From: Eric Biggers Date: Sun, 5 May 2013 17:19:26 +0000 (-0500) Subject: imagex-extract initial implementation X-Git-Tag: v1.4.0~117 X-Git-Url: https://wimlib.net/git/?p=wimlib;a=commitdiff_plain;h=eaf5b4c85a3b7b853317b887867c18a4865a83e2 imagex-extract initial implementation --- diff --git a/programs/imagex.c b/programs/imagex.c index 0bba3546..275f48d1 100644 --- a/programs/imagex.c +++ b/programs/imagex.c @@ -68,6 +68,7 @@ enum imagex_op_type { DELETE, DIR, EXPORT, + EXTRACT, INFO, JOIN, MOUNT, @@ -122,6 +123,12 @@ IMAGEX_PROGNAME" export SRC_WIMFILE (SRC_IMAGE_NUM | SRC_IMAGE_NAME | all ) \n" " [--boot] [--check] [--compress=TYPE] [--ref=\"GLOB\"]\n" " [--threads=NUM_THREADS] [--rebuild]\n" ), +[EXTRACT] = +T( +IMAGEX_PROGNAME" extract SRC_WIMFILE (SRC_IMAGE_NUM | SRC_IMAGE_NAME) [PATH...]\n" +" [--check] [--ref=\"GLOB\"] [--verbose] [--unix-data] [--no-acls]\n" +" [--strict-acls] [--to-stdout] [--dest-dir=DIR]\n" +), [INFO] = T( IMAGEX_PROGNAME" info WIMFILE [IMAGE_NUM | IMAGE_NAME] [NEW_NAME]\n" @@ -168,14 +175,15 @@ enum { IMAGEX_CONFIG_OPTION, IMAGEX_DEBUG_OPTION, IMAGEX_DEREFERENCE_OPTION, + IMAGEX_DEST_DIR_OPTION, IMAGEX_EXTRACT_XML_OPTION, IMAGEX_FLAGS_OPTION, IMAGEX_HARDLINK_OPTION, IMAGEX_HEADER_OPTION, IMAGEX_LOOKUP_TABLE_OPTION, IMAGEX_METADATA_OPTION, - IMAGEX_NO_ACLS_OPTION, IMAGEX_NORPFIX_OPTION, + IMAGEX_NO_ACLS_OPTION, IMAGEX_REBULID_OPTION, IMAGEX_RECOMPRESS_OPTION, IMAGEX_REF_OPTION, @@ -187,6 +195,7 @@ enum { IMAGEX_STRICT_ACLS_OPTION, IMAGEX_SYMLINK_OPTION, IMAGEX_THREADS_OPTION, + IMAGEX_TO_STDOUT_OPTION, IMAGEX_UNIX_DATA_OPTION, IMAGEX_VERBOSE_OPTION, IMAGEX_XML_OPTION, @@ -241,6 +250,19 @@ static const struct option export_options[] = { {NULL, 0, NULL, 0}, }; +static const struct option extract_options[] = { + {T("check"), no_argument, NULL, IMAGEX_CHECK_OPTION}, + {T("verbose"), no_argument, NULL, IMAGEX_VERBOSE_OPTION}, + {T("ref"), required_argument, NULL, IMAGEX_REF_OPTION}, + {T("unix-data"), no_argument, NULL, IMAGEX_UNIX_DATA_OPTION}, + {T("noacls"), no_argument, NULL, IMAGEX_NO_ACLS_OPTION}, + {T("no-acls"), no_argument, NULL, IMAGEX_NO_ACLS_OPTION}, + {T("strict-acls"), no_argument, NULL, IMAGEX_STRICT_ACLS_OPTION}, + {T("dest-dir"), required_argument, NULL, IMAGEX_DEST_DIR_OPTION}, + {T("to-stdout"), no_argument, NULL, IMAGEX_TO_STDOUT_OPTION}, + {NULL, 0, NULL, 0}, +}; + static const struct option info_options[] = { {T("boot"), no_argument, NULL, IMAGEX_BOOT_OPTION}, {T("check"), no_argument, NULL, IMAGEX_CHECK_OPTION}, @@ -991,7 +1013,7 @@ imagex_progress_func(enum wimlib_progress_msg msg, tputchar(T('\n')); break; case WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN: - tprintf(T("Applying image %d (%"TS") from \"%"TS"\" " + tprintf(T("Applying image %d (\"%"TS"\") from \"%"TS"\" " "to %"TS" \"%"TS"\"\n"), info->extract.image, info->extract.image_name, @@ -1000,6 +1022,15 @@ imagex_progress_func(enum wimlib_progress_msg msg, T("NTFS volume") : T("directory")), info->extract.target); break; + case WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN: + tprintf(T("Extracting \"%"TS"\" from image %d (\"%"TS"\") " + "in \"%"TS"\" to \"%"TS"\"\n"), + info->extract.extract_root_wim_source_path, + info->extract.image, + info->extract.image_name, + info->extract.wimfile_name, + info->extract.target); + break; /*case WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN:*/ /*tprintf(T("Applying directory structure to %"TS"\n"),*/ /*info->extract.target);*/ @@ -1019,7 +1050,8 @@ imagex_progress_func(enum wimlib_progress_msg msg, tprintf(T("%"TS"\n"), info->extract.cur_path); break; case WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS: - tprintf(T("Setting timestamps on all extracted files...