]> wimlib.net Git - wimlib/blobdiff - src/extract_image.c
extract cleanups and fixes
[wimlib] / src / extract_image.c
index 17b12809f010a38fff6245371232f978a9c10c84..69a4b9e8d0fac4298f376197f69b88c3d4ecf40b 100644 (file)
 #  include <alloca.h>
 #endif
 
-#if TCHAR_IS_UTF16LE
-#  include <wchar.h>
-#endif
 
 #ifndef __WIN32__
+
+/* Returns the number of components of @path.  */
+static unsigned
+get_num_path_components(const char *path)
+{
+       unsigned num_components = 0;
+       while (*path) {
+               while (*path == '/')
+                       path++;
+               if (*path)
+                       num_components++;
+               while (*path && *path != '/')
+                       path++;
+       }
+       return num_components;
+}
+
+static const char *
+path_next_part(const char *path)
+{
+       while (*path && *path != '/')
+               path++;
+       while (*path && *path == '/')
+               path++;
+       return path;
+}
+
 static int
 extract_regular_file_linked(struct wim_dentry *dentry,
                            const char *output_path,
@@ -88,10 +112,8 @@ extract_regular_file_linked(struct wim_dentry *dentry,
                const char *p2;
                size_t i;
 
-               num_path_components =
-                       get_num_path_components(dentry->full_path) - 1;
-               num_output_dir_path_components =
-                       get_num_path_components(args->target);
+               num_path_components = get_num_path_components(dentry->_full_path) - 1;
+               num_output_dir_path_components = get_num_path_components(args->target);
 
                if (args->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
                        num_path_components++;
@@ -110,8 +132,10 @@ extract_regular_file_linked(struct wim_dentry *dentry,
                p2 = lte->extracted_file;
                while (*p2 == '/')
                        p2++;
-               while (num_output_dir_path_components--)
-                       p2 = path_next_part(p2, NULL);
+               while (num_output_dir_path_components > 0) {
+                       p2 = path_next_part(p2);
+                       num_output_dir_path_components--;
+               }
                strcpy(p, p2);
                if (symlink(buf, output_path) != 0) {
                        ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
@@ -123,15 +147,17 @@ extract_regular_file_linked(struct wim_dentry *dentry,
 }
 
 static int
-symlink_apply_unix_data(const mbchar *link,
+symlink_apply_unix_data(const char *link,
                        const struct wimlib_unix_data *unix_data)
 {
        if (lchown(link, unix_data->uid, unix_data->gid)) {
                if (errno == EPERM) {
                        /* Ignore */
-                       WARNING_WITH_ERRNO("failed to set symlink UNIX owner/group");
+                       WARNING_WITH_ERRNO("failed to set symlink UNIX "
+                                          "owner/group on \"%s\"", link);
                } else {
-                       ERROR_WITH_ERRNO("failed to set symlink UNIX owner/group");
+                       ERROR_WITH_ERRNO("failed to set symlink UNIX "
+                                        "owner/group on \"%s\"", link);
                        return WIMLIB_ERR_INVALID_DENTRY;
                }
        }
@@ -139,24 +165,29 @@ symlink_apply_unix_data(const mbchar *link,
 }
 
 static int
-fd_apply_unix_data(int fd, const struct wimlib_unix_data *unix_data)
+fd_apply_unix_data(int fd, const char *path,
+                  const struct wimlib_unix_data *unix_data)
 {
        if (fchown(fd, unix_data->uid, unix_data->gid)) {
                if (errno == EPERM) {
-                       WARNING_WITH_ERRNO("failed to set file UNIX owner/group");
+                       WARNING_WITH_ERRNO("failed to set file UNIX "
+                                          "owner/group on \"%s\"", path);
                        /* Ignore? */
                } else {
-                       ERROR_WITH_ERRNO("failed to set file UNIX owner/group");
+                       ERROR_WITH_ERRNO("failed to set file UNIX "
+                                        "owner/group on \"%s\"", path);
                        return WIMLIB_ERR_INVALID_DENTRY;
                }
        }
 
        if (fchmod(fd, unix_data->mode)) {
                if (errno == EPERM) {
-                       WARNING_WITH_ERRNO("failed to set UNIX file mode");
+                       WARNING_WITH_ERRNO("failed to set UNIX file mode "
+                                          "on \"%s\"", path);
                        /* Ignore? */
                } else {
-                       ERROR_WITH_ERRNO("failed to set UNIX file mode");
+                       ERROR_WITH_ERRNO("failed to set UNIX file mode "
+                                        "on \"%s\"", path);
                        return WIMLIB_ERR_INVALID_DENTRY;
                }
        }
@@ -169,7 +200,7 @@ dir_apply_unix_data(const char *dir, const struct wimlib_unix_data *unix_data)
        int dfd = open(dir, O_RDONLY);
        int ret;
        if (dfd >= 0) {
-               ret = fd_apply_unix_data(dfd, unix_data);
+               ret = fd_apply_unix_data(dfd, dir, unix_data);
                if (close(dfd)) {
                        ERROR_WITH_ERRNO("can't close directory `%s'", dir);
                        ret = WIMLIB_ERR_MKDIR;
@@ -243,7 +274,7 @@ extract_regular_file_unlinked(struct wim_dentry *dentry,
        }
 
