X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fextract.c;h=51684f0e2b9c3249fc4865516eac1748e757e114;hp=95071bd7bfbb35b45dec106755a88b1c0db204e8;hb=773fafce334bd429d35365adfd377a0b63c66953;hpb=61db93f82eca3fe9f7676355c709c58cc425a6ad diff --git a/src/extract.c b/src/extract.c index 95071bd7..51684f0e 100644 --- a/src/extract.c +++ b/src/extract.c @@ -1067,10 +1067,63 @@ dentry_extract_skeleton(struct wim_dentry *dentry, void *_ctx) { struct apply_ctx *ctx = _ctx; tchar path[ctx->ops->path_max]; + struct wim_dentry *orig_dentry; + struct wim_dentry *other_dentry; + int ret; + /* Here we may re-order the extraction of multiple names (hard links) + * for the same file in the same directory in order to ensure the short + * (DOS) name is set correctly. A short name is always associated with + * exactly one long name, and at least on NTFS, only one long name for a + * file can have a short name associated with it. (More specifically, + * there can be unlimited names in the POSIX namespace, but only one + * name can be in the Win32+DOS namespace, or one name in the Win32 + * namespace with a corresponding name in the DOS namespace.) To ensure + * the short name of a file is associated with the correct long name in + * a directory, we extract the long name with a corresponding short name + * before any additional names. This can affect NTFS-3g extraction + * (which uses ntfs_set_ntfs_dos_name(), which doesn't allow specifying + * the long name to associate with a short name) and may affect Win32 + * extraction as well (which uses SetFileShortName()). */ + + if (dentry->skeleton_extracted) + return 0; + orig_dentry = NULL; + if (ctx->supported_features.short_names + && !dentry_has_short_name(dentry) + && !dentry->d_inode->i_dos_name_extracted) + { + inode_for_each_dentry(other_dentry, dentry->d_inode) { + if (dentry_has_short_name(other_dentry) + && !other_dentry->skeleton_extracted + && other_dentry->parent == dentry->parent) + { + DEBUG("Creating %"TS" before %"TS" " + "to guarantee correct DOS name extraction", + dentry_full_path(other_dentry), + dentry_full_path(dentry)); + orig_dentry = dentry; + dentry = other_dentry; + break; + } + } + } +again: if (!build_extraction_path(path, dentry, ctx)) return 0; - return do_dentry_extract_skeleton(path, dentry, ctx); + ret = do_dentry_extract_skeleton(path, dentry, ctx); + if (ret) + return ret; + + dentry->skeleton_extracted = 1; + + if (orig_dentry) { + dentry = orig_dentry; + orig_dentry = NULL; + goto again; + } + dentry->d_inode->i_dos_name_extracted = 1; + return 0; } /* Create a file or directory, then immediately extract all streams. This @@ -1083,13 +1136,13 @@ dentry_extract(struct wim_dentry *dentry, void *_ctx) tchar path[ctx->ops->path_max]; int ret; - if (!build_extraction_path(path, dentry, ctx)) - return 0; - - ret = do_dentry_extract_skeleton(path, dentry, ctx); + ret = dentry_extract_skeleton(dentry, ctx); if (ret) return ret; + if (!build_extraction_path(path, dentry, ctx)) + return 0; + return extract_streams(path, ctx, dentry, NULL, NULL); } @@ -1203,8 +1256,7 @@ extract_stream_list(struct apply_ctx *ctx) bool can_seek; int ret; - can_seek = !(lseek(ctx->wim->in_fd.fd, 0, SEEK_CUR) == (off_t)-1 && - errno == ESPIPE); + can_seek = (lseek(ctx->wim->in_fd.fd, 0, SEEK_CUR) != -1); list_for_each_entry(lte, &ctx->stream_list, extraction_list) { ret = extract_stream_instances(lte, ctx, can_seek); if (ret) @@ -1213,10 +1265,13 @@ extract_stream_list(struct apply_ctx *ctx) return 0; } +#define PWM_ALLOW_WIM_HDR 0x00001 +#define PWM_SILENT_EOF 0x00002 + /* Read the header from a stream in a pipable WIM. */ static int read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte, - bool allow_header) + int flags) { struct pwm_stream_hdr stream_hdr; int ret; @@ -1225,7 +1280,7 @@ read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte, if (ret) goto read_error; - if (allow_header && stream_hdr.magic == PWM_MAGIC) { + if ((flags & PWM_ALLOW_WIM_HDR) && stream_hdr.magic == PWM_MAGIC) { u8 buf[WIM_HEADER_DISK_SIZE - sizeof(stream_hdr)]; ret = full_read(&pwm->in_fd, buf, sizeof(buf)); if (ret) @@ -1256,7 +1311,8 @@ read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte, return 0; read_error: - ERROR_WITH_ERRNO("Error reading pipable WIM from pipe"); + if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF)) + ERROR_WITH_ERRNO("Error reading pipable WIM from pipe"); return ret; } @@ -1277,6 +1333,7 @@ extract_streams_from_pipe(struct apply_ctx *ctx) struct wim_lookup_table_entry *needed_lte; struct wim_lookup_table *lookup_table; int ret; + int pwm_flags; ret = WIMLIB_ERR_NOMEM; found_lte = new_lookup_table_entry(); @@ -1284,11 +1341,19 @@ extract_streams_from_pipe(struct apply_ctx *ctx) goto out; lookup_table = ctx->wim->lookup_table; - + pwm_flags = PWM_ALLOW_WIM_HDR; + if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) + pwm_flags |= PWM_SILENT_EOF; while (ctx->num_streams_remaining) { - ret = read_pwm_stream_header(ctx->wim, found_lte, true); - if (ret) + ret = read_pwm_stream_header(ctx->wim, found_lte, pwm_flags); + if (ret) { + if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE && + (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) + { + goto resume_done; + } goto out_free_found_lte; + } if ((found_lte->resource_location != RESOURCE_NONEXISTENT) && !(found_lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) @@ -1317,6 +1382,10 @@ out_free_found_lte: free_lookup_table_entry(found_lte); out: return ret; + +resume_done: + /* TODO */ + return 0; } /* Finish extracting a file, directory, or symbolic link by setting file @@ -1338,42 +1407,6 @@ dentry_extract_final(struct wim_dentry *dentry, void *_ctx) return extract_timestamps(path, ctx, dentry); } -/* Sorts a list of streams in ascending order of their offset in the WIM file in - * order to prepare for sequential extraction. */ -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, 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 a WIM dentry to standard output. * @@ -1615,9 +1648,11 @@ dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore) dentry->extraction_skipped = 0; dentry->was_hardlinked = 0; + dentry->skeleton_extracted = 0; inode->i_visited = 0; FREE(inode->i_extracted_file); inode->i_extracted_file = NULL; + inode->i_dos_name_extracted = 0; if ((void*)dentry->extraction_name != (void*)dentry->file_name) FREE(dentry->extraction_name); dentry->extraction_name = NULL; @@ -1686,76 +1721,90 @@ static int do_feature_check(const struct wim_features *required_features, const struct wim_features *supported_features, int extract_flags, - const struct apply_operations *ops) + const struct apply_operations *ops, + const tchar *wim_source_path) { + const tchar *loc; + const tchar *mode = "this extraction mode"; + + if (wim_source_path[0] == '\0') + loc = "the WIM image"; + else + loc = wim_source_path; + + /* We're an archive program, so theoretically we can do what we want + * with FILE_ATTRIBUTE_ARCHIVE (which is a dumb flag anyway). Don't + * bother the user about it. */ +#if 0 if (required_features->archive_files && !supported_features->archive_files) { WARNING( - "%lu files are marked as archived, but this attribute\n" -" is not supported in this extraction mode or volume.", - required_features->archive_files); + "%lu files in %"TS" are marked as archived, but this attribute\n" +" is not supported in %"TS".", + required_features->archive_files, loc, mode); } +#endif if (required_features->hidden_files && !supported_features->hidden_files) { WARNING( - "%lu files are marked as hidden, but this attribute\n" -" is not supported in this extraction mode or volume.", - required_features->hidden_files); + "%lu files in %"TS" are marked as hidden, but this\n" +" attribute is not supported in %"TS".", + required_features->hidden_files, loc, mode); } if (required_features->system_files && !supported_features->system_files) { WARNING( - "%lu files are marked as system files, but this attribute\n" -" is not supported in this extraction mode or volume.", - required_features->system_files); + "%lu files in %"TS" are marked as system files,\n" +" but this attribute is not supported in %"TS".", + required_features->system_files, loc, mode); } if (required_features->compressed_files && !supported_features->compressed_files) { WARNING( - "%lu files are marked as being transparently compressed, but\n" -" transparent compression is not supported in this extraction\n" -" mode or volume. These files will be extracted as uncompressed.", - required_features->compressed_files); + "%lu files in %"TS" are marked as being transparently\n" +" compressed, but transparent compression is not supported in\n" +" %"TS". These files will be extracted as uncompressed.", + required_features->compressed_files, loc, mode); } if (required_features->encrypted_files && !supported_features->encrypted_files) { WARNING( - "%lu files are marked as being encrypted, but encryption is not\n" -" supported in this extraction mode or volume. These files will be\n" -" extracted as raw encrypted data instead.", - required_features->encrypted_files); + "%lu files in %"TS" are marked as being encrypted,\n" +" but encryption is not supported in %"TS". These files\n" +" will be extracted as raw encrypted data instead.", + required_features->encrypted_files, loc, mode); } if (required_features->not_context_indexed_files && !supported_features->not_context_indexed_files) { WARNING( - "%lu files are marked as not content indexed, but this attribute\n" -" is not supported in this extraction mode or volume.", - required_features->not_context_indexed_files); + "%lu files in %"TS" are marked as not content indexed,\n" +" but this attribute is not supported in %"TS".", + required_features->not_context_indexed_files, loc, mode); } if (required_features->sparse_files && !supported_features->sparse_files) { WARNING( - "%lu files are marked as sparse, but creating sparse files is not\n" -" supported in this extraction mode or volume. These files will be\n" -" extracted as non-sparse.", - required_features->not_context_indexed_files); + "%lu files in %"TS" are marked as sparse, but creating\n" +" sparse files is not supported in %"TS". These files\n" +" will be extracted as non-sparse.", + required_features->sparse_files, loc, mode); } if (required_features->named_data_streams && !supported_features->named_data_streams) { WARNING( - "%lu files contain one or more alternate (named) data streams,\n" -" which are not supported in this extraction mode or volume.\n" + "%lu files in %"TS" contain one or more alternate (named)\n" +" data streams, which are not supported in %"TS".\n" " Alternate data streams will NOT be extracted.", - required_features->named_data_streams); + required_features->named_data_streams, loc, mode); } if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK | @@ -1764,19 +1813,19 @@ do_feature_check(const struct wim_features *required_features, supported_features->named_data_streams) { WARNING( - "%lu files contain one or more alternate (named) data streams,\n" -" which are not supported in linked extraction mode.\n" + "%lu files in %"TS" contain one or more alternate (named)\n" +" data streams, which are not supported in linked extraction mode.\n" " Alternate data streams will NOT be extracted.", - required_features->named_data_streams); + required_features->named_data_streams, loc); } if (required_features->hard_links && !supported_features->hard_links) { WARNING( - "%lu files are hard links, but hard links are not supported in\n" -" this extraction mode or volume. Hard links will be extracted as\n" + "%lu files in %"TS" are hard links, but hard links are\n" +" not supported in %"TS". Hard links will be extracted as\n" " duplicate copies of the linked files.", - required_features->hard_links); + required_features->hard_links, loc, mode); } if (required_features->reparse_points && !supported_features->reparse_points) @@ -1784,16 +1833,17 @@ do_feature_check(const struct wim_features *required_features, if (supported_features->symlink_reparse_points) { if (required_features->other_reparse_points) { WARNING( - "%lu files are reparse points that are neither symbolic links\n" -" nor junction points and are not supported in this extraction mode\n" -" or volume. These reparse points will not be extracted.", - required_features->other_reparse_points); + "%lu files in %"TS" are reparse points that are neither\n" +" symbolic links nor junction points and are not supported in\n" +" %"TS". These reparse points will not be extracted.", + required_features->other_reparse_points, loc, + mode); } } else { WARNING( - "%lu files are reparse points, which are not supported in this\n" -" extraction mode or volume and will not be extracted.", - required_features->reparse_points); + "%lu files in %"TS" are reparse points, which are\n" +" not supported in %"TS" and will not be extracted.", + required_features->reparse_points, loc, mode); } } @@ -1801,40 +1851,37 @@ do_feature_check(const struct wim_features *required_features, !supported_features->security_descriptors) { WARNING( - "%lu files have Windows NT security descriptors, but extracting\n" -" security descriptors is not supported in this extraction mode\n" -" or volume. No security descriptors will be extracted.", - required_features->security_descriptors); + "%lu files in %"TS" have Windows NT security descriptors,\n" +" but extracting security descriptors is not supported in\n" +" %"TS". No security descriptors will be extracted.", + required_features->security_descriptors, loc, mode); } if (required_features->short_names && !supported_features->short_names) { WARNING( - "%lu files have short (DOS) names, but extracting short names\n" -" is not supported in this extraction mode or volume. Short names\n" -" will not be extracted.\n", - required_features->short_names); + "%lu files in %"TS" have short (DOS) names, but\n" +" extracting short names is not supported in %"TS".\n" +" Short names will not be extracted.\n", + required_features->short_names, loc, mode); } if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) && required_features->unix_data && !supported_features->unix_data) { - ERROR("UNIX data not supported in this extraction mode " - "or volume", ops->name); + ERROR("Extracting UNIX data is not supported in %"TS, mode); return WIMLIB_ERR_UNSUPPORTED; } if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) && required_features->short_names && !supported_features->short_names) { - ERROR("Short names are not supported in this extraction " - "mode or volume", ops->name); + ERROR("Extracting short names is not supported in %"TS"", mode); return WIMLIB_ERR_UNSUPPORTED; } if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) && !ops->set_timestamps) { - ERROR("Timestamps are not supported in this extraction " - "mode or volume", ops->name); + ERROR("Extracting timestamps is not supported in %"TS"", mode); return WIMLIB_ERR_UNSUPPORTED; } if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS | @@ -1843,8 +1890,7 @@ do_feature_check(const struct wim_features *required_features, required_features->security_descriptors && !supported_features->security_descriptors) { - ERROR("Security descriptors not supported in this extraction " - "mode or volume."); + ERROR("Extracting security descriptors is not supported in %"TS, mode); return WIMLIB_ERR_UNSUPPORTED; } @@ -1852,7 +1898,7 @@ do_feature_check(const struct wim_features *required_features, !supported_features->hard_links) { ERROR("Hard link extraction mode requested, but " - "extraction mode or volume does not support hard links!"); + "%"TS" does not support hard links!", mode); return WIMLIB_ERR_UNSUPPORTED; } @@ -1860,8 +1906,8 @@ do_feature_check(const struct wim_features *required_features, !supported_features->symlink_reparse_points) { ERROR("Symbolic link extraction mode requested, but " - "extraction mode or volume does not support symbolic " - "links!"); + "%"TS" does not support symbolic " + "links!", mode); return WIMLIB_ERR_UNSUPPORTED; } return 0; @@ -1975,7 +2021,7 @@ extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target, dentry_tree_get_features(root, &required_features); ret = do_feature_check(&required_features, &ctx.supported_features, - extract_flags, ctx.ops); + extract_flags, ctx.ops, wim_source_path); if (ret) goto out_finish_or_abort_extract; @@ -2022,8 +2068,7 @@ extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target, * output. In that case, "root" should be a single file, not a * directory tree. (If not, extract_dentry_to_stdout() will * return an error.) */ - if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT && - !(extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)) { + if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) { ret = extract_dentry_to_stdout(root); goto out_teardown_stream_list; } @@ -2035,7 +2080,10 @@ extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target, WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_SEQUENTIAL) { - ret = sort_stream_list_by_wim_position(&ctx.stream_list); + ret = sort_stream_list_by_sequential_order( + &ctx.stream_list, + offsetof(struct wim_lookup_table_entry, + extraction_list)); if (ret) goto out_teardown_stream_list; } @@ -2097,9 +2145,12 @@ extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target, if (progress_func) progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN, &ctx.progress); - ret = for_dentry_in_tree(root, dentry_extract_skeleton, &ctx); - if (ret) - goto out_free_realtarget; + + if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) { + ret = for_dentry_in_tree(root, dentry_extract_skeleton, &ctx); + if (ret) + goto out_free_realtarget; + } if (progress_func) progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END, &ctx.progress); @@ -2214,6 +2265,11 @@ check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags) WIMLIB_EXTRACT_FLAG_NORPFIX)) return WIMLIB_ERR_INVALID_PARAM; + if ((extract_flags & + (WIMLIB_EXTRACT_FLAG_RESUME | + WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME) + return WIMLIB_ERR_INVALID_PARAM; + if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) { #ifndef WITH_NTFS_3G ERROR("wimlib was compiled without support for NTFS-3g, so\n" @@ -2276,8 +2332,7 @@ do_wimlib_extract_files(WIMStruct *wim, return ret; /* Make sure there are no streams in the WIM that have not been - * checksummed yet. Needed at least because 'unhashed_list' aliases - * 'extraction_list' in `struct wim_lookup_table_entry'. */ + * checksummed yet. */ ret = wim_checksum_unhashed_streams(wim); if (ret) return ret; @@ -2461,12 +2516,10 @@ extract_all_images(WIMStruct *wim, extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE; -#ifdef WITH_NTFS_3G if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) { ERROR("Cannot extract multiple images in NTFS extraction mode."); return WIMLIB_ERR_INVALID_PARAM; } -#endif if (tstat(target, &stbuf)) { if (errno == ENOENT) { @@ -2602,7 +2655,7 @@ wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, * WIMs.) */ { struct wim_lookup_table_entry xml_lte; - ret = read_pwm_stream_header(pwm, &xml_lte, false); + ret = read_pwm_stream_header(pwm, &xml_lte, 0); if (ret) goto out_wimlib_free; @@ -2629,16 +2682,26 @@ wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, /* Get image index (this may use the XML data that was just read to * resolve an image name). */ - image = wimlib_resolve_image(pwm, image_num_or_name); - if (image == WIMLIB_NO_IMAGE) { - ERROR("\"%"TS"\" is not a valid image in the pipable WIM!", - image_num_or_name); - ret = WIMLIB_ERR_INVALID_IMAGE; - goto out_wimlib_free; - } else if (image == WIMLIB_ALL_IMAGES) { - ERROR("Applying all images from a pipe is not supported."); - ret = WIMLIB_ERR_INVALID_IMAGE; - goto out_wimlib_free; + if (image_num_or_name) { + image = wimlib_resolve_image(pwm, image_num_or_name); + if (image == WIMLIB_NO_IMAGE) { + ERROR("\"%"TS"\" is not a valid image in the pipable WIM!", + image_num_or_name); + ret = WIMLIB_ERR_INVALID_IMAGE; + goto out_wimlib_free; + } else if (image == WIMLIB_ALL_IMAGES) { + ERROR("Applying all images from a pipe is not supported."); + ret = WIMLIB_ERR_INVALID_IMAGE; + goto out_wimlib_free; + } + } else { + if (pwm->hdr.image_count != 1) { + ERROR("No image was specified, but the pipable WIM " + "did not contain exactly 1 image"); + ret = WIMLIB_ERR_INVALID_IMAGE; + goto out_wimlib_free; + } + image = 1; } /* Load the needed metadata resource. */ @@ -2652,7 +2715,7 @@ wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, goto out_wimlib_free; } - ret = read_pwm_stream_header(pwm, metadata_lte, false); + ret = read_pwm_stream_header(pwm, metadata_lte, 0); imd = pwm->image_metadata[i - 1]; imd->metadata_lte = metadata_lte; if (ret)