\n")); + if (info->extract.extract_root_wim_source_path[0] == T('\0')) + tprintf(T("Setting timestamps on all extracted files...\n")); break; case WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END: if (info->extract.extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) { @@ -1815,6 +1847,183 @@ out: return ret; } +static bool +is_root_wim_path(const tchar *path) +{ + const tchar *p; + for (p = path; *p; p++) + if (*p != T('\\') && *p != T('\\')) + return false; + return true; +} + +static void +free_extract_commands(struct wimlib_extract_command *cmds, size_t num_cmds, + tchar *dest_dir) +{ + for (size_t i = 0; i < num_cmds; i++) + if (cmds[i].fs_dest_path != dest_dir) + free(cmds[i].fs_dest_path); + free(cmds); +} + +static struct wimlib_extract_command * +prepare_extract_commands(tchar **argv, int argc, int extract_flags, + tchar *dest_dir, size_t *num_cmds_ret) +{ + struct wimlib_extract_command *cmds; + size_t num_cmds; + tchar *emptystr = T(""); + + num_cmds = argc; + if (argc == 0) { + num_cmds = 1; + argv = &emptystr; + } + cmds = calloc(num_cmds, sizeof(cmds[0])); + if (!cmds) { + imagex_error("Out of memory!"); + return NULL; + } + + for (size_t i = 0; i < num_cmds; i++) { + cmds[i].extract_flags = extract_flags; + cmds[i].wim_source_path = argv[i]; + if (is_root_wim_path(argv[i])) { + cmds[i].fs_dest_path = dest_dir; + } else { + size_t len = tstrlen(dest_dir) + 1 + tstrlen(argv[i]); + cmds[i].fs_dest_path = malloc((len + 1) * sizeof(tchar)); + if (!cmds[i].fs_dest_path) + goto oom; + tsprintf(cmds[i].fs_dest_path, "%"TS"/%"TS, dest_dir, tbasename(argv[i])); + } + } + + *num_cmds_ret = num_cmds; + return cmds; +oom: + free_extract_commands(cmds, num_cmds, dest_dir); + return NULL; +} + +static int +imagex_extract(int argc, tchar **argv) +{ + int c; + int open_flags = WIMLIB_OPEN_FLAG_SPLIT_OK; + int image; + WIMStruct *wim; + int ret; + const tchar *wimfile; + const tchar *image_num_or_name; + tchar *dest_dir = T("."); + int extract_flags = WIMLIB_EXTRACT_FLAG_SEQUENTIAL; + + const tchar *swm_glob = NULL; + WIMStruct **additional_swms = NULL; + unsigned num_additional_swms = 0; + + struct wimlib_extract_command *cmds; + size_t num_cmds; + + for_opt(c, extract_options) { + switch (c) { + case IMAGEX_CHECK_OPTION: + open_flags |= WIMLIB_OPEN_FLAG_CHECK_INTEGRITY; + break; + case IMAGEX_VERBOSE_OPTION: + extract_flags |= WIMLIB_EXTRACT_FLAG_VERBOSE; + break; + case IMAGEX_REF_OPTION: + swm_glob = optarg; + break; + case IMAGEX_UNIX_DATA_OPTION: + extract_flags |= WIMLIB_EXTRACT_FLAG_UNIX_DATA; + break; + case IMAGEX_NO_ACLS_OPTION: + extract_flags |= WIMLIB_EXTRACT_FLAG_NO_ACLS; + break; + case IMAGEX_STRICT_ACLS_OPTION: + extract_flags |= WIMLIB_EXTRACT_FLAG_STRICT_ACLS; + break; + case IMAGEX_DEST_DIR_OPTION: + dest_dir = optarg; + break; + case IMAGEX_TO_STDOUT_OPTION: + extract_flags |= WIMLIB_EXTRACT_FLAG_TO_STDOUT; + break; + default: + usage(EXTRACT); + ret = -1; + goto out; + } + } + argc -= optind; + argv += optind; + + if (argc < 2) { + usage(EXTRACT); + ret = -1; + goto out; + } + wimfile = argv[0]; + image_num_or_name = argv[1]; + + argc -= 2; + argv += 2; + + cmds = prepare_extract_commands(argv, argc, extract_flags, dest_dir, + &num_cmds); + if (!cmds) { + ret = -1; + goto out; + } + + ret = wimlib_open_wim(wimfile, open_flags, &wim, imagex_progress_func); + if (ret) + goto out_free_cmds; + + image = wimlib_resolve_image(wim, image_num_or_name); + ret = verify_image_exists_and_is_single(image, + image_num_or_name, + wimfile); + if (ret) + goto out_wimlib_free; + + if (swm_glob) { + ret = open_swms_from_glob(swm_glob, wimfile, open_flags, + &additional_swms, + &num_additional_swms); + if (ret) + goto out_wimlib_free; + } + +#ifdef __WIN32__ + win32_acquire_restore_privileges(); +#endif + + ret = wimlib_extract_files(wim, image, 0, cmds, num_cmds, + additional_swms, num_additional_swms, + imagex_progress_func); + if (ret == 0) + tprintf(T("Done extracting files.\n")); +#ifdef __WIN32__ + win32_release_restore_privileges(); +#endif + if (additional_swms) { + for (unsigned i = 0; i < num_additional_swms; i++) + wimlib_free(additional_swms[i]); + free(additional_swms); + } +out_wimlib_free: + wimlib_free(wim); +out_free_cmds: + free_extract_commands(cmds, num_cmds, dest_dir); +out: + return ret; +} + /* Prints information about a WIM file; also can mark an image as bootable, * change the name of an image, or change the description of an image. */ static int @@ -2443,6 +2652,7 @@ static const struct imagex_command imagex_commands[] = { {T("delete"), imagex_delete, DELETE}, {T("dir"), imagex_dir, DIR}, {T("export"), imagex_export, EXPORT}, + {T("extract"), imagex_extract, EXTRACT}, {T("info"), imagex_info, INFO}, {T("join"), imagex_join, JOIN}, {T("mount"), imagex_mount_rw_or_ro, MOUNT}, diff --git a/src/add_image.c b/src/add_image.c index 9bd597a1..c39ded67 100644 --- a/src/add_image.c +++ b/src/add_image.c @@ -478,29 +478,6 @@ exclude_path(const tchar *path, size_t path_len, } -/* Strip leading and trailing forward slashes from a string. Modifies it in - * place and returns the stripped string. */ -static const tchar * -canonicalize_target_path(tchar *target_path) -{ - tchar *p; - if (target_path == NULL) - return T(""); - for (;;) { - if (*target_path == T('\0')) - return target_path; - else if (*target_path == T('/')) - target_path++; - else - break; - } - - p = tstrchr(target_path, T('\0')) - 1; - while (*p == T('/')) - *p-- = T('\0'); - return target_path; -} - /* Strip leading and trailing slashes from the target paths, and translate all * backslashes in the source and target paths into forward slashes. */ static void @@ -515,12 +492,8 @@ canonicalize_sources_and_targets(struct wimlib_capture_source *sources, /* The Windows API can handle forward slashes. Just get rid of * backslashes to avoid confusing other parts of the library * code. */ - zap_backslashes(sources->fs_source_path); - if (sources->wim_target_path) - zap_backslashes(sources->wim_target_path); - - sources->wim_target_path = - (tchar*)canonicalize_target_path(sources->wim_target_path); + sources->fs_source_path = canonicalize_fs_path(sources->fs_source_path); + sources->wim_target_path = canonicalize_wim_path(sources->wim_target_path); DEBUG("Canonical target: \"%"TS"\"", sources->wim_target_path); sources++; } diff --git a/src/dentry.c b/src/dentry.c index 73dba52b..2e5090f2 100644 --- a/src/dentry.c +++ b/src/dentry.c @@ -1270,13 +1270,13 @@ replace_forbidden_characters(utf16lechar *name) #ifdef __WIN32__ if (wcschr(L"<>:\"/\\|?