        ret = extract_wim_resource_to_fd(lte, out_fd, wim_resource_size(lte));
-       if (ret != 0) {
+       if (ret) {
                ERROR("Failed to extract resource to `%s'", output_path);
                goto out;
        }
@@ -257,8 +288,8 @@ out_extract_unix_data:
                else if (ret < 0)
                        ret = 0;
                else
-                       ret = fd_apply_unix_data(out_fd, &unix_data);
-               if (ret != 0)
+                       ret = fd_apply_unix_data(out_fd, output_path, &unix_data);
+               if (ret)
                        goto out;
        }
        if (lte)
@@ -301,36 +332,53 @@ extract_symlink(struct wim_dentry *dentry,
                struct apply_args *args,
                const char *output_path)
 {
-       char target[4096];
-       ssize_t ret = inode_readlink(dentry->d_inode, target,
-                                    sizeof(target), args->w, 0);
+       char target[4096 + args->target_realpath_len];
+       char *fixed_target;
+       const struct wim_inode *inode = dentry->d_inode;
+
+       ssize_t ret = wim_inode_readlink(inode,
+                                        target + args->target_realpath_len,
+                                        sizeof(target) - args->target_realpath_len - 1);
        struct wim_lookup_table_entry *lte;
 
        if (ret <= 0) {
                ERROR("Could not read the symbolic link from dentry `%s'",
-                     dentry->full_path);
+                     dentry->_full_path);
                return WIMLIB_ERR_INVALID_DENTRY;
        }
-       ret = symlink(target, output_path);
-       if (ret != 0) {
+       target[args->target_realpath_len + ret] = '\0';
+       if (target[args->target_realpath_len] == '/' &&
+           args->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)
+       {
+               /* Fix absolute symbolic link target to point into the actual
+                * extraction destination */
+               memcpy(target, args->target_realpath,
+                      args->target_realpath_len);
+               fixed_target = target;
+       } else {
+               /* Keep same link target */
+               fixed_target = target + args->target_realpath_len;
+       }
+       ret = symlink(fixed_target, output_path);
+       if (ret) {
                ERROR_WITH_ERRNO("Failed to symlink `%s' to `%s'",
-                                output_path, target);
+                                output_path, fixed_target);
                return WIMLIB_ERR_LINK;
        }
-       lte = inode_unnamed_lte_resolved(dentry->d_inode);
-       wimlib_assert(lte != NULL);
        if (args->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
                struct wimlib_unix_data unix_data;
-               ret = inode_get_unix_data(dentry->d_inode, &unix_data, NULL);
+               ret = inode_get_unix_data(inode, &unix_data, NULL);
                if (ret > 0)
                        ;
                else if (ret < 0)
                        ret = 0;
                else
                        ret = symlink_apply_unix_data(output_path, &unix_data);
-               if (ret != 0)
+               if (ret)
                        return ret;
        }
+       lte = inode_unnamed_lte_resolved(inode);
+       wimlib_assert(lte != NULL);
        args->progress.extract.completed_bytes += wim_resource_size(lte);
        return 0;
 }
@@ -384,10 +432,9 @@ dir_exists:
 }
 
 #ifndef __WIN32__
-static int unix_do_apply_dentry(const char *output_path,
-                               size_t output_path_len,
-                               struct wim_dentry *dentry,
-                               struct apply_args *args)
+static int
+unix_do_apply_dentry(const char *output_path, size_t output_path_len,
+                    struct wim_dentry *dentry, struct apply_args *args)
 {
        const struct wim_inode *inode = dentry->d_inode;
 
@@ -404,7 +451,7 @@ static int unix_do_apply_dentry(const char *output_path,
 static int
 unix_do_apply_dentry_timestamps(const char *output_path,
                                size_t output_path_len,
-                               const struct wim_dentry *dentry,
+                               struct wim_dentry *dentry,
                                struct apply_args *args)
 {
        int ret;
@@ -462,29 +509,49 @@ unix_do_apply_dentry_timestamps(const char *output_path,
 }
 #endif /* !__WIN32__ */
 
+static int
+do_apply_op(struct wim_dentry *dentry, struct apply_args *args,
+           int (*apply_dentry_func)(const tchar *, size_t,
+                                    struct wim_dentry *, struct apply_args *))
+{
+       tchar *p;
+       const tchar *full_path;
+       size_t full_path_nchars;
+
+       wimlib_assert(dentry->_full_path != NULL);
+       full_path = dentry->_full_path + 1;
+       full_path_nchars = dentry->full_path_nbytes / sizeof(tchar) - 1;
+       tchar output_path[args->target_nchars + 1 +
+                        (full_path_nchars - args->wim_source_path_nchars) + 1];
+       p = output_path;
+
+       /*print_dentry(dentry, NULL);*/
+       /*ERROR("%"TS" %"TS, args->target, dentry->_full_path);*/
+       /*ERROR("");*/
+
+       tmemcpy(p, args->target, args->target_nchars);
+       p += args->target_nchars;
+
+       if (dentry != args->extract_root) {
+               *p++ = T('/');
+               tmemcpy(p, full_path + args->wim_source_path_nchars,
+                       full_path_nchars - args->wim_source_path_nchars);
+               p += full_path_nchars - args->wim_source_path_nchars;
+       }
+       *p = T('\0');
+       return (*apply_dentry_func)(output_path, p - output_path,
+                                   dentry, args);
+}
+
+
 /* Extracts a file, directory, or symbolic link from the WIM archive. */
 static int
 apply_dentry_normal(struct wim_dentry *dentry, void *arg)
 {
-       struct apply_args *args = arg;
-       tchar *output_path;
-       size_t len;
-
-       len = tstrlen(args->target);
-       if (dentry_is_root(dentry)) {
-               output_path = (tchar*)args->target;
-       } else {
-               output_path = alloca(len * sizeof(tchar) + dentry->full_path_nbytes +
-                                    sizeof(tchar));
-               memcpy(output_path, args->target, len * sizeof(tchar));
-               memcpy(output_path + len, dentry->full_path, dentry->full_path_nbytes);
-               len += dentry->full_path_nbytes / sizeof(tchar);
-               output_path[len] = T('\0');
-       }
 #ifdef __WIN32__
-       return win32_do_apply_dentry(output_path, len, dentry, args);
+       return do_apply_op(dentry, arg, win32_do_apply_dentry);
 #else
-       return unix_do_apply_dentry(output_path, len, dentry, args);
+       return do_apply_op(dentry, arg, unix_do_apply_dentry);
 #endif
 }
 
