X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fextract.c;h=e80bbe94a4d2e65f0b23cc0b8ce4ee96c94d53fb;hp=bc769eec798d284f61ddee9f28fd006304b616bd;hb=8b709192cd2811b83c248fbe61ca4f11ee9de797;hpb=50407ddd59aa2d0906225a4526eaefbf6da8450f diff --git a/src/extract.c b/src/extract.c index bc769eec..e80bbe94 100644 --- a/src/extract.c +++ b/src/extract.c @@ -6,22 +6,20 @@ */ /* - * Copyright (C) 2012, 2013, 2014 Eric Biggers + * Copyright (C) 2012, 2013, 2014, 2015 Eric Biggers * - * This file is part of wimlib, a library for working with WIM files. + * This file is free software; you can redistribute it and/or modify it under + * the terms of the GNU Lesser General Public License as published by the Free + * Software Foundation; either version 3 of the License, or (at your option) any + * later version. * - * wimlib is free software; you can redistribute it and/or modify it under the - * terms of the GNU General Public License as published by the Free - * Software Foundation; either version 3 of the License, or (at your option) - * any later version. - * - * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY - * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR - * A PARTICULAR PURPOSE. See the GNU General Public License for more + * This file is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS + * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more * details. * - * You should have received a copy of the GNU General Public License - * along with wimlib; if not, see http://www.gnu.org/licenses/. + * You should have received a copy of the GNU Lesser General Public License + * along with this file; if not, see http://www.gnu.org/licenses/. */ /* @@ -41,42 +39,36 @@ # include "config.h" #endif +#include +#include +#include +#include + #include "wimlib/apply.h" +#include "wimlib/assert.h" +#include "wimlib/blob_table.h" #include "wimlib/dentry.h" #include "wimlib/encoding.h" #include "wimlib/endianness.h" #include "wimlib/error.h" -#include "wimlib/lookup_table.h" #include "wimlib/metadata.h" #include "wimlib/pathlist.h" #include "wimlib/paths.h" +#include "wimlib/pattern.h" #include "wimlib/reparse.h" #include "wimlib/resource.h" #include "wimlib/security.h" -#ifdef __WIN32__ -# include "wimlib/win32.h" /* for realpath() equivalent */ -#endif -#include "wimlib/xml.h" -#include "wimlib/wildcard.h" +#include "wimlib/unix_data.h" #include "wimlib/wim.h" +#include "wimlib/win32.h" /* for realpath() equivalent */ +#include "wimlib/xml.h" -#include -#include -#include -#include -#include - -#define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000 -#define WIMLIB_EXTRACT_FLAG_FROM_PIPE 0x40000000 -#define WIMLIB_EXTRACT_FLAG_IMAGEMODE 0x20000000 +#define WIMLIB_EXTRACT_FLAG_FROM_PIPE 0x80000000 +#define WIMLIB_EXTRACT_FLAG_IMAGEMODE 0x40000000 /* Keep in sync with wimlib.h */ #define WIMLIB_EXTRACT_MASK_PUBLIC \ (WIMLIB_EXTRACT_FLAG_NTFS | \ - WIMLIB_EXTRACT_FLAG_HARDLINK | \ - WIMLIB_EXTRACT_FLAG_SYMLINK | \ - WIMLIB_EXTRACT_FLAG_VERBOSE | \ - WIMLIB_EXTRACT_FLAG_SEQUENTIAL | \ WIMLIB_EXTRACT_FLAG_UNIX_DATA | \ WIMLIB_EXTRACT_FLAG_NO_ACLS | \ WIMLIB_EXTRACT_FLAG_STRICT_ACLS | \ @@ -88,1078 +80,182 @@ WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS | \ WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES | \ WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS | \ - WIMLIB_EXTRACT_FLAG_RESUME | \ - WIMLIB_EXTRACT_FLAG_FILE_ORDER | \ WIMLIB_EXTRACT_FLAG_GLOB_PATHS | \ WIMLIB_EXTRACT_FLAG_STRICT_GLOB | \ WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES | \ WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE | \ - WIMLIB_EXTRACT_FLAG_WIMBOOT) - -static bool -dentry_in_list(const struct wim_dentry *dentry) -{ - return dentry->extraction_list.next != NULL; -} - -static inline bool -is_linked_extraction(const struct apply_ctx *ctx) -{ - return 0 != (ctx->extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK | - WIMLIB_EXTRACT_FLAG_SYMLINK)); -} - -static inline bool -can_extract_named_data_streams(const struct apply_ctx *ctx) -{ - return ctx->supported_features.named_data_streams && - !is_linked_extraction(ctx); -} -/* Inform library user of progress of stream extraction following the successful - * extraction of a copy of the stream specified by @lte. */ -static void -update_extract_progress(struct apply_ctx *ctx, - const struct wim_lookup_table_entry *lte) -{ - wimlib_progress_func_t progress_func = ctx->progress_func; - union wimlib_progress_info *progress = &ctx->progress; - - progress->extract.completed_bytes += lte->size; - if (progress_func && - progress->extract.completed_bytes >= ctx->next_progress) - { - progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, progress); - if (progress->extract.completed_bytes >= - progress->extract.total_bytes) - { - ctx->next_progress = ~0ULL; - } else { - ctx->next_progress += progress->extract.total_bytes / 128; - if (ctx->next_progress > progress->extract.total_bytes) - ctx->next_progress = progress->extract.total_bytes; - } - } -} - -#ifndef __WIN32__ -/* Extract a symbolic link (not directly as reparse data), handling fixing up - * the target of absolute symbolic links and updating the extract progress. - * - * @inode must specify the WIM inode for a symbolic link or junction reparse - * point. - * - * @lte_override overrides the resource used as the reparse data for the - * symbolic link. */ -static int -extract_symlink(const tchar *path, struct apply_ctx *ctx, - struct wim_inode *inode, - struct wim_lookup_table_entry *lte_override) -{ - ssize_t bufsize = ctx->ops->path_max; - tchar target[bufsize]; - tchar *buf = target; - tchar *fixed_target; - ssize_t sret; - int ret; - - /* If absolute symbolic link fixups requested, reserve space in the link - * target buffer for the absolute path of the target directory. */ - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) - { - buf += ctx->realtarget_nchars; - bufsize -= ctx->realtarget_nchars; - } - - /* Translate the WIM inode's reparse data into the link target. */ - sret = wim_inode_readlink(inode, buf, bufsize - 1, lte_override); - if (sret < 0) { - errno = -sret; - return WIMLIB_ERR_READLINK; - } - buf[sret] = '\0'; - - if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) && - buf[0] == '/') - { - /* Fix absolute symbolic link target to point into the - * actual extraction destination. */ - tmemcpy(target, ctx->realtarget, ctx->realtarget_nchars); - fixed_target = target; - } else { - /* Keep same link target. */ - fixed_target = buf; - } - - /* Call into the apply_operations to create the symbolic link. */ - DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"", - path, fixed_target); - ret = ctx->ops->create_symlink(fixed_target, path, ctx); - if (ret) { - ERROR_WITH_ERRNO("Failed to create symlink " - "\"%"TS"\" => \"%"TS"\"", path, fixed_target); - return ret; - } - - /* Account for reparse data consumed. */ - update_extract_progress(ctx, - (lte_override ? lte_override : - inode_unnamed_lte_resolved(inode))); - return 0; -} -#endif /* !__WIN32__ */ - -/* Create a file, directory, or symbolic link. */ -static int -extract_inode(const tchar *path, struct apply_ctx *ctx, struct wim_inode *inode) -{ - int ret; - -#ifndef __WIN32__ - if (ctx->supported_features.symlink_reparse_points && - !ctx->supported_features.reparse_points && - inode_is_symlink(inode)) - { - ret = extract_symlink(path, ctx, inode, NULL); - } else -#endif /* !__WIN32__ */ - if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) { - ret = ctx->ops->create_directory(path, ctx, &inode->extract_cookie); - if (ret) { - ERROR_WITH_ERRNO("Failed to create the directory " - "\"%"TS"\"", path); - } - } else if ((inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) && - ctx->ops->extract_encrypted_stream_creates_file && - ctx->supported_features.encrypted_files) { - ret = ctx->ops->extract_encrypted_stream( - path, inode_unnamed_lte_resolved(inode), ctx); - if (ret) { - ERROR_WITH_ERRNO("Failed to create and extract " - "encrypted file \"%"TS"\"", path); - } - } else { - ret = ctx->ops->create_file(path, ctx, &inode->extract_cookie); - if (ret) { - ERROR_WITH_ERRNO("Failed to