]> wimlib.net Git - wimlib/blobdiff - src/extract.c
More accurate timestamps
[wimlib] / src / extract.c
index a04e3f2ea81dc96c8c0704677255ed9c73b25e07..121966022b77bcc7c089b059e87c97fcef156622 100644 (file)
 #include <ntfs-3g/security.h>
 #endif
 
-/* Sets and creates the directory to which files are to be extracted when
- * extracting files from the WIM. */
-static int set_output_dir(WIMStruct *w, const char *dir)
+/* Internal */
+#define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
+
+static int extract_regular_file_linked(const struct dentry *dentry, 
+                                      const char *output_dir,
+                                      const char *output_path,
+                                      int extract_flags,
+                                      struct lookup_table_entry *lte)
 {
-       char *p;
-       DEBUG("Setting output directory to `%s'", dir);
+       /* This mode overrides the normal hard-link extraction and
+        * instead either symlinks or hardlinks *all* identical files in
+        * the WIM, even if they are in a different image (in the case
+        * of a multi-image extraction) */
 
-       p = STRDUP(dir);
-       if (!p) {
-               ERROR("Out of memory");
-               return WIMLIB_ERR_NOMEM;
-       }
+       wimlib_assert(lte->file_on_disk);
 
-       if (mkdir(dir, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH) != 0) {
-               if (errno == EEXIST) {
-                       DEBUG("`%s' already exists", dir);
-                       goto done;
+
+       if (extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
+               if (link(lte->file_on_disk, output_path) != 0) {
+                       ERROR_WITH_ERRNO("Failed to hard link "
+                                        "`%s' to `%s'",
+                                        output_path, lte->file_on_disk);
+                       return WIMLIB_ERR_LINK;
                }
-               ERROR_WITH_ERRNO("Cannot create directory `%s'", dir);
-               FREE(p);
-               return WIMLIB_ERR_MKDIR;
        } else {
-               DEBUG("Created directory `%s'", dir);
+               int num_path_components;
+               int num_output_dir_path_components;
+               size_t file_on_disk_len;
+               char *p;
+               const char *p2;
+               size_t i;
+
+               wimlib_assert(extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK);
+
+               num_path_components = 
+                       get_num_path_components(dentry->full_path_utf8) - 1;
+               num_output_dir_path_components =
+                       get_num_path_components(output_dir);
+
+               if (extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
+                       num_path_components++;
+                       num_output_dir_path_components--;
+               }
+               file_on_disk_len = strlen(lte->file_on_disk);
+
+               char buf[file_on_disk_len + 3 * num_path_components + 1];
+               p = &buf[0];
+
+               for (i = 0; i < num_path_components; i++) {
+                       *p++ = '.';
+                       *p++ = '.';
+                       *p++ = '/';
+               }
+               p2 = lte->file_on_disk;
+               while (*p2 == '/')
+                       p2++;
+               while (num_output_dir_path_components--)
+                       p2 = path_next_part(p2, NULL);
+               strcpy(p, p2);
+               if (symlink(buf, output_path) != 0) {
+                       ERROR_WITH_ERRNO("Failed to symlink `%s' to "
+                                        "`%s'",
+                                        buf, lte->file_on_disk);
+                       return WIMLIB_ERR_LINK;
+               }
+
        }
-done:
-       FREE(w->output_dir);
-       w->output_dir = p;
        return 0;
 }
 