@@ -493,32 +560,29 @@ apply_dentry_normal(struct wim_dentry *dentry, void *arg)
 static int
 apply_dentry_timestamps_normal(struct wim_dentry *dentry, void *arg)
 {
-       struct apply_args *args = arg;
-       size_t len;
-       tchar *output_path;
-
-       len = tstrlen(args->target);
-       if (dentry_is_root(dentry)) {
-               output_path = (tchar*)args->target;
-       } else {
-               output_path = alloca(len * sizeof(tchar) + dentry->full_path_nbytes +
-                                    sizeof(tchar));
-               memcpy(output_path, args->target, len * sizeof(tchar));
-               memcpy(output_path + len, dentry->full_path, dentry->full_path_nbytes);
-               len += dentry->full_path_nbytes / sizeof(tchar);
-               output_path[len] = T('\0');
-       }
-
-
 #ifdef __WIN32__
-       return win32_do_apply_dentry_timestamps(output_path, len, dentry, args);
+       return do_apply_op(dentry, arg, win32_do_apply_dentry_timestamps);
 #else
-       return unix_do_apply_dentry_timestamps(output_path, len, dentry, args);
+       return do_apply_op(dentry, arg, unix_do_apply_dentry_timestamps);
 #endif
 }
 
-/* Extract a dentry if it hasn't already been extracted, and either the dentry
- * has no streams or WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified. */
+static bool
+dentry_is_descendent(const struct wim_dentry *dentry,
+                    const struct wim_dentry *ancestor)
+{
+       for (;;) {
+               if (dentry == ancestor)
+                       return true;
+               if (dentry_is_root(dentry))
+                       return false;
+               dentry = dentry->parent;
+       }
+}
+
+/* Extract a dentry if it hasn't already been extracted and either
+ * WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified, or the dentry is a directory
+ * and/or has no unnamed stream. */
 static int
 maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
 {
@@ -528,13 +592,17 @@ maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
        if (dentry->is_extracted)
                return 0;
 
-       if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS)
-               if (inode_unnamed_lte_resolved(dentry->d_inode))
-                       return 0;
+       if (!dentry_is_descendent(dentry, args->extract_root))
+               return 0;
+
+       if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS &&
+           !dentry_is_directory(dentry) &&
+           inode_unnamed_lte_resolved(dentry->d_inode) != NULL)
+               return 0;
 
        if ((args->extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) &&
             args->progress_func) {
-               args->progress.extract.cur_path = dentry->full_path;
+               args->progress.extract.cur_path = dentry->_full_path;
                args->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY,
                                    &args->progress);
        }
@@ -544,54 +612,6 @@ maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
        return ret;
 }
 
-static int
-cmp_streams_by_wim_position(const void *p1, const void *p2)
-{
-       const struct wim_lookup_table_entry *lte1, *lte2;
-       lte1 = *(const struct wim_lookup_table_entry**)p1;
-       lte2 = *(const struct wim_lookup_table_entry**)p2;
-       if (lte1->resource_entry.offset < lte2->resource_entry.offset)
-               return -1;
-       else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
-               return 1;
-       else
-               return 0;
-}
-
-static int
-sort_stream_list_by_wim_position(struct list_head *stream_list)
-{
-       struct list_head *cur;
-       size_t num_streams;
-       struct wim_lookup_table_entry **array;
-       size_t i;
-       size_t array_size;
-
-       num_streams = 0;
-       list_for_each(cur, stream_list)
-               num_streams++;
-       array_size = num_streams * sizeof(array[0]);
-       array = MALLOC(array_size);
-       if (!array) {
-               ERROR("Failed to allocate %zu bytes to sort stream entries",
-                     array_size);
-               return WIMLIB_ERR_NOMEM;
-       }
-       cur = stream_list->next;
-       for (i = 0; i < num_streams; i++) {
-               array[i] = container_of(cur, struct wim_lookup_table_entry, staging_list);
-               cur = cur->next;
-       }
-
-       qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
-
-       INIT_LIST_HEAD(stream_list);
-       for (i = 0; i < num_streams; i++)
-               list_add_tail(&array[i]->staging_list, stream_list);
-       FREE(array);
-       return 0;
-}
-
 static void
 calculate_bytes_to_extract(struct list_head *stream_list,
                           int extract_flags,
@@ -602,7 +622,7 @@ calculate_bytes_to_extract(struct list_head *stream_list,
        u64 num_streams = 0;
 
        /* For each stream to be extracted... */
-       list_for_each_entry(lte, stream_list, staging_list) {
+       list_for_each_entry(lte, stream_list, extraction_list) {
                if (extract_flags &
                    (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
                {
@@ -630,7 +650,7 @@ maybe_add_stream_for_extraction(struct wim_lookup_table_entry *lte,
 {
        if (++lte->out_refcnt == 1) {
                INIT_LIST_HEAD(&lte->inode_list);
-               list_add_tail(&lte->staging_list, stream_list);
+               list_add_tail(&lte->extraction_list, stream_list);
        }
 }
 