create the file " - "\"%"TS"\"", path); - } - } - return ret; -} - -static int -extract_hardlink(const tchar *oldpath, const tchar *newpath, - struct apply_ctx *ctx) -{ - int ret; - - DEBUG("Creating hardlink \"%"TS"\" => \"%"TS"\"", newpath, oldpath); - ret = ctx->ops->create_hardlink(oldpath, newpath, ctx); - if (ret) { - ERROR_WITH_ERRNO("Failed to create hardlink " - "\"%"TS"\" => \"%"TS"\"", - newpath, oldpath); - } - return ret; -} - -#ifdef __WIN32__ -static int -try_extract_rpfix(u8 *rpbuf, - u16 *rpbuflen_p, - const wchar_t *extract_root_realpath, - unsigned extract_root_realpath_nchars) -{ - struct reparse_data rpdata; - wchar_t *target; - size_t target_nchars; - size_t stripped_nchars; - wchar_t *stripped_target; - wchar_t stripped_target_nchars; - int ret; - - utf16lechar *new_target; - utf16lechar *new_print_name; - size_t new_target_nchars; - size_t new_print_name_nchars; - utf16lechar *p; - - ret = parse_reparse_data(rpbuf, *rpbuflen_p, &rpdata); - if (ret) - return ret; - - if (extract_root_realpath[0] == L'\0' || - extract_root_realpath[1] != L':' || - extract_root_realpath[2] != L'\\') - return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED; - - ret = parse_substitute_name(rpdata.substitute_name, - rpdata.substitute_name_nbytes, - rpdata.rptag); - if (ret < 0) - return 0; - stripped_nchars = ret; - target = rpdata.substitute_name; - target_nchars = rpdata.substitute_name_nbytes / sizeof(utf16lechar); - stripped_target = target + stripped_nchars; - stripped_target_nchars = target_nchars - stripped_nchars; - - new_target = alloca((6 + extract_root_realpath_nchars + - stripped_target_nchars) * sizeof(utf16lechar)); - - p = new_target; - if (stripped_nchars == 6) { - /* Include \??\ prefix if it was present before */ - p = wmempcpy(p, L"\\??\\", 4); - } - - /* Print name excludes the \??\ if present. */ - new_print_name = p; - if (stripped_nchars != 0) { - /* Get drive letter from real path to extract root, if a drive - * letter was present before. */ - *p++ = extract_root_realpath[0]; - *p++ = extract_root_realpath[1]; - } - /* Copy the rest of the extract root */ - p = wmempcpy(p, extract_root_realpath + 2, extract_root_realpath_nchars - 2); - - /* Append the stripped target */ - p = wmempcpy(p, stripped_target, stripped_target_nchars); - new_target_nchars = p - new_target; - new_print_name_nchars = p - new_print_name; - - if (new_target_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE || - new_print_name_nchars * sizeof(utf16lechar) >= REPARSE_POINT_MAX_SIZE) - return WIMLIB_ERR_REPARSE_POINT_FIXUP_FAILED; - - rpdata.substitute_name = new_target; - rpdata.substitute_name_nbytes = new_target_nchars * sizeof(utf16lechar); - rpdata.print_name = new_print_name; - rpdata.print_name_nbytes = new_print_name_nchars * sizeof(utf16lechar); - return make_reparse_buffer(&rpdata, rpbuf, rpbuflen_p); -} -#endif /* __WIN32__ */ - -/* Set reparse data on extracted file or directory that has - * FILE_ATTRIBUTE_REPARSE_POINT set. */ -static int -extract_reparse_data(const tchar *path, struct apply_ctx *ctx, - struct wim_inode *inode, - struct wim_lookup_table_entry *lte_override) -{ - int ret; - u8 rpbuf[REPARSE_POINT_MAX_SIZE]; - u16 rpbuflen; - - ret = wim_inode_get_reparse_data(inode, rpbuf, &rpbuflen, lte_override); - if (ret) - goto error; - -#ifdef __WIN32__ - /* Fix up target of absolute symbolic link or junction points so - * that they point into the actual extraction target. */ - if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) && - (inode->i_reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK || - inode->i_reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT) && - !inode->i_not_rpfixed) - { - ret = try_extract_rpfix(rpbuf, &rpbuflen, ctx->realtarget, - ctx->realtarget_nchars); - if (ret && !