*", (wchar_t)*p)) #else - if (*p == '/') + if (*p == cpu_to_le16('/')) #endif { #ifdef __WIN32__ *p = cpu_to_le16(0xfffd); #else - *p = '?'; + *p = cpu_to_le16('?'); #endif if (name) { WARNING("File, directory, or stream name \"%"WS"\"\n" diff --git a/src/dentry.h b/src/dentry.h index 1629111d..d3060be9 100644 --- a/src/dentry.h +++ b/src/dentry.h @@ -232,6 +232,8 @@ struct wim_inode { /* %true iff verify_inode() has run on this inode. */ u8 i_verified : 1; + u8 i_visited : 1; + /* Used only in NTFS-mode extraction */ u8 i_dos_name_extracted : 1; diff --git a/src/extract_image.c b/src/extract_image.c index cd23e45d..5522a1fc 100644 --- a/src/extract_image.c +++ b/src/extract_image.c @@ -57,6 +57,7 @@ # include #endif + #ifndef __WIN32__ /* Returns the number of components of @path. */ @@ -450,7 +451,7 @@ unix_do_apply_dentry(const char *output_path, size_t output_path_len, 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; @@ -508,29 +509,42 @@ 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 = dentry->_full_path + 1; + size_t 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; + + 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; - 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(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 } @@ -539,26 +553,10 @@ 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 } @@ -679,25 +677,62 @@ inode_find_streams_for_extraction(struct wim_inode *inode, } } +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 wim_image_metadata *imd, +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 wim_dentry *dentry; + struct find_streams_ctx ctx; - for_lookup_table_entry(lookup_table, lte_zero_out_refcnt, NULL); - INIT_LIST_HEAD(stream_list); - image_for_each_inode(inode, imd) { - 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 { @@ -806,32 +841,33 @@ sort_stream_list_by_wim_position(struct list_head *stream_list) } -/* 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) +extract_tree(WIMStruct *wim, int image, + 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 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.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.wimfile_name = wim->filename; args.progress.extract.image = 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, image); + args.progress.extract.extract_root_wim_source_path = wim_source_path; args.progress.extract.target = target; } @@ -841,28 +877,35 @@ 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) - goto out; + root = get_dentry(wim, wim_source_path); + if (!root) { + ERROR("Path \"%"TS"\" does not exist in WIM image %d", + wim_source_path, 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(wim_get_current_image_metadata(w), + find_streams_for_extraction(root, &stream_list, - w->lookup_table, extract_flags); + 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); } @@ -882,17 +925,17 @@ extract_single_image(WIMStruct *w, int image, &args.progress); } - ret = calculate_dentry_tree_full_paths(wim_root_dentry(w)); + ret = calculate_dentry_tree_full_paths(root); if (ret) - goto out; + 