LINK;
  66                 }
  67         } else {
  68                 int num_path_components;
  69                 int num_output_dir_path_components;
  70                 size_t file_on_disk_len;
  71                 char *p;
  72                 const char *p2;
  73                 size_t i;
  74
  75                 wimlib_assert(extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK);
  76
  77                 num_path_components =
  78                         get_num_path_components(dentry->full_path_utf8) - 1;
  79                 num_output_dir_path_components =
  80                         get_num_path_components(output_dir);
  81
  82                 if (extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
  83                         num_path_components++;
  84                         num_output_dir_path_components--;
  85                 }
  86                 file_on_disk_len = strlen(lte->file_on_disk);
  87
  88                 char buf[file_on_disk_len + 3 * num_path_components + 1];
  89                 p = &buf[0];
  90
  91                 for (i = 0; i < num_path_components; i++) {
  92                         *p++ = '.';
  93                         *p++ = '.';
  94                         *p++ = '/';
  95                 }
  96                 p2 = lte->file_on_disk;
  97                 while (*p2 == '/')
  98                         p2++;
  99                 while (num_output_dir_path_components--)
 100                         p2 = path_next_part(p2, NULL);
 101                 strcpy(p, p2);
 102                 if (symlink(buf, output_path) != 0) {
 103                         ERROR_WITH_ERRNO("Failed to symlink `%s' to "
 104                                          "`%s'",
 105                                          buf, lte->file_on_disk);
 106                         return WIMLIB_ERR_LINK;
 107                 }
 108
 109         }
 110         return 0;
 111 }
 112
 113 static int extract_regular_file_unlinked(WIMStruct *w,
 114                                          struct dentry *dentry,
 115                                          const char *output_path,
 116                                          int extract_flags,
 117                                          struct lookup_table_entry *lte)
 118 {
 119         /* Normal mode of extraction.  Regular files and hard links are
 120          * extracted in the way that they appear in the WIM. */
 121
 122         int out_fd;
 123         const struct resource_entry *res_entry;
 124         int ret;
 125         const struct list_head *head = &dentry->link_group_list;
 126
 127         if (head->next != head) {
 128                 /* This dentry is one of a hard link set of at least 2 dentries.
 129                  * If one of the other dentries has already been extracted, make
 130                  * a hard link to the file corresponding to this
 131                  * already-extracted directory.  Otherwise, extract the
 132                  * file, and set the dentry->extracted_file field so that other
 133                  * dentries in the hard link group can link to it. */
 134                 struct dentry *other;
 135                 list_for_each_entry(other, head, link_group_list) {
 136                         if (other->extracted_file) {
 137                                 DEBUG("Extracting hard link `%s' => `%s'",
 138                                       output_path, other->extracted_file);
 139                                 if (link(other->extracted_file, output_path) != 0) {
 140                                         ERROR_WITH_ERRNO("Failed to hard link "
 141                                                          "`%s' to `%s'",
 142                                                          output_path,
 143                                                          other->extracted_file);
 144                                         return WIMLIB_ERR_LINK;
 145                                 }
 146                                 return 0;
 147                         }
 148                 }
 149                 FREE(dentry->extracted_file);
 150                 dentry->extracted_file = STRDUP(output_path);
 151                 if (!dentry->extracted_file) {
 152                         ERROR("Failed to allocate memory for filename");
 153                         return WIMLIB_ERR_NOMEM;
 154                 }
 155         }
 156
 157         /* Extract the contents of the file to @output_path. */
 158
 159         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
 160         if (out_fd == -1) {
 161                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
 162                                  output_path);
 163                 return WIMLIB_ERR_OPEN;
 164         }
 165
 166         if (!lte) {
 167                 /* Empty file with no lookup table entry */
 168                 DEBUG("Empty file `%s'.", output_path);
 169                 ret = 0;
 170                 goto done;
 171         }
 172
 173         ret = extract_full_wim_resource_to_fd(lte, out_fd);
 174         if (ret != 0) {
 175                 ERROR("Failed to extract resource to `%s'", output_path);
 176                 goto done;
 177         }
 178
 179         if (extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
 180                 /* Mark the lookup table entry to indicate this file has been
 181                  * extracted. */
 182                 lte->out_refcnt++;
 183                 FREE(lte->file_on_disk);
 184                 lte->file_on_disk = STRDUP(output_path);
 185                 if (!lte->file_on_disk)
 186                         ret = WIMLIB_ERR_NOMEM;
 187         }
 188 done:
 189         if (close(out_fd) != 0) {
 190                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
 191                 ret = WIMLIB_ERR_WRITE;
 192         }
 193         return ret;
 194 }
 195
 196 /*
 197  * Extracts a regular file from the WIM archive.