@@ -648,8 +668,21 @@ inode_find_streams_for_extraction(struct wim_inode *inode,
                list_add_tail(&inode->i_lte_inode_list, &lte->inode_list);
                inode_added = true;
        }
-#ifdef WITH_NTFS_3G
-       if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
+
+       /* Determine whether to include alternate data stream entries or not.
+        *
+        * UNIX:  Include them if extracting using NTFS-3g.
+        *
+        * Windows: Include them undconditionally, although if the filesystem is
+        * not NTFS we won't actually be able to extract them. */
+#if defined(WITH_NTFS_3G)
+       if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
+#elif defined(__WIN32__)
+       if (1)
+#else
+       if (0)
+#endif
+       {
                for (unsigned i = 0; i < inode->i_num_ads; i++) {
                        if (inode->i_ads_entries[i].stream_name_nbytes != 0) {
                                lte = inode->i_ads_entries[i].lte;
@@ -665,29 +698,64 @@ inode_find_streams_for_extraction(struct wim_inode *inode,
                        }
                }
        }
-#endif
+}
+
+struct find_streams_ctx {
+       struct list_head stream_list;
+       int extract_flags;
+};
+
+static int
+dentry_find_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
+{
+       struct find_streams_ctx *ctx = _ctx;
+       struct wim_inode *inode = dentry->d_inode;
+
+       dentry->is_extracted = 0;
+       if (!inode->i_visited) {
+               inode_find_streams_for_extraction(inode, &ctx->stream_list,
+                                                 ctx->extract_flags);
+               inode->i_visited = 1;
+       }
+       return 0;
+}
+
+static int
+dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_lookup_table)
+{
+       struct wim_inode *inode = dentry->d_inode;
+       struct wim_lookup_table *lookup_table = _lookup_table;
+       struct wim_lookup_table_entry *lte;
+
+       inode_resolve_ltes(inode, lookup_table);
+       for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+               lte = inode_stream_lte_resolved(inode, i);
+               if (lte)
+                       lte->out_refcnt = 0;
+       }
+       return 0;
 }
 
 static void
-find_streams_for_extraction(struct hlist_head *inode_list,
+find_streams_for_extraction(struct wim_dentry *root,
                            struct list_head *stream_list,
                            struct wim_lookup_table *lookup_table,
                            int extract_flags)
 {
-       struct wim_inode *inode;
-       struct hlist_node *cur;
-       struct wim_dentry *dentry;
+       struct find_streams_ctx ctx;
 
-       for_lookup_table_entry(lookup_table, lte_zero_out_refcnt, NULL);
-       INIT_LIST_HEAD(stream_list);
-       hlist_for_each_entry(inode, cur, inode_list, i_hlist) {
-               if (!inode->i_resolved)
-                       inode_resolve_ltes(inode, lookup_table);
-               inode_for_each_dentry(dentry, inode)
-                       dentry->is_extracted = 0;
-               inode_find_streams_for_extraction(inode, stream_list,
-                                                 extract_flags);
-       }
+       INIT_LIST_HEAD(&ctx.stream_list);
+       ctx.extract_flags = extract_flags;
+       for_dentry_in_tree(root, dentry_resolve_and_zero_lte_refcnt, lookup_table);
+       for_dentry_in_tree(root, dentry_find_streams_to_extract, &ctx);
+       list_transfer(&ctx.stream_list, stream_list);
+}
+
+static int
+dentry_mark_inode_unvisited(struct wim_dentry *dentry, void *_ignore)
+{
+       dentry->d_inode->i_visited = 0;
+       return 0;
 }
 
 struct apply_operations {
@@ -729,7 +797,7 @@ apply_stream_list(struct list_head *stream_list,
         * sequential reading of the WIM can be implemented. */
 
        /* For each distinct stream to be extracted */
-       list_for_each_entry(lte, stream_list, staging_list) {
+       list_for_each_entry(lte, stream_list, extraction_list) {
                /* For each inode that contains the stream */
                list_for_each_entry(inode, &lte->inode_list, i_lte_inode_list) {
                        /* For each dentry that points to the inode */
@@ -737,7 +805,7 @@ apply_stream_list(struct list_head *stream_list,
                                /* Extract the dentry if it was not already
                                 * extracted */
                                ret = maybe_apply_dentry(dentry, args);
-                               if (ret != 0)
+                               if (ret)
                                        return ret;
                                if (progress_func &&
                                    args->progress.extract.completed_bytes >= next_progress)
@@ -761,33 +829,102 @@ apply_stream_list(struct list_head *stream_list,
        return 0;
 }
 