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) - goto out; + goto out_mark_inodes_unvisited; if (progress_func) { progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END, @@ -901,8 +944,10 @@ extract_single_image(WIMStruct *w, int image, if (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) { args.target_realpath = realpath(target, NULL); - if (!args.target_realpath) - return WIMLIB_ERR_NOMEM; + if (!args.target_realpath) { + ret = WIMLIB_ERR_NOMEM; + goto out_mark_inodes_unvisited; + } args.target_realpath_len = tstrlen(args.target_realpath); } @@ -917,18 +962,21 @@ 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) 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_free_target_realpath: FREE(args.target_realpath); -out: +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) { @@ -940,6 +988,237 @@ out: } } #endif +out: + return ret; +} + +static int +check_extract_command(struct wimlib_extract_command *cmd, + bool multiple_commands, + int wim_header_flags) +{ + int extract_flags; + bool is_entire_image = (cmd->wim_source_path == T('\0')); + + if (cmd->fs_dest_path[0] == T('\0')) + return WIMLIB_ERR_INVALID_PARAM; + + extract_flags = cmd->extract_flags; + + 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 (!is_entire_image && + (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)) + { + ERROR("Can only extract entire 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 + } + + 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) + { + 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; +} + + +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; + + ret = select_wim_image(wim, image); + if (ret) + return ret; + + ret = wim_checksum_unhashed_streams(wim); + if (ret) + return ret; + + for (size_t i = 0; i < num_cmds; i++) { + ret = check_extract_command(&cmds[i], num_cmds > 1, + 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; + } + } + + for (size_t i = 0; i < num_cmds; i++) { + ret = extract_tree(wim, image, + cmds[i].wim_source_path, + cmds[i].fs_dest_path, + cmds[i].extract_flags, + progress_func); + if (ret) + return ret; + } + return 0; +} + +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; +} + +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); + 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; } @@ -964,12 +1243,12 @@ image_name_ok_as_dir(const tchar *image_name) /* 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; @@ -982,8 +1261,8 @@ extract_all_images(WIMStruct *w, 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 { @@ -991,9 +1270,9 @@ extract_all_images(WIMStruct *w, * characters. */ 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; @@ -1002,7 +1281,7 @@ 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, @@ -1010,103 +1289,44 @@ 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 - } - - if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX | - WIMLIB_EXTRACT_FLAG_RPFIX)) == - (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) - if (w->hdr.flags & WIM_HDR_FLAG_RP_FIX) - extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX; - - ret = verify_swm_set(w, additional_swms, num_additional_swms); - if (ret) - return ret; - - ret = wim_checksum_unhashed_streams(w); + 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) 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; } diff --git