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                 /* This dentry is one of a hard link set of at least 2 dentries.
 131                  * If one of the other dentries has already been extracted, make
 132                  * a hard link to the file corresponding to this
 133                  * already-extracted directory.  Otherwise, extract the
 134                  * file, and set the dentry->extracted_file field so that other
 135                  * dentries in the hard link group can link to it. */
 136                 struct dentry *other;
 137                 if (inode->extracted_file) {
 138                         DEBUG("Extracting hard link `%s' => `%s'",
 139                               output_path, inode->extracted_file);
 140                         if (link(inode->extracted_file, output_path) != 0) {
 141                                 ERROR_WITH_ERRNO("Failed to hard link "
 142                                                  "`%s' to `%s'",
 143                                                  output_path,
 144                                                  inode->extracted_file);
 145                                 return WIMLIB_ERR_LINK;
 146                         }
 147                         return 0;
 148                 }
 149                 if (inode->link_count > 1) {
 150                         FREE(inode->extracted_file);
 151                         inode->extracted_file = STRDUP(output_path);
 152                         if (!inode->extracted_file) {
 153                                 ERROR("Failed to allocate memory for filename");
 154                                 return WIMLIB_ERR_NOMEM;
 155                         }
 156                 }
 157         }
 158
 159         /* Extract the contents of the file to @output_path. */
 160
 161         out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
 162         if (out_fd == -1) {
 163                 ERROR_WITH_ERRNO("Failed to open the file `%s' for writing",
 164                                  output_path);
 165                 return WIMLIB_ERR_OPEN;
 166         }
 167
 168         if (!lte) {
 169                 /* Empty file with no lookup table entry */
 170                 DEBUG("Empty file `%s'.", output_path);
 171                 ret = 0;
 172                 goto done;
 173         }
 174
 175         ret = extract_full_wim_resource_to_fd(lte, out_fd);
 176         if (ret != 0) {
 177                 ERROR("Failed to extract resource to `%s'", output_path);
 178                 goto done;
 179         }
 180
 181 done:
 182         if (close(out_fd) != 0) {
 183                 ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
 184                 ret = WIMLIB_ERR_WRITE;
 185         }
 186         return ret;
 187 }
 188
 189 /*
 190  * Extracts a regular file from the WIM archive.
 191  */
 192 static int extract_regular_file(WIMStruct *w,
 193                                 struct dentry *dentry,
 194                                 const char *output_dir,
 195                                 const char *output_path,
 196                                 int extract_flags)
 197 {
 198         struct lookup_table_entry *lte;
 199         const struct inode *inode = dentry->inode;
 200
 201         lte = inode_unnamed_lte(inode, w->lookup_table);
 202
 203         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
 204                               WIMLIB_EXTRACT_FLAG_HARDLINK)) && lte) {
 205                 if (++lte->out_refcnt != 1)
 206                         return extract_regular_file_linked(dentry, output_dir,
 207                                                            output_path,
 208                                                            extract_flags, lte);
 209                 lte->extracted_file = STRDUP(output_path);
 210                 if (!lte->extracted_file)
 211                         return WIMLIB_ERR_NOMEM;
 212         }
 213
 214         return extract_regular_file_unlinked(w, dentry, output_path,
 215                                              extract_flags, lte);
 216
 217 }
 218
 219 static int extract_symlink(const struct dentry *dentry, const char *output_path,
 220                            const WIMStruct *w)
 221 {
 222         char target[4096];
 223         ssize_t ret = inode_readlink(dentry->inode, target, sizeof(target), w);
 224         if (ret <= 0) {
 225                 ERROR("Could not read the symbolic link from dentry `%s'",
 226                       dentry->full_path_utf8);
 227                 return WIMLIB_ERR_INVALID_DENTRY;
 228         }
 229         ret = symlink(target, output_path);
 230         if (ret != 0) {
 231                 ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
 232                                  output_path, target);
 233                 return WIMLIB_ERR_LINK;
 234         }
 235         return 0;
 236 }
 237
 238 /*
 239  * Extracts a directory from the WIM archive.
 240  *
 241  * @dentry:             The directory entry for the directory.
 242  * @output_path:        The path to which the directory is to be extracted to.
 243  * @return:             True on success, false on failure.
 244  */
 245 static int extract_directory(const char *output_path, bool is_root)
 246 {
 247         int ret;
 248         struct stat stbuf;
 249         ret = stat(output_path, &stbuf);
 250         if (ret == 0) {
 251                 if (S_ISDIR(stbuf.st_mode)) {
 252                         if (!is_root)
 253                                 WARNING("`%s' already exists", output_path);
 254                         return 0;
 255                 } else {
 256                         ERROR("`%s' is not a directory", output_path);
 257                         return WIMLIB_ERR_MKDIR;
 258                 }
 259         } else {
 260                 if (errno != ENOENT) {
 261                         ERROR_WITH_ERRNO("Failed to stat `%s'", output_path);
 262                         return WIMLIB_ERR_STAT;
 263                 }
 264         }
 265         /* Compute the output path directory to the directory. */
 266         if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP |
 267                                S_IROTH | S_IXOTH) != 0) {
 268                 ERROR_WITH_ERRNO("Cannot create directory `%s'",
 269                                  output_path);
 270                 return WIMLIB_ERR_MKDIR;
 271         }
 272         return 0;
 273 }
 274
 275 struct extract_args {
 276         WIMStruct *w;
 277         int extract_flags;
 278         const char *output_dir;
 279 };
 280
 281 /*
 282  * Extracts a file, directory, or symbolic link from the WIM archive.  For use
 283  * in for_dentry_in_tree().