-/* 
- * Extracts a regular file from the WIM archive. 
- *
- * @dentry:            The directory entry for the file, which must be a
- *                             regular file.
- * @output_path:       The path to which the file is to be extracted.
- * @lookup_table:      The lookup table for the WIM file.
- * @wim_fp:            The FILE* for the WIM, opened for reading.
- * @wim_ctype:         The type of compression used in the WIM.
- * @link_type:         One of WIM_LINK_TYPE_*; specifies what to do with
- *                     files that are hard-linked inside the WIM.
- * @is_multi_image_extraction: 
- *                     True if the image currently being extracted is just one 
- *                     image of a multi-image extraction.  This is needed so
- *                     that cross-image symbolic links can be created
- *                     correctly.
- */
-static int extract_regular_file(WIMStruct *w, 
-                               const struct dentry *dentry, 
-                               const char *output_path,
-                               int extract_flags)
+static int extract_regular_file_unlinked(WIMStruct *w,
+                                        struct dentry *dentry, 
+                                        const char *output_path,
+                                        int extract_flags,
+                                        struct lookup_table_entry *lte)
 {
-       struct lookup_table_entry *lte;
-       int ret;
+       /* Normal mode of extraction.  Regular files and hard links are
+        * extracted in the way that they appear in the WIM. */
+
        int out_fd;
        const struct resource_entry *res_entry;
-
-       lte = lookup_resource(w->lookup_table, dentry_hash(dentry));
-
-       /* If we already extracted the same file or a hard link copy of it, we
-        * may be able to simply create a link.  The exact action is specified
-        * by the current @link_type. */
-       if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK)) &&
-             lte && lte->out_refcnt != 0)
-       {
-               wimlib_assert(lte->file_on_disk);
-
-               if (extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) {
-                       if (link(lte->file_on_disk, output_path) != 0) {
-                               ERROR_WITH_ERRNO("Failed to hard link "
-                                                "`%s' to `%s'",
-                                                output_path, lte->file_on_disk);
-                               return WIMLIB_ERR_LINK;
-                       }
-               } else {
-                       int num_path_components;
-                       int num_output_dir_path_components;
-                       size_t file_on_disk_len;
-                       char *p;
-                       const char *p2;
-                       size_t i;
-
-                       num_path_components = 
-                               get_num_path_components(dentry->full_path_utf8) - 1;
-                       num_output_dir_path_components =
-                               get_num_path_components(w->output_dir);
-
-                       if (w->is_multi_image_extraction) {
-                               num_path_components++;
-                               num_output_dir_path_components--;
-                       }
-                       file_on_disk_len = strlen(lte->file_on_disk);
-
-                       char buf[file_on_disk_len + 3 * num_path_components + 1];
-                       p = &buf[0];
-
-                       for (i = 0; i < num_path_components; i++) {
-                               *p++ = '.';
-                               *p++ = '.';
-                               *p++ = '/';
-                       }
-                       p2 = lte->file_on_disk;
-                       while (*p2 == '/')
-                               p2++;
-                       while (num_output_dir_path_components--)
-                               p2 = path_next_part(p2, NULL);
-                       strcpy(p, p2);
-                       if (symlink(buf, output_path) != 0) {
-                               ERROR_WITH_ERRNO("Failed to symlink `%s' to "
-                                                "`%s'",
-                                                buf, lte->file_on_disk);
-                               return WIMLIB_ERR_LINK;
+       int ret;
+       const struct list_head *head = &dentry->link_group_list;
+
+       if (head->next != head) {
+               /* This dentry is one of a hard link set of at least 2 dentries.
+                * If one of the other dentries has already been extracted, make
+                * a hard link to the file corresponding to this
+                * already-extracted directory.  Otherwise, extract the
+                * file, and set the dentry->extracted_file field so that other
+                * dentries in the hard link group can link to it. */
+               struct dentry *other;
+               list_for_each_entry(other, head, link_group_list) {
+                       if (other->extracted_file) {
+                               DEBUG("Extracting hard link `%s' => `%s'",
+                                     output_path, other->extracted_file);
+                               if (link(other->extracted_file, output_path) != 0) {
+                                       ERROR_WITH_ERRNO("Failed to hard link "
+                                                        "`%s' to `%s'",
+                                                        output_path,
+                                                        other->extracted_file);
+                                       return WIMLIB_ERR_LINK;
+                               }
+                               return 0;
                        }
-
                }
-               return 0;
-       } 
+               FREE(dentry->extracted_file);
+               dentry->extracted_file = STRDUP(output_path);
+               if (!dentry->extracted_file) {
+                       ERROR("Failed to allocate memory for filename");
+                       return WIMLIB_ERR_NOMEM;
+               }
+       }
 
-       /* Otherwise, we must actually extract the file contents. */
+       /* Extract the contents of the file to @output_path. */
 
        out_fd = open(output_path, O_WRONLY | O_CREAT | O_TRUNC, 0644);
        if (out_fd == -1) {
@@ -167,13 +163,14 @@ static int extract_regular_file(WIMStruct *w,
                return WIMLIB_ERR_OPEN;
        }
 
-       /* Extract empty file, with no lookup table entry... */
        if (!lte) {
+               /* Empty file with no lookup table entry */
                DEBUG("Empty file `%s'.", output_path);
                ret = 0;
                goto done;
        }
 
+
        res_entry = &lte->resource_entry;
 
        ret = extract_resource_to_fd(w, res_entry, out_fd, 
@@ -184,19 +181,71 @@ static int extract_regular_file(WIMStruct *w,
                goto done;
        }
 