-/* Extracts the image @image from the WIM @w to the directory or NTFS volume
- * @target. */
 static int
-extract_single_image(WIMStruct *w, int image,
-                    const tchar *target, int extract_flags,
-                    wimlib_progress_func_t progress_func)
+sort_stream_list_by_wim_position(struct list_head *stream_list)
+{
+       struct list_head *cur;
+       size_t num_streams;
+       struct wim_lookup_table_entry **array;
+       size_t i;
+       size_t array_size;
+
+       num_streams = 0;
+       list_for_each(cur, stream_list)
+               num_streams++;
+       array_size = num_streams * sizeof(array[0]);
+       array = MALLOC(array_size);
+       if (!array) {
+               ERROR("Failed to allocate %zu bytes to sort stream entries",
+                     array_size);
+               return WIMLIB_ERR_NOMEM;
+       }
+       cur = stream_list->next;
+       for (i = 0; i < num_streams; i++) {
+               array[i] = container_of(cur, struct wim_lookup_table_entry, extraction_list);
+               cur = cur->next;
+       }
+
+       qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
+
+       INIT_LIST_HEAD(stream_list);
+       for (i = 0; i < num_streams; i++)
+               list_add_tail(&array[i]->extraction_list, stream_list);
+       FREE(array);
+       return 0;
+}
+
+/*
+ * extract_tree - Extract a file or directory tree from the currently selected
+ *               WIM image.
+ *
+ * @wim:       WIMStruct for the WIM file, with the desired image selected
+ *             (as wim->current_image).
+ * @wim_source_path:
+ *             "Canonical" (i.e. no leading or trailing slashes, path
+ *             separators forwald slashes) path inside the WIM image to
+ *             extract.  An empty string means the full image.
+ * @target:
+ *             Filesystem path to extract the file or directory tree to.
+ *
+ * @extract_flags:
+ *             WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
+ *             WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
+ *             called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
+ *             the image.
+ *
+ * @progress_func:
+ *             If non-NULL, progress function for the extraction.  The messages
+ *             we may in this function are:
+ *
+ *             WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
+ *                     WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
+ *             WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
+ *             WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
+ *             WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
+ *             WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
+ *             WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
+ *             WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
+ *                     WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
+ *
+ * Returns 0 on success; nonzero on failure.
+ */
+static int
+extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
+            int extract_flags, wimlib_progress_func_t progress_func)
 {
        int ret;
        struct list_head stream_list;
-       struct hlist_head *inode_list;
-
        struct apply_args args;
        const struct apply_operations *ops;
+       struct wim_dentry *root;
+
+       memset(&args, 0, sizeof(args));
 
-       args.w                  = w;
-       args.target             = target;
-       args.extract_flags      = extract_flags;
-       args.num_utime_warnings = 0;
-       args.stream_list        = &stream_list;
-       args.progress_func      = progress_func;
+       args.w                      = wim;
+       args.target                 = target;
+       args.extract_flags          = extract_flags;
+       args.progress_func          = progress_func;
+       args.target_nchars          = tstrlen(target);
+       args.wim_source_path_nchars = tstrlen(wim_source_path);
 
        if (progress_func) {
-               args.progress.extract.wimfile_name = w->filename;
-               args.progress.extract.image = image;
+               args.progress.extract.wimfile_name = wim->filename;
+               args.progress.extract.image = wim->current_image;
                args.progress.extract.extract_flags = (extract_flags &
                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
-               args.progress.extract.image_name = wimlib_get_image_name(w, image);
+               args.progress.extract.image_name = wimlib_get_image_name(wim,
+                                                                        wim->current_image);
+               args.progress.extract.extract_root_wim_source_path = wim_source_path;
                args.progress.extract.target = target;
        }
 
@@ -797,29 +934,35 @@ extract_single_image(WIMStruct *w, int image,
                if (!args.vol) {
                        ERROR_WITH_ERRNO("Failed to mount NTFS volume `%"TS"'",
                                         target);
-                       return WIMLIB_ERR_NTFS_3G;
+                       ret = WIMLIB_ERR_NTFS_3G;
+                       goto out;
                }
                ops = &ntfs_apply_operations;
        } else
 #endif
                ops = &normal_apply_operations;
 
-       ret = select_wim_image(w, image);
-       if (ret != 0)
-               goto out;
-
-       inode_list = &w->image_metadata[image - 1].inode_list;
+       root = get_dentry(wim, wim_source_path);
+       if (!root) {
+               ERROR("Path \"%"TS"\" does not exist in WIM image %d",
+                     wim_source_path, wim->current_image);
+               ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
+               goto out_ntfs_umount;
+       }
+       args.extract_root = root;
 
        /* Build a list of the streams that need to be extracted */
-       find_streams_for_extraction(inode_list, &stream_list,
-                                   w->lookup_table, extract_flags);
+       find_streams_for_extraction(root,
+                                   &stream_list,
+                                   wim->lookup_table, extract_flags);
 
        /* Calculate the number of bytes of data that will be extracted */
        calculate_bytes_to_extract(&stream_list, extract_flags,
                                   &args.progress);
 
        if (progress_func) {
-               progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
+               progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
+                             WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
                              &args.progress);
        }
 
@@ -839,23 +982,36 @@ extract_single_image(WIMStruct *w, int image,
                              &args.progress);
        }
 
+       ret = calculate_dentry_tree_full_paths(root);
+       if (ret)
+               goto out_mark_inodes_unvisited;
+
        /* Make the directory structure and extract empty files */
        args.extract_flags |= WIMLIB_EXTRACT_FLAG_NO_STREAMS;
        args.apply_dentry = ops->apply_dentry;
-       ret = for_dentry_in_tree(wim_root_dentry(w), maybe_apply_dentry, &args);
+       ret = for_dentry_in_tree(root, maybe_apply_dentry, &args);
        args.extract_flags &= ~WIMLIB_EXTRACT_FLAG_NO_STREAMS;
-       if (ret != 0)
-               goto out;
+       if (ret)
+               goto out_mark_inodes_unvisited;
 
        if (progress_func) {
                progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
                              &args.progress);
        }
 