 198  */
 199 static int extract_regular_file(WIMStruct *w,
 200                                 struct dentry *dentry,
 201                                 const char *output_dir,
 202                                 const char *output_path,
 203                                 int extract_flags)
 204 {
 205         struct lookup_table_entry *lte;
 206
 207         lte = dentry_first_lte(dentry, w->lookup_table);
 208
 209         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 210                               WIMLIB_EXTRACT_FLAG_HARDLINK)) && lte) {
 211                 if (lte->out_refcnt++ != 0)
 212                         return extract_regular_file_linked(dentry, output_dir,
 213                                                            output_path,
 214                                                            extract_flags, lte);
 215                 lte->file_on_disk = STRDUP(output_path);
 216                 if (!lte->file_on_disk)
 217                         return WIMLIB_ERR_NOMEM;
 218         }
 219
 220         return extract_regular_file_unlinked(w, dentry, output_path,
 221                                              extract_flags, lte);
 222
 223 }
 224
 225 static int extract_symlink(const struct dentry *dentry, const char *output_path,
 226                            const WIMStruct *w)
 227 {
 228         char target[4096];
 229         ssize_t ret = dentry_readlink(dentry, target, sizeof(target), w);
 230         if (ret <= 0) {
 231                 ERROR("Could not read the symbolic link from dentry `%s'",
 232                       dentry->full_path_utf8);
 233                 return WIMLIB_ERR_INVALID_DENTRY;
 234         }
 235         ret = symlink(target, output_path);
 236         if (ret != 0) {
 237                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 238                                  output_path, target);
 239                 return WIMLIB_ERR_LINK;
 240         }
 241         return 0;
 242 }
 243
 244 /*
 245  * Extracts a directory from the WIM archive.
 246  *
 247  * @dentry:             The directory entry for the directory.
 248  * @output_path:        The path to which the directory is to be extracted to.
 249  * @return:             True on success, false on failure.
 250  */
 251 static int extract_directory(const char *output_path, bool is_root)
 252 {
 253         int ret;
 254         struct stat stbuf;
 255         ret = stat(output_path, &stbuf);
 256         if (ret == 0) {
 257                 if (S_ISDIR(stbuf.st_mode)) {
 258                         if (!is_root)
 259                                 WARNING("`%s' already exists", output_path);
 260                         return 0;
 261                 } else {
 262                         ERROR("`%s' is not a directory", output_path);
 263                         return WIMLIB_ERR_MKDIR;
 264                 }
 265         } else {
 266                 if (errno != ENOENT) {
 267                         ERROR_WITH_ERRNO("Failed to stat `%s'", output_path);
 268                         return WIMLIB_ERR_STAT;
 269                 }
 270         }
 271         /* Compute the output path directory to the directory. */
 272         if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP |
 273                                S_IROTH | S_IXOTH) != 0) {
 274                 ERROR_WITH_ERRNO("Cannot create directory `%s'",
 275                                  output_path);
 276                 return WIMLIB_ERR_MKDIR;
 277         }
 278         return 0;
 279 }
 280
 281 struct extract_args {
 282         WIMStruct *w;
 283         int extract_flags;
 284         const char *output_dir;
 285 #ifdef WITH_NTFS_3G
 286         struct SECURITY_API *scapi;
 287 #endif
 288 };
 289
 290 /*
 291  * Extracts a file, directory, or symbolic link from the WIM archive.  For use
 292  * in for_dentry_in_tree().