-       /* Mark the lookup table entry to indicate this file has been extracted. */
-       lte->out_refcnt++;
-       FREE(lte->file_on_disk);
-       lte->file_on_disk = STRDUP(output_path);
-       if (lte->file_on_disk)
-               ret = 0;
-       else
-               ret = WIMLIB_ERR_NOMEM;
+       if (extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
+               /* Mark the lookup table entry to indicate this file has been
+                * extracted. */
+               lte->out_refcnt++;
+               FREE(lte->file_on_disk);
+               lte->file_on_disk = STRDUP(output_path);
+               if (!lte->file_on_disk)
+                       ret = WIMLIB_ERR_NOMEM;
+       }
 done:
-       close(out_fd);
+       if (close(out_fd) != 0) {
+               ERROR_WITH_ERRNO("Failed to close file `%s'", output_path);
+               ret = WIMLIB_ERR_WRITE;
+       }
        return ret;
 }
 
+/* 
+ * Extracts a regular file from the WIM archive. 
+ */
+static int extract_regular_file(WIMStruct *w, 
+                               struct dentry *dentry, 
+                               const char *output_dir,
+                               const char *output_path,
+                               int extract_flags)
+{
+       struct lookup_table_entry *lte;
+
+       lte = dentry_first_lte(dentry, w->lookup_table);
+
+       if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
+                             WIMLIB_EXTRACT_FLAG_HARDLINK)) && lte) {
+               if (lte->out_refcnt++ != 0)
+                       return extract_regular_file_linked(dentry, output_dir,
+                                                          output_path,
+                                                          extract_flags, lte);
+               lte->file_on_disk = STRDUP(output_path);
+               if (!lte->file_on_disk)
+                       return WIMLIB_ERR_NOMEM;
+       }
+
+       return extract_regular_file_unlinked(w, dentry, output_path,
+                                            extract_flags, lte);
+
+}
+
+static int extract_symlink(const struct dentry *dentry, const char *output_path,
+                          const WIMStruct *w)
+{
+       char target[4096];
+       ssize_t ret = dentry_readlink(dentry, target, sizeof(target), w);
+       if (ret <= 0) {
+               ERROR("Could not read the symbolic link from dentry `%s'",
+                     dentry->full_path_utf8);
+               return WIMLIB_ERR_INVALID_DENTRY;
+       }
+       ret = symlink(target, output_path);
+       if (ret != 0) {
+               ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
+                                output_path, target);
+               return WIMLIB_ERR_LINK;
+       }
+       return 0;
+}
+
 /* 
  * Extracts a directory from the WIM archive. 
  *
@@ -204,7 +253,7 @@ done:
  * @output_path:       The path to which the directory is to be extracted to.
  * @return:            True on success, false on failure. 
  */
-static int extract_directory(struct dentry *dentry, const char *output_path)
+static int extract_directory(const char *output_path)
 {
        /* Compute the output path directory to the directory. */
        if (mkdir(output_path, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH) != 0) 
@@ -227,45 +276,45 @@ static int extract_directory(struct dentry *dentry, const char *output_path)
 struct extract_args {
        WIMStruct *w;
        int extract_flags;
+       const char *output_dir;
 #ifdef WITH_NTFS_3G
        struct SECURITY_API *scapi;
 #endif
 };
 
 /* 
- * Extracts a file or directory from the WIM archive.  For use in
- * for_dentry_in_tree().
- *
- * @dentry:    The dentry to extract.
- * @arg:       A pointer to the WIMStruct for the WIM file.
+ * Extracts a file, directory, or symbolic link from the WIM archive.  For use
+ * in for_dentry_in_tree().
  */
-static int extract_regular_file_or_directory(struct dentry *dentry, void *arg)
+static int extract_dentry(struct dentry *dentry, void *arg)
 {
        struct extract_args *args = arg;
        WIMStruct *w = args->w;
        int extract_flags = args->extract_flags;
-       size_t len = strlen(w->output_dir);
+       size_t len = strlen(args->output_dir);
        char output_path[len + dentry->full_path_utf8_len + 1];
 
-       if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE)
+       if (extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) {
+               wimlib_assert(dentry->full_path_utf8);
                puts(dentry->full_path_utf8);
+       }
 
-       memcpy(output_path, w->output_dir, len);
+       memcpy(output_path, args->output_dir, len);
        memcpy(output_path + len, dentry->full_path_utf8, dentry->full_path_utf8_len);
        output_path[len + dentry->full_path_utf8_len] = '\0';
 
-       if (dentry_is_regular_file(dentry)) {
-               return extract_regular_file(w, dentry, output_path, extract_flags);
-       } else {
-               if (dentry_is_root(dentry)) /* Root doesn't need to be extracted. */
-                       return 0;
-               else
-                       return extract_directory(dentry, output_path);
-       }
+       if (dentry_is_symlink(dentry))
+               return extract_symlink(dentry, output_path, w);
+       else if (dentry_is_directory(dentry))
+               return extract_directory(output_path);
+       else
+               return extract_regular_file(w, dentry, args->output_dir,
+                                           output_path, extract_flags);
 }
 