+       if (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) {
+               args.target_realpath = realpath(target, NULL);
+               if (!args.target_realpath) {
+                       ret = WIMLIB_ERR_NOMEM;
+                       goto out_mark_inodes_unvisited;
+               }
+               args.target_realpath_len = tstrlen(args.target_realpath);
+       }
+
        /* Extract non-empty files */
        ret = apply_stream_list(&stream_list, &args, ops, progress_func);
-       if (ret != 0)
-               goto out;
+       if (ret)
+               goto out_free_target_realpath;
 
        if (progress_func) {
                progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
@@ -863,16 +1019,21 @@ extract_single_image(WIMStruct *w, int image,
        }
 
        /* Apply timestamps */
-       ret = for_dentry_in_tree_depth(wim_root_dentry(w),
+       ret = for_dentry_in_tree_depth(root,
                                       ops->apply_dentry_timestamps, &args);
-       if (ret != 0)
-               goto out;
+       if (ret)
+               goto out_free_target_realpath;
 
        if (progress_func) {
-               progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
+               progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
+                             WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
                              &args.progress);
        }
-out:
+out_free_target_realpath:
+       FREE(args.target_realpath);
+out_mark_inodes_unvisited:
+       for_dentry_in_tree(root, dentry_mark_inode_unvisited, NULL);
+out_ntfs_umount:
 #ifdef WITH_NTFS_3G
        /* Unmount the NTFS volume */
        if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
@@ -884,33 +1045,300 @@ out:
                }
        }
 #endif
+out:
+       return ret;
+}
+
+/* Validates a single wimlib_extract_command, mostly checking to make sure the
+ * extract flags make sense. */
+static int
+check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
+{
+       int extract_flags;
+       bool is_entire_image = (cmd->wim_source_path[0] == T('\0'));
+
+       /* Empty destination path? */
+       if (cmd->fs_dest_path[0] == T('\0'))
+               return WIMLIB_ERR_INVALID_PARAM;
+
+       extract_flags = cmd->extract_flags;
+
+       /* Specified both symlink and hardlink modes? */
+       if ((extract_flags &
+            (WIMLIB_EXTRACT_FLAG_SYMLINK |
+             WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
+                                                WIMLIB_EXTRACT_FLAG_HARDLINK))
+               return WIMLIB_ERR_INVALID_PARAM;
+
+#ifdef __WIN32__
+       /* Wanted UNIX data on Win32? */
+       if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
+               ERROR("Extracting UNIX data is not supported on Windows");
+               return WIMLIB_ERR_INVALID_PARAM;
+       }
+       /* Wanted linked extraction on Windows?  (XXX This is possible, just not
+        * implemented yet.) */
+       if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
+                            WIMLIB_EXTRACT_FLAG_HARDLINK))
+       {
+               ERROR("Linked extraction modes are not supported on Windows");
+               return WIMLIB_ERR_INVALID_PARAM;
+       }
+#endif
+
+       if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
+               /* NTFS-3g extraction mode requested */
+#ifdef WITH_NTFS_3G
+               if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
+                                     WIMLIB_EXTRACT_FLAG_HARDLINK))) {
+                       ERROR("Cannot specify symlink or hardlink flags when applying\n"
+                             "        directly to a NTFS volume");
+                       return WIMLIB_ERR_INVALID_PARAM;
+               }
+               if (!is_entire_image &&
+                   (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS))
+               {
+                       ERROR("When applying directly to a NTFS volume you can "
+                             "only extract a full image, not part of one");
+                       return WIMLIB_ERR_INVALID_PARAM;
+               }
+               if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
+                       ERROR("Cannot restore UNIX-specific data in "
+                             "the NTFS extraction mode");
+                       return WIMLIB_ERR_INVALID_PARAM;
+               }
+#else
+               ERROR("wimlib was compiled without support for NTFS-3g, so");
+               ERROR("we cannot apply a WIM image directly to a NTFS volume");
+               return WIMLIB_ERR_UNSUPPORTED;
+#endif
+       }
+
+       if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
+                             WIMLIB_EXTRACT_FLAG_NORPFIX)) ==
+               (WIMLIB_EXTRACT_FLAG_RPFIX | WIMLIB_EXTRACT_FLAG_NORPFIX))
+       {
+               ERROR("Cannot specify RPFIX and NORPFIX flags at the same time!");
+               return WIMLIB_ERR_INVALID_PARAM;
+       }
+
+       if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
+                             WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
+       {
+               /* Do reparse point fixups by default if the WIM header says
+                * they are enabled and we are extracting a full image. */
+               if ((wim_header_flags & WIM_HDR_FLAG_RP_FIX) && is_entire_image)
+                       extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
+       }
+
+       if (!is_entire_image && (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)) {
+               ERROR("Cannot specify --rpfix when not extracting entire image");
+               return WIMLIB_ERR_INVALID_PARAM;
+       }
+
+       cmd->extract_flags = extract_flags;
+       return 0;
+}
+
+
+/* Internal function to execute extraction commands for a WIM image. */
+static int
+do_wimlib_extract_files(WIMStruct *wim,
+                       int image,
+                       struct wimlib_extract_command *cmds,
+                       size_t num_cmds,
+                       wimlib_progress_func_t progress_func)
+{
+       int ret;
+       bool found_link_cmd = false;
+       bool found_nolink_cmd = false;
+
+       /* Select the image from which we are extracting files */
+       ret = select_wim_image(wim, image);
+       if (ret)
+               return ret;
+
+       /* Make sure there are no streams in the WIM that have not been
+        * checksummed yet. */
+       ret = wim_checksum_unhashed_streams(wim);
+       if (ret)
+               return ret;
+
+       /* Check for problems with the extraction commands */
+       for (size_t i = 0; i < num_cmds; i++) {
+               ret = check_extract_command(&cmds[i], wim->hdr.flags);
+               if (ret)
+                       return ret;
+               if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
+                                            WIMLIB_EXTRACT_FLAG_HARDLINK)) {
+                       found_link_cmd = true;
+               } else {
+                       found_nolink_cmd = true;
+               }
+               if (found_link_cmd && found_nolink_cmd) {
+                       ERROR("Symlink or hardlink extraction mode must "
+                             "be set on all extraction commands");
+                       return WIMLIB_ERR_INVALID_PARAM;
+               }
+       }
+
+       /* Execute the extraction commands */
+       for (size_t i = 0; i < num_cmds; i++) {
+               ret = extract_tree(wim,
+                                  cmds[i].wim_source_path,
+                                  cmds[i].fs_dest_path,
+                                  cmds[i].extract_flags,
+                                  progress_func);
+               if (ret)
+                       return ret;
+       }
+       return 0;
+}
+
+/* Extract files or directories from a WIM image. */
+WIMLIBAPI int
+wimlib_extract_files(WIMStruct *wim,
+                    int image,
+                    int default_extract_flags,
+                    const struct wimlib_extract_command *cmds,
+                    size_t num_cmds,
+                    WIMStruct **additional_swms,
+                    unsigned num_additional_swms,
+                    wimlib_progress_func_t progress_func)
+{
+       int ret;
+       struct wimlib_extract_command *cmds_copy;
+       struct wim_lookup_table *wim_tab_save, *joined_tab;
+       int all_flags = 0;
+
+       default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
+
+       ret = verify_swm_set(wim, additional_swms, num_additional_swms);
+       if (ret)
+               goto out;
+
+       if (num_additional_swms) {
+               ret = new_joined_lookup_table(wim, additional_swms,
+                                             num_additional_swms,
+                                             &joined_tab);
+               if (ret)
+                       goto out;
+               wim_tab_save = wim->lookup_table;
+               wim->lookup_table = joined_tab;
+       }
+
+       cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
+       if (!cmds_copy) {
+               ret = WIMLIB_ERR_NOMEM;
+               goto out_restore_lookup_table;
+       }
+
+       for (size_t i = 0; i < num_cmds; i++) {
+               cmds_copy[i].extract_flags = (default_extract_flags |
+                                                cmds[i].extract_flags)
+                                               & WIMLIB_EXTRACT_MASK_PUBLIC;
+               all_flags |= cmds_copy[i].extract_flags;
+
+               cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
+               if (!cmds_copy[i].wim_source_path) {
+                       ret = WIMLIB_ERR_NOMEM;
+                       goto out_free_cmds_copy;
+               }
+
+               cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
+               if (!cmds_copy[i].fs_dest_path) {
+                       ret = WIMLIB_ERR_NOMEM;
+                       goto out_free_cmds_copy;
+               }
+
+       }
+       ret = do_wimlib_extract_files(wim, image,
+                                     cmds_copy, num_cmds,
+                                     progress_func);
+
+       if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
+                        WIMLIB_EXTRACT_FLAG_HARDLINK))
+       {
+               for_lookup_table_entry(wim->lookup_table,
+                                      lte_free_extracted_file, NULL);
+       }
+out_free_cmds_copy:
+       for (size_t i = 0; i < num_cmds; i++) {
+               FREE(cmds_copy[i].wim_source_path);
+               FREE(cmds_copy[i].fs_dest_path);
+       }
+       FREE(cmds_copy);
+out_restore_lookup_table:
+       if (num_additional_swms) {
+               free_lookup_table(wim->lookup_table);
+               wim->lookup_table = wim_tab_save;
+       }
+out:
        return ret;
 }
 