(ctx->extract_flags & - WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS)) - { - WARNING("Reparse point fixup of \"%"TS"\" " - "failed", path); - ret = 0; - } - if (ret) - goto error; - } -#endif - - ret = ctx->ops->set_reparse_data(path, rpbuf, rpbuflen, ctx); - - /* On Windows, the SeCreateSymbolicLink privilege is required to create - * symbolic links. To be more friendly towards non-Administrator users, - * we merely warn the user if symbolic links cannot be created due to - * insufficient permissions or privileges, unless - * WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS was provided. */ -#ifdef __WIN32__ - if (ret && inode_is_symlink(inode) && - (errno == EACCES || errno == EPERM) && - !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS)) - { - WARNING("Can't set reparse data on \"%"TS"\": " - "Access denied!\n" - " You may be trying to " - "extract a symbolic link without the\n" - " SeCreateSymbolicLink privilege, " - "which by default non-Administrator\n" - " accounts do not have.", - path); - ret = 0; - } -#endif - if (ret) - goto error; - - /* Account for reparse data consumed. */ - update_extract_progress(ctx, - (lte_override ? lte_override : - inode_unnamed_lte_resolved(inode))); - return 0; - -error: - ERROR_WITH_ERRNO("Failed to set reparse data on \"%"TS"\"", path); - return ret; -} + WIMLIB_EXTRACT_FLAG_WIMBOOT | \ + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K | \ + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K | \ + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K | \ + WIMLIB_EXTRACT_FLAG_COMPACT_LZX \ + ) -/* - * Extract zero or more streams to a file. - * - * This function operates slightly differently depending on whether @lte_spec is - * NULL or not. When @lte_spec is NULL, the behavior is to extract the default - * file contents (unnamed stream), and, if named data streams are supported in - * the extract mode and volume, any named data streams. When @lte_spec is not - * NULL, the behavior is to extract only all copies of the stream @lte_spec, and - * in addition use @lte_spec to set the reparse data or create the symbolic link - * if appropriate. - * - * @path - * Path to file to extract (as can be passed to apply_operations - * functions). - * @ctx - * Apply context. - * @dentry - * WIM dentry that corresponds to the file being extracted. - * @lte_spec - * If non-NULL, specifies the lookup table entry for a stream to extract, - * and only that stream will be extracted (although there may be more than - * one instance of it). - * @lte_override - * Used only if @lte_spec != NULL; it is passed to the extraction functions - * rather than @lte_spec, allowing the location of the stream to be - * overridden. (This is used when the WIM is being read from a nonseekable - * file, such as a pipe, when streams need to be used more than once; each - * such stream is extracted to a temporary file.) - */ -static int -extract_streams(const tchar *path, struct apply_ctx *ctx, - struct wim_dentry *dentry, - struct wim_lookup_table_entry *lte_spec, - struct wim_lookup_table_entry *lte_override) +/* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or + * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA. */ +int +do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg) { - struct wim_inode *inode = dentry->d_inode; - struct wim_lookup_table_entry *lte; - file_spec_t file_spec; - int ret; - - if (dentry->was_linked) - return 0; - -#ifdef ENABLE_DEBUG - if (lte_spec) { - char sha1_str[100]; - char *p = sha1_str; - for (unsigned i = 0; i < SHA1_HASH_SIZE; i++) - p += sprintf(p, "%02x", lte_override->hash[i]); - DEBUG("Extracting stream SHA1=%s to \"%"TS"\"", - sha1_str, path, inode->i_ino); - } else { - DEBUG("Extracting streams to \"%"TS"\"", path, inode->i_ino); - } -#endif - - if (ctx->ops->uses_cookies) - file_spec.cookie = inode->extract_cookie; - else - file_spec.path = path; - - /* Unnamed data stream. */ - lte = inode_unnamed_lte_resolved(inode); - if (lte && (!lte_spec || lte == lte_spec)) { - if (lte_spec) - lte = lte_override; - if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY | - FILE_ATTRIBUTE_REPARSE_POINT))) - { - if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED && - ctx->supported_features.encrypted_files) { - if (!ctx->ops->extract_encrypted_stream_creates_file) { - ret = ctx->ops->extract_encrypted_stream( - path, lte, ctx); - if (ret) - goto error; - } - } else { - ret = ctx->ops->extract_unnamed_stream( - file_spec, lte, ctx, - dentry); - if (ret) - goto error; - } - update_extract_progress(ctx, lte); - } - else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) - { - ret = 0; - if (ctx->supported_features.reparse_points) - ret = extract_reparse_data(path, ctx, inode, lte); - #ifndef __WIN32__ - else if ((inode_is_symlink(inode) && - ctx->supported_features.symlink_reparse_points)) - ret = extract_symlink(path, ctx, inode, lte); - #endif - if (ret) - return ret; - } - } - - /* Named data streams. */ - if (can_extract_named_data_streams(ctx)) { - for (u16 i = 0; i < inode->i_num_ads; i++) { - struct wim_ads_entry *entry = &inode->i_ads_entries[i]; - - if (!ads_entry_is_named_stream(entry)) - continue; - lte = entry->lte; - if (!lte) - continue; - if (lte_spec && lte_spec != lte) - continue; - if (lte_spec) - lte = lte_override; - ret = ctx->ops->extract_named_stream(file_spec, entry->stream_name, - entry->stream_name_nbytes / 2, - lte, ctx); - if (ret) - goto error; - update_extract_progress(ctx, lte); - } - } - return 0; - -error: - ERROR_WITH_ERRNO("Failed to extract data of \"%"TS"\"", path); - return ret; + ctx->count_until_file_progress = 500; /* Arbitrary value to limit calls */ + return extract_progress(ctx, msg); } -/* Set attributes on an extracted file or directory if supported by the - * extraction mode. */ static int -extract_file_attributes(const tchar *path, struct apply_ctx *ctx, - struct wim_dentry *dentry, unsigned pass) +start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg) { - int ret; - - if (ctx->ops->set_file_attributes && - !(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES) && - !(dentry == ctx->target_dentry && ctx->root_dentry_is_special)) { - u32 attributes = dentry->d_inode->i_attributes; - - /* Clear unsupported attributes. */ - attributes &= ctx->supported_attributes_mask; - - if ((attributes & FILE_ATTRIBUTE_DIRECTORY && - !ctx->supported_features.encrypted_directories) || - (!(attributes & FILE_ATTRIBUTE_DIRECTORY) && - !ctx->supported_features.encrypted_files)) - { - attributes &= ~FILE_ATTRIBUTE_ENCRYPTED; - } - - if (attributes == 0) - attributes = FILE_ATTRIBUTE_NORMAL; - - ret = ctx->ops->set_file_attributes(path, attributes, ctx, pass); - if (ret) { - ERROR_WITH_ERRNO("Failed to set attributes on " - "\"%"TS"\"", path); - return ret; - } - } - return 0; + ctx->progress.extract.current_file_count = 0; + ctx->progress.extract.end_file_count = end_file_count; + return do_file_extract_progress(ctx, msg); } - -/* Set or remove the short (DOS) name on an extracted file or directory if - * supported by the extraction mode. Since DOS names are unimportant and it's - * easy to run into problems setting them on Windows (SetFileShortName() - * requires SE_RESTORE privilege, which only the Administrator can request, and - * also requires DELETE access to the file), failure is ignored unless - * WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES is set. */ -static int -extract_short_name(const tchar *path, struct apply_ctx *ctx, - struct wim_dentry *dentry) +int +start_file_structure_phase(struct apply_ctx *ctx, u64 end_file_count) { - int ret; - - /* The root of the dentry tree being extracted may not be extracted to - * its original name, so its short name should be ignored. */ - if (dentry == ctx->target_dentry) - return 0; - - if (ctx->supported_features.short_names) { - ret = ctx->ops->set_short_name(path, - dentry->short_name, - dentry->short_name_nbytes / 2, - ctx); - if (ret && (ctx->extract_flags & - WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES)) - { - ERROR_WITH_ERRNO("Failed to set short name of " - "\"%"TS"\"", path); - return ret; - } - } - return 0; + return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE); } -/* Set security descriptor, UNIX data, or neither on an extracted file, taking - * into account the current extraction mode and flags. */ -static int -extract_security(const tchar *path, struct