 
-static int extract_single_image(WIMStruct *w, int image, int extract_flags)
+static int extract_single_image(WIMStruct *w, int image,
+                               const char *output_dir, int extract_flags)
 {
        DEBUG("Extracting image %d", image);
 
@@ -277,22 +326,23 @@ static int extract_single_image(WIMStruct *w, int image, int extract_flags)
        struct extract_args args = {
                .w = w,
                .extract_flags = extract_flags,
+               .output_dir = output_dir,
        #ifdef WITH_NTFS_3G
                .scapi = NULL
        #endif
        };
 
-       return for_dentry_in_tree(wim_root_dentry(w),
-                                 extract_regular_file_or_directory, &args);
+       return for_dentry_in_tree(wim_root_dentry(w), extract_dentry, &args);
 }
 
 
-/* Extracts all images from the WIM to w->output_dir, with the images placed in
+/* Extracts all images from the WIM to @output_dir, with the images placed in
  * subdirectories named by their image names. */
-static int extract_all_images(WIMStruct *w, int extract_flags)
+static int extract_all_images(WIMStruct *w, const char *output_dir,
+                             int extract_flags)
 {
        size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
-       size_t output_path_len = strlen(w->output_dir);
+       size_t output_path_len = strlen(output_dir);
        char buf[output_path_len + 1 + image_name_max_len + 1];
        int ret;
        int image;
@@ -300,7 +350,11 @@ static int extract_all_images(WIMStruct *w, int extract_flags)
 
        DEBUG("Attempting to extract all images from `%s'", w->filename);
 
-       memcpy(buf, w->output_dir, output_path_len);
+       ret = extract_directory(output_dir);
+       if (ret != 0)
+               return ret;
+
+       memcpy(buf, output_dir, output_path_len);
        buf[output_path_len] = '/';
        for (image = 1; image <= w->hdr.image_count; image++) {
                
@@ -311,10 +365,7 @@ static int extract_all_images(WIMStruct *w, int extract_flags)
                        /* Image name is empty. Use image number instead */
                        sprintf(buf + output_path_len + 1, "%d", image);
                }
-               ret = set_output_dir(w, buf);
-               if (ret != 0)
-                       goto done;
-               ret = extract_single_image(w, image, extract_flags);
+               ret = extract_single_image(w, image, buf, extract_flags);
                if (ret != 0)
                        goto done;
        }
@@ -328,21 +379,18 @@ done:
 WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
                                   const char *output_dir, int flags)
 {
-       int ret;
+
        if (!output_dir)
                return WIMLIB_ERR_INVALID_PARAM;
+
        if ((flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
                        == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
                return WIMLIB_ERR_INVALID_PARAM;
-       
-       ret = set_output_dir(w, output_dir);
-       if (ret != 0)
-               return ret;
 
        if ((flags & WIMLIB_EXTRACT_FLAG_NTFS)) {
        #ifdef WITH_NTFS_3G
                unsigned long mnt_flags;
-               ret = ntfs_check_if_mounted(output_dir, &mnt_flags);
+               int ret = ntfs_check_if_mounted(output_dir, &mnt_flags);
                if (ret != 0) {
                        ERROR_WITH_ERRNO("NTFS-3g: Cannot determine if `%s' "
                                         "is mounted", output_dir);
@@ -351,6 +399,7 @@ WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
                if (!(mnt_flags & NTFS_MF_MOUNTED)) {
                        ERROR("NTFS-3g: Filesystem on `%s' is not mounted ",
                              output_dir);
+                       return WIMLIB_ERR_NTFS_3G;
                }
                if (mnt_flags & NTFS_MF_READONLY) {
                        ERROR("NTFS-3g: Filesystem on `%s' is mounted "
@@ -364,13 +413,15 @@ WIMLIBAPI int wimlib_extract_image(WIMStruct *w, int image,
                return WIMLIB_ERR_UNSUPPORTED;
        #endif
        }
+
+       for_lookup_table_entry(w->lookup_table, zero_out_refcnts, NULL);
+
        if (image == WIM_ALL_IMAGES) {
-               w->is_multi_image_extraction = true;
-               ret = extract_all_images(w, flags);
+               flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
+               return extract_all_images(w, output_dir, flags);
        } else {
-               w->is_multi_image_extraction = false;
-               ret = extract_single_image(w, image, flags);
+               flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
+               return extract_single_image(w, image, output_dir, flags);
        }
-       return ret;
 
 }