+/*
+ * Extracts an image from a WIM file.
+ *
+ * @wim:               WIMStruct for the WIM file.
+ *
+ * @image:             Number of the single image to extract.
+ *
+ * @target:            Directory or NTFS volume to extract the image to.
+ *
+ * @extract_flags:     Bitwise or of WIMLIB_EXTRACT_FLAG_*.
+ *
+ * @progress_func:     If non-NULL, a progress function to be called
+ *                     periodically.
+ *
+ * Returns 0 on success; nonzero on failure.
+ */
+static int
+extract_single_image(WIMStruct *wim, int image,
+                    const tchar *target, int extract_flags,
+                    wimlib_progress_func_t progress_func)
+{
+       int ret;
+       tchar *target_copy = canonicalize_fs_path(target);
+       if (!target_copy)
+               return WIMLIB_ERR_NOMEM;
+       struct wimlib_extract_command cmd = {
+               .wim_source_path = T(""),
+               .fs_dest_path = target_copy,
+               .extract_flags = extract_flags,
+       };
+       ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
+       FREE(target_copy);
+       return ret;
+}
 
+static const tchar * const filename_forbidden_chars =
+T(
+#ifdef __WIN32__
+"<>:\"/\\|?*"
+#else
+"/"
+#endif
+);
+
+/* This function checks if it is okay to use a WIM image's name as a directory
+ * name.  */
 static bool
 image_name_ok_as_dir(const tchar *image_name)
 {
-       if (image_name == NULL)
-               return false;
-       if (image_name[0] == T('\0'))
-               return false;
-       if (tstrchr(image_name, T('/')))
-               return false;
-       if (tstrchr(image_name, T('\\')))
-               return false;
-       return true;
+       return image_name && *image_name &&
+               !tstrpbrk(image_name, filename_forbidden_chars);
 }
 