a/src/util.c b/src/util.c index 5421a6b5..35fbf6f1 100644 --- a/src/util.c +++ b/src/util.c @@ -347,6 +347,8 @@ static const tchar *error_strings[] = { = T("Failed to open a file"), [WIMLIB_ERR_OPENDIR] = T("Failed to open a directory"), + [WIMLIB_ERR_PATH_DOES_NOT_EXIST] + = T("The path does not exist in the WIM image"), [WIMLIB_ERR_READ] = T("Could not read data from a file"), [WIMLIB_ERR_READLINK] @@ -570,6 +572,45 @@ zap_backslashes(tchar *s) } } +tchar * +canonicalize_fs_path(const tchar *fs_path) +{ + tchar *canonical_path; + + if (!fs_path) + fs_path = T(""); + canonical_path = TSTRDUP(fs_path); + zap_backslashes(canonical_path); + return canonical_path; +} + +/* Strip leading and trailing slashes from a string. Also translates + * backslashes into forward slashes. */ +tchar * +canonicalize_wim_path(const tchar *wim_path) +{ + tchar *p; + tchar *canonical_path; + + if (wim_path == NULL) { + wim_path = T(""); + } else { + while (*wim_path == T('/') || *wim_path == T('\\')) + wim_path++; + } + canonical_path = TSTRDUP(wim_path); + if (canonical_path) { + zap_backslashes(canonical_path); + for (p = tstrchr(canonical_path, T('\0')) - 1; + p >= canonical_path && *p == T('/'); + p--) + { + *p = T('\0'); + } + } + return canonical_path; +} + /* Like read(), but keep trying until everything has been written or we know for * sure that there was an error (or end-of-file). */ size_t diff --git a/src/util.h b/src/util.h index 660dcf99..3d22cf40 100644 --- a/src/util.h +++ b/src/util.h @@ -279,6 +279,12 @@ wimlib_printf(const tchar *format, ...) FORMAT(printf, 1, 2); extern void zap_backslashes(tchar *s); +extern tchar * +canonicalize_wim_path(const tchar *wim_path); + +extern tchar * +canonicalize_fs_path(const tchar *fs_path); + static inline u64 hash_u64(u64 n) { diff --git a/src/wimlib.h b/src/wimlib.h index 7a2fad7d..c4979458 100644 --- a/src/wimlib.h +++ b/src/wimlib.h @@ -302,6 +302,9 @@ enum wimlib_progress_msg { * ::wimlib_progress_info.extract. */ WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN, + /** XXX */ + WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN, + /** The directory structure of the WIM image is about to be extracted. * @a info will point to ::wimlib_progress_info.extract. */ WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN, @@ -328,6 +331,9 @@ enum wimlib_progress_msg { * ::wimlib_progress_info.extract. */ WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END, + /** XXX */ + WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END, + /** The directory or NTFS volume is about to be scanned to build a tree * of WIM dentries in-memory. @a info will point to * ::wimlib_progress_info.scan. */ @@ -497,6 +503,11 @@ union wimlib_progress_info { * special cases (hard links, symbolic links, and alternate data * streams.) */ uint64_t num_streams; + + /** Path to the root dentry within the WIM for the tree that is + * being extracted. Will be the empty string when extracting a + * full image. */ + const wimlib_tchar *extract_root_wim_source_path; } extract; /** Valid on messages ::WIMLIB_PROGRESS_MSG_RENAME. */ @@ -767,6 +778,9 @@ struct wimlib_capture_config { * WIM_HDR_FLAG_RP_FIX flag in the WIM header. */ #define WIMLIB_EXTRACT_FLAG_NORPFIX 0x00000200 +/** Ignore the target directory; only extract file data to standard output. */ +#define