 284  */
 285 static int extract_dentry(struct dentry *dentry, void *arg)
 286 {
 287         struct extract_args *args = arg;
 288         WIMStruct *w = args->w;
 289         int extract_flags = args->extract_flags;
 290         size_t len = strlen(args->output_dir);
 291         char output_path[len + dentry->full_path_utf8_len + 1];
 292
 293         if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) {
 294                 wimlib_assert(dentry->full_path_utf8);
 295                 puts(dentry->full_path_utf8);
 296         }
 297
 298         memcpy(output_path, args->output_dir, len);
 299         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 300         output_path[len + dentry->full_path_utf8_len] = '\0';
 301
 302         if (dentry_is_symlink(dentry))
 303                 return extract_symlink(dentry, output_path, w);
 304         else if (dentry_is_directory(dentry))
 305                 return extract_directory(output_path, dentry_is_root(dentry));
 306         else
 307                 return extract_regular_file(w, dentry, args->output_dir,
 308                                             output_path, extract_flags);
 309 }
 310
 311 /* Apply timestamp to extracted file */
 312 static int apply_dentry_timestamps(struct dentry *dentry, void *arg)
 313 {
 314         struct extract_args *args = arg;
 315         size_t len = strlen(args->output_dir);
 316         char output_path[len + dentry->full_path_utf8_len + 1];
 317
 318         memcpy(output_path, args->output_dir, len);
 319         memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
 320         output_path[len + dentry->full_path_utf8_len] = '\0';
 321
 322         struct timeval tv[2];
 323         wim_timestamp_to_timeval(dentry->inode->last_access_time, &tv[0]);
 324         wim_timestamp_to_timeval(dentry->inode->last_write_time, &tv[1]);
 325         if (lutimes(output_path, tv) != 0) {
 326                 WARNING("Failed to set timestamp on file `%s': %s",
 327                         output_path, strerror(errno));
 328         }
 329         return 0;
 330 }
 331
 332
 333 static int extract_single_image(WIMStruct *w, int image,
 334                                 const char *output_dir, int extract_flags)
 335 {
 336         DEBUG("Extracting image %d", image);
 337
 338         int ret;
 339         ret = wimlib_select_image(w, image);
 340         if (ret != 0)
 341                 return ret;
 342
 343         struct extract_args args = {
 344                 .w = w,
 345                 .extract_flags = extract_flags,
 346                 .output_dir = output_dir,
 347         };
 348
 349         ret = for_dentry_in_tree(wim_root_dentry(w), extract_dentry, &args);
 350         if (ret != 0)
 351                 return ret;
 352         return for_dentry_in_tree_depth(wim_root_dentry(w),
 353                                         apply_dentry_timestamps, &args);
 354
 355 }
 356
 357
 358 /* Extracts all images from the WIM to @output_dir, with the images placed in
 359  * subdirectories named by their image names. */
 360 static int extract_all_images(WIMStruct *w, const char *output_dir,
 361                               int extract_flags)
 362 {
 363         size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
 364         size_t output_path_len = strlen(output_dir);
 365         char buf[output_path_len + 1 + image_name_max_len + 1];
 366         int ret;
 367         int image;
 368         const char *image_name;
 369
 370         DEBUG("Attempting to extract all images from `%s' to `%s'",
 371               w->filename, output_dir);
 372
 373         ret = extract_directory(output_dir, true);
 374         if (ret != 0)
 375                 return ret;
 376
 377         memcpy(buf, output_dir, output_path_len);
 378         buf[output_path_len] = '/';
 379         for (image = 1; image <= w->hdr.image_count; image++) {
 380
 381                 image_name = wimlib_get_image_name(w, image);
 382                 if (*image_name) {
 383                         strcpy(buf + output_path_len + 1, image_name);
 384                 } else {
 385                         /* Image name is empty. Use image number instead */
 386                         sprintf(buf + output_path_len + 1, "%d", image);
 387                 }
 388                 ret = extract_single_image(w, image, buf, extract_flags);
 389                 if (ret != 0)
 390                         return ret;
 391         }
 392         return 0;
 393 }
 394
 395
 396 /* Extracts a single image or all images from a WIM file. */
 397 WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
 398                                    const char *output_dir, int flags,
 399                                    WIMStruct **additional_swms,
 400                                    unsigned num_additional_swms)
 401 {
 402         struct lookup_table *joined_tab, *w_tab_save;
 403         int ret;
 404
 405         DEBUG("w->filename = %s, image = %d, output_dir = %s, flags = 0x%x, "
 406               "num_additional_swms = %u",
 407               w->filename, image, output_dir, flags, num_additional_swms);
 408
 409         if (!w || !output_dir)
 410                 return WIMLIB_ERR_INVALID_PARAM;
 411
 412         if ((flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 413                         == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
 414                 return WIMLIB_ERR_INVALID_PARAM;
 415
 416         ret = verify_swm_set(w, additional_swms, num_additional_swms);
 417         if (ret != 0)
 418                 return ret;
 419
 420         if (num_additional_swms) {
 421                 ret = new_joined_lookup_table(w, additional_swms,
 422                                               num_additional_swms, &joined_tab);
 423                 if (ret != 0)
 424                         return ret;
 425                 w_tab_save = w->lookup_table;
 426                 w->lookup_table = joined_tab;
 427         }
 428
 429
 430         for_lookup_table_entry(w->lookup_table, zero_out_refcnts, NULL);
 431
 432         if (image == WIM_ALL_IMAGES) {
 433                 flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 434                 ret = extract_all_images(w, output_dir, flags);
 435         } else {
 436                 flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
 437                 ret = extract_single_image(w, image, output_dir, flags);
 438         }
 439         if (num_additional_swms) {
 440                 free_lookup_table(w->lookup_table);
 441                 w->lookup_table = w_tab_save;
 442         }
 443         return ret;
 444
 445 }