 293  */
 294 static int extract_dentry(struct dentry *dentry, void *arg)
 295 {
 296         struct extract_args *args = arg;
 297         WIMStruct *w = args->w;
 298         int extract_flags = args->extract_flags;
 299         size_t len = strlen(args->output_dir);
 300         char output_path[len + dentry->full_path_utf8_len + 1];
 301         int ret;
 302
 303         if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) {
 304                 wimlib_assert(dentry->full_path_utf8);
 305                 puts(dentry->full_path_utf8);
 306         }
 307
 308         memcpy(output_path, args->output_dir, len);
 309         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 310         output_path[len + dentry->full_path_utf8_len] = '\0';
 311
 312         if (dentry_is_symlink(dentry))
 313                 ret = extract_symlink(dentry, output_path, w);
 314         else if (dentry_is_directory(dentry))
 315                 ret = extract_directory(output_path, dentry_is_root(dentry));
 316         else
 317                 ret = extract_regular_file(w, dentry, args->output_dir,
 318                                             output_path, extract_flags);
 319         if (ret != 0)
 320                 return ret;
 321
 322         return 0;
 323 }
 324
 325 static int apply_dentry_timestamps(struct dentry *dentry, void *arg)
 326 {
 327         struct extract_args *args = arg;
 328         size_t len = strlen(args->output_dir);
 329         char output_path[len + dentry->full_path_utf8_len + 1];
 330
 331         memcpy(output_path, args->output_dir, len);
 332         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 333         output_path[len + dentry->full_path_utf8_len] = '\0';
 334
 335         struct timeval tv[2];
 336         wim_timestamp_to_timeval(dentry->last_access_time, &tv[0]);
 337         wim_timestamp_to_timeval(dentry->last_write_time, &tv[1]);
 338         if (lutimes(output_path, tv) != 0) {
 339                 WARNING("Failed to set timestamp on file `%s': %s",
 340                         output_path, strerror(errno));
 341         }
 342         return 0;
 343 }
 344
 345
 346 static int extract_single_image(WIMStruct *w, int image,
 347                                 const char *output_dir, int extract_flags)
 348 {
 349         DEBUG("Extracting image %d", image);
 350
 351         int ret;
 352         ret = wimlib_select_image(w, image);
 353         if (ret != 0)
 354                 return ret;
 355
 356         struct extract_args args = {
 357                 .w = w,
 358                 .extract_flags = extract_flags,
 359                 .output_dir = output_dir,
 360         #ifdef WITH_NTFS_3G
 361                 .scapi = NULL
 362         #endif
 363         };
 364
 365         ret = for_dentry_in_tree(wim_root_dentry(w), extract_dentry, &args);
 366         if (ret != 0)
 367                 return ret;
 368         return for_dentry_in_tree_depth(wim_root_dentry(w),
 369                                         apply_dentry_timestamps, &args);
 370
 371 }
 372
 373
 374 /* Extracts all images from the WIM to @output_dir, with the images placed in
 375  * subdirectories named by their image names. */
 376 static int extract_all_images(WIMStruct *w, const char *output_dir,
 377                               int extract_flags)
 378 {
 379         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
 380         size_t output_path_len = strlen(output_dir);
 381         char buf[output_path_len + 1 + image_name_max_len + 1];
 382         int ret;
 383         int image;
 384         const char *image_name;
 385
 386         DEBUG("Attempting to extract all images from `%s'", w->filename);
 387
 388         ret = extract_directory(output_dir, true);
 389         if (ret != 0)
 390                 return ret;
 391
 392         memcpy(buf, output_dir, output_path_len);
 393         buf[output_path_len] = '/';
 394         for (image = 1; image <= w->hdr.image_count; image++) {
 395
 396                 image_name = wimlib_get_image_name(w, image);
 397                 if (*image_name) {
 398                         strcpy(buf + output_path_len + 1, image_name);
 399                 } else {
 400                         /* Image name is empty. Use image number instead */
 401                         sprintf(buf + output_path_len + 1, "%d", image);
 402                 }
 403                 ret = extract_single_image(w, image, buf, extract_flags);
 404                 if (ret != 0)
 405                         goto done;
 406         }
 407 done:
 408         /* Restore original output directory */
 409         buf[output_path_len + 1] = '\0';
 410         return 0;
 411 }
 412
 413 /* Extracts a single image or all images from a WIM file. */
 414 WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
 415                                    const char *output_dir, int flags)
 416 {
 417         if (!output_dir)
 418                 return WIMLIB_ERR_INVALID_PARAM;
 419
 420         if ((flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 421                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 422                 return WIMLIB_ERR_INVALID_PARAM;
 423
 424         for_lookup_table_entry(w->lookup_table, zero_out_refcnts, NULL);
 425
 426         if (image == WIM_ALL_IMAGES) {
 427                 flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 428                 return extract_all_images(w, output_dir, flags);
 429         } else {
 430                 flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 431                 return extract_single_image(w, image, output_dir, flags);
 432         }
 433
 434 }