WIMLIB_EXTRACT_FLAG_TO_STDOUT 0x00000400 + /****************************** * WIMLIB_MOUNT_FLAG_* * ******************************/ @@ -858,67 +872,38 @@ struct wimlib_capture_config { * locale's character encoding. */ #define WIMLIB_INIT_FLAG_ASSUME_UTF8 0x00000001 - -#if 0 -/**************************************************************** - * Definition of struct wimlib_modify_command, with various flags - ****************************************************************/ - -enum { - WIMLIB_MOVE_TREE_FLAG_OVERWRITE_ALL = 0x1, - WIMLIB_MOVE_TREE_FLAG_OVERWRITE_NONDIRECTORIES = 0x2, - WIMLIB_MOVE_TREE_FLAG_OVERWRITE_EMPTY_DIRECTORIES = 0x4, - WIMLIB_MOVE_TREE_FLAG_OVERWRITE_DIRECTORIES = 0x8, -}; - -enum { - WIMLIB_DELETE_TREE_FLAG_FORCE = 0x1, - WIMLIB_DELETE_TREE_FLAG_RECURSIVE = 0x2, - WIMLIB_DELETE_TREE_FLAG_REMOVE_EMPTY_DIR = 0x4, -}; - -enum { - WIMLIB_ADD_TREE_FLAG_DEREFERENCE = 0x1, - WIMLIB_ADD_TREE_FLAG_VERBOSE = 0x2, - WIMLIB_ADD_TREE_FLAG_UNIX_DATA = 0x4, - WIMLIB_ADD_TREE_FLAG_NOACLS = 0x8, - WIMLIB_ADD_TREE_FLAG_NTFS_VOLUME = 0x01, - WIMLIB_ADD_TREE_FLAG_OVERLAY = 0x02, - WIMLIB_ADD_TREE_FLAG_MAKE_NECESSARY_DIRS = 0x04, -}; - -enum wimlib_modify_op { - WIMLIB_MODIFY_OP_DELETE_TREE, - WIMLIB_MODIFY_OP_ADD_TREE, - WIMLIB_MODIFY_OP_MOVE_TREE, -}; - -struct wimlib_modify_command { - enum wimlib_modify_op op; +/** XXX */ +struct wimlib_update_command { + enum { + WIMLIB_UPDATE_OP_ADD, + WIMLIB_UPDATE_OP_DELETE, + WIMLIB_UPDATE_OP_MOVE, + } op; union { - struct wimlib_modify_command_delete_tree { - int delete_tree_flags; - const wimlib_tchar *tree_wim_path; - unsigned long reserved; - } delete_tree; - - struct wimlib_modify_command_add_tree { - int add_tree_flags; + struct { const wimlib_tchar *fs_source_path; const wimlib_tchar *wim_target_path; - unsigned long reserved; - } add_tree; - - struct wimlib_modify_command_move_tree { - int move_tree_flags; + const struct wimlib_capture_config *config; + int add_flags; + } add; + struct { + const wimlib_tchar *path_in_wim; + int delete_flags; + } delete; + struct { const wimlib_tchar *wim_source_path; const wimlib_tchar *wim_target_path; - unsigned long reserved; - } move_tree; + int rename_flags; + } rename; }; }; -#endif +/** XXX */ +struct wimlib_extract_command { + wimlib_tchar *wim_source_path; + wimlib_tchar *fs_dest_path; + int extract_flags; +}; /** * Possible values of the error code returned by many functions in wimlib. @@ -991,6 +976,7 @@ enum wimlib_error_code { WIMLIB_ERR_VOLUME_LACKS_FEATURES, WIMLIB_ERR_WRITE, WIMLIB_ERR_XML, + WIMLIB_ERR_PATH_DOES_NOT_EXIST, }; @@ -1302,6 +1288,17 @@ wimlib_export_image(WIMStruct *src_wim, int src_image, unsigned num_additional_swms, wimlib_progress_func_t progress_func); +/** XXX */ +extern 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); + /** * Extracts an image, or all images, from a standalone or split WIM file to a * directory or a NTFS volume. diff --git a/src/wimlib_internal.h b/src/wimlib_internal.h index 82eb42ee..1ddec8e3 100644 --- a/src/wimlib_internal.h +++ b/src/wimlib_internal.h @@ -539,6 +539,9 @@ write_metadata_resource(WIMStruct *w); struct apply_args { WIMStruct *w; const tchar *target; + unsigned target_nchars; + unsigned wim_source_path_nchars; + struct wim_dentry *extract_root; tchar *target_realpath; unsigned target_realpath_len; int extract_flags;