 /* Extracts all images from the WIM to the directory @target, with the images
  * placed in subdirectories named by their image names. */
 static int
-extract_all_images(WIMStruct *w,
+extract_all_images(WIMStruct *wim,
                   const tchar *target,
                   int extract_flags,
                   wimlib_progress_func_t progress_func)
 {
-       size_t image_name_max_len = max(xml_get_max_image_name_len(w), 20);
+       size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
        size_t output_path_len = tstrlen(target);
        tchar buf[output_path_len + 1 + image_name_max_len + 1];
        int ret;
@@ -921,21 +1349,20 @@ extract_all_images(WIMStruct *w,
        if (ret)
                return ret;
 
-       wmemcpy(buf, target, output_path_len);
+       tmemcpy(buf, target, output_path_len);
        buf[output_path_len] = T('/');
-       for (image = 1; image <= w->hdr.image_count; image++) {
-               image_name = wimlib_get_image_name(w, image);
+       for (image = 1; image <= wim->hdr.image_count; image++) {
+               image_name = wimlib_get_image_name(wim, image);
                if (image_name_ok_as_dir(image_name)) {
                        tstrcpy(buf + output_path_len + 1, image_name);
                } else {
-                       /* Image name is empty, or may not be representable in
-                        * the current locale, or contains path separators.  Use
-                        * the image number instead. */
+                       /* Image name is empty or contains forbidden characters.
+                        * Use image number instead. */
                        tsprintf(buf + output_path_len + 1, T("%d"), image);
                }
-               ret = extract_single_image(w, image, buf, extract_flags,
+               ret = extract_single_image(wim, image, buf, extract_flags,
                                           progress_func);
-               if (ret != 0)
+               if (ret)
                        return ret;
        }
        return 0;
@@ -944,7 +1371,7 @@ extract_all_images(WIMStruct *w,
 /* Extracts a single image or all images from a WIM file to a directory or NTFS
  * volume. */
 WIMLIBAPI int
-wimlib_extract_image(WIMStruct *w,
+wimlib_extract_image(WIMStruct *wim,
                     int image,
                     const tchar *target,
                     int extract_flags,
@@ -952,86 +1379,44 @@ wimlib_extract_image(WIMStruct *w,
                     unsigned num_additional_swms,
                     wimlib_progress_func_t progress_func)
 {
-       struct wim_lookup_table *joined_tab, *w_tab_save;
+       struct wim_lookup_table *joined_tab, *wim_tab_save;
        int ret;
 
-       if (!target)
-               return WIMLIB_ERR_INVALID_PARAM;
-
        extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
 
-       if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
-                       == (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
-               return WIMLIB_ERR_INVALID_PARAM;
-
-#ifdef __WIN32__
-       if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
-               ERROR("Extracting UNIX data is not supported on Windows");
-               return WIMLIB_ERR_INVALID_PARAM;
-       }
-       if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK)) {
-               ERROR("Linked extraction modes are not supported on Windows");
-               return WIMLIB_ERR_INVALID_PARAM;
-       }
-#endif
-
-       if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
-#ifdef WITH_NTFS_3G
-               if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))) {
-                       ERROR("Cannot specify symlink or hardlink flags when applying\n"
-                             "        directly to a NTFS volume");
-                       return WIMLIB_ERR_INVALID_PARAM;
-               }
-               if (image == WIMLIB_ALL_IMAGES) {
-                       ERROR("Can only apply a single image when applying "
-                             "directly to a NTFS volume");
-                       return WIMLIB_ERR_INVALID_PARAM;
-               }
-               if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
-                       ERROR("Cannot restore UNIX-specific data in the NTFS extraction mode");
-                       return WIMLIB_ERR_INVALID_PARAM;
-               }
-#else
-               ERROR("wimlib was compiled without support for NTFS-3g, so");
-               ERROR("we cannot apply a WIM image directly to a NTFS volume");
-               return WIMLIB_ERR_UNSUPPORTED;
-#endif
-       }
-
-       ret = verify_swm_set(w, additional_swms, num_additional_swms);
-       if (ret != 0)
+       ret = verify_swm_set(wim, additional_swms, num_additional_swms);
+       if (ret)
                return ret;
 
        if (num_additional_swms) {
-               ret = new_joined_lookup_table(w, additional_swms,
+               ret = new_joined_lookup_table(wim, additional_swms,
                                              num_additional_swms, &joined_tab);
-               if (ret != 0)
+               if (ret)
                        return ret;
-               w_tab_save = w->lookup_table;
-               w->lookup_table = joined_tab;
+               wim_tab_save = wim->lookup_table;
+               wim->lookup_table = joined_tab;
        }
 
        if (image == WIMLIB_ALL_IMAGES) {
                extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
-               ret = extract_all_images(w, target, extract_flags,
+               ret = extract_all_images(wim, target, extract_flags,
                                         progress_func);
        } else {
                extract_flags &= ~WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
-               ret = extract_single_image(w, image, target, extract_flags,
+               ret = extract_single_image(wim, image, target, extract_flags,
                                           progress_func);
        }
 
        if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
                             WIMLIB_EXTRACT_FLAG_HARDLINK))
        {
-               for_lookup_table_entry(w->lookup_table,
+               for_lookup_table_entry(wim->lookup_table,
                                       lte_free_extracted_file,
                                       NULL);
        }
-
        if (num_additional_swms) {
-               free_lookup_table(w->lookup_table);
-               w->lookup_table = w_tab_save;
+               free_lookup_table(wim->lookup_table);
+               wim->lookup_table = wim_tab_save;
        }
        return ret;
 }