apply_ctx *ctx, - struct wim_dentry *dentry) +int +start_file_metadata_phase(struct apply_ctx *ctx, u64 end_file_count) { - int ret; - struct wim_inode *inode = dentry->d_inode; - - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) - return 0; - - if ((ctx->target_dentry == dentry) && ctx->root_dentry_is_special) - return 0; - -#ifndef __WIN32__ - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) { - struct wimlib_unix_data data; - - ret = inode_get_unix_data(inode, &data, NULL); - if (ret < 0) - ret = 0; - else if (ret == 0) - ret = ctx->ops->set_unix_data(path, &data, ctx); - if (ret) { - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) { - ERROR_WITH_ERRNO("Failed to set UNIX owner, " - "group, and/or mode on " - "\"%"TS"\"", path); - return ret; - } else { - WARNING_WITH_ERRNO("Failed to set UNIX owner, " - "group, and/or/mode on " - "\"%"TS"\"", path); - } - } - } - else -#endif /* __WIN32__ */ - if (ctx->supported_features.security_descriptors && - inode->i_security_id != -1) - { - const struct wim_security_data *sd; - const u8 *desc; - size_t desc_size; - - sd = wim_const_security_data(ctx->wim); - desc = sd->descriptors[inode->i_security_id]; - desc_size = sd->sizes[inode->i_security_id]; - - ret = ctx->ops->set_security_descriptor(path, desc, - desc_size, ctx); - if (ret) { - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) { - ERROR_WITH_ERRNO("Failed to set security " - "descriptor on \"%"TS"\"", path); - return ret; - } else { - #if 0 - if (errno != EACCES) { - WARNING_WITH_ERRNO("Failed to set " - "security descriptor " - "on \"%"TS"\"", path); - } - #endif - ctx->no_security_descriptors++; - } - } - } - return 0; + return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA); } -/* Set timestamps on an extracted file. Failure is warning-only unless - * WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS is set. */ static int -extract_timestamps(const tchar *path, struct apply_ctx *ctx, - struct wim_dentry *dentry) +end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg) { - struct wim_inode *inode = dentry->d_inode; - int ret; - - if ((ctx->target_dentry == dentry) && ctx->root_dentry_is_special) - return 0; - - if (ctx->ops->set_timestamps) { - ret = ctx->ops->set_timestamps(path, - inode->i_creation_time, - inode->i_last_write_time, - inode->i_last_access_time, - ctx); - if (ret) { - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) { - ERROR_WITH_ERRNO("Failed to set timestamps " - "on \"%"TS"\"", path); - return ret; - } else { - WARNING_WITH_ERRNO("Failed to set timestamps " - "on \"%"TS"\"", path); - } - } - } - return 0; + ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count; + return do_file_extract_progress(ctx, msg); } -/* Check whether the extraction of a dentry should be skipped completely. */ -static bool -dentry_is_supported(struct wim_dentry *dentry, - const struct wim_features *supported_features) +int +end_file_structure_phase(struct apply_ctx *ctx) { - struct wim_inode *inode = dentry->d_inode; - - if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) { - return supported_features->reparse_points || - (inode_is_symlink(inode) && - supported_features->symlink_reparse_points); - } - if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) { - if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) - return supported_features->encrypted_directories != 0; - else - return supported_features->encrypted_files != 0; - } - return true; + return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE); } -/* Given a WIM dentry to extract, build the path to which to extract it, in the - * format understood by the callbacks in the apply_operations being used. - * - * Write the resulting path into @path, which must have room for at least - * ctx->ops->path_max characters. - * - * Return %true if successful; %false if this WIM dentry doesn't actually need - * to be extracted or if the calculated path exceeds ctx->ops->max_path - * characters. - * - * This function clobbers the tmp_list member of @dentry and its ancestors up - * until the extraction root. */ -static bool -build_extraction_path(tchar path[], struct wim_dentry *dentry, - const struct apply_ctx *ctx) +int +end_file_metadata_phase(struct apply_ctx *ctx) { - size_t path_nchars; - LIST_HEAD(ancestor_list); - tchar *p = path; - const tchar *target_prefix; - size_t target_prefix_nchars; - struct wim_dentry *d; - - path_nchars = ctx->ops->path_prefix_nchars; - - if (ctx->ops->requires_realtarget_in_paths) { - target_prefix = ctx->realtarget; - target_prefix_nchars = ctx->realtarget_nchars; - } else if (ctx->ops->requires_target_in_paths) { - target_prefix = ctx->target; - target_prefix_nchars = ctx->target_nchars; - } else { - target_prefix = NULL; - target_prefix_nchars = 0; - } - path_nchars += target_prefix_nchars; - - for (d = dentry; d != ctx->target_dentry; d = d->parent) { - if (!dentry_in_list(d)) - break; - - path_nchars += d->extraction_name_nchars + 1; - list_add(&d->tmp_list, &ancestor_list); - } - - path_nchars++; /* null terminator */ - - if (path_nchars > ctx->ops->path_max) { - WARNING("\"%"TS"\": Path too long to extract", - dentry_full_path(dentry)); - return false; - } - - p = tmempcpy(p, ctx->ops->path_prefix, ctx->ops->path_prefix_nchars); - p = tmempcpy(p, target_prefix, target_prefix_nchars); - list_for_each_entry(d, &ancestor_list, tmp_list) { - *p++ = ctx->ops->path_separator; - p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars); - } - *p++ = T('\0'); - wimlib_assert(p - path == path_nchars); - return true; -} - -static unsigned -get_num_path_components(const tchar *path, tchar path_separator) -{ - unsigned num_components = 0; -#ifdef __WIN32__ - /* Ignore drive letter. */ - if (path[0] != L'\0' && path[1] == L':') - path += 2; -#endif - - while (*path) { - while (*path == path_separator) - path++; - if (*path) - num_components++; - while (*path && *path != path_separator) - path++; - } - return num_components; -} - -static int -extract_multiimage_symlink(const tchar *oldpath, const tchar *newpath, - struct apply_ctx *ctx, struct wim_dentry *dentry) -{ - size_t num_raw_path_components; - const struct wim_dentry *d; - size_t num_target_path_components; - tchar *p; - const tchar *p_old; - int ret; - - num_raw_path_components = 0; - for (d = dentry; d != ctx->target_dentry; d = d->parent) - num_raw_path_components++; - - if (ctx->ops->requires_realtarget_in_paths) - num_target_path_components = get_num_path_components(ctx->realtarget, - ctx->ops->path_separator); - else if (ctx->ops->requires_target_in_paths) - num_target_path_components = get_num_path_components(ctx->target, - ctx->ops->path_separator); - else - num_target_path_components = 0; - - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) { - wimlib_assert(num_target_path_components > 0); - num_raw_path_components++; - num_target_path_components--; - } - - p_old = oldpath + ctx->ops->path_prefix_nchars; -#ifdef __WIN32__ - if (p_old[0] != L'\0' && p_old[1] == ':') - p_old += 2; -#endif - while (*p_old == ctx->ops->path_separator) - p_old++; - while (--num_target_path_components) { - while (*p_old != ctx->ops->path_separator) - p_old++; - while (*p_old == ctx->ops->path_separator) - p_old++; - } - - tchar symlink_target[tstrlen(p_old) + 3 * num_raw_path_components + 1]; - - p = &symlink_target[0]; - while (num_raw_path_components--) { - *p++ = '.'; - *p++ = '.'; - *p++ = ctx->ops->path_separator; - } - tstrcpy(p, p_old); - DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"", - newpath, symlink_target); - ret = ctx->ops->create_symlink(symlink_target, newpath, ctx); - if (ret) { - ERROR_WITH_ERRNO("Failed to create symlink " - "\"%"TS"\" => \"%"TS"\"", - newpath, symlink_target); - } - return ret; + return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA); } -/* Create the "skeleton" of an extracted file or directory. Don't yet extract - * data streams, reparse data (including symbolic links), timestamps, and - * security descriptors. Basically, everything that doesn't require reading - * non-metadata resources from the WIM file and isn't delayed until the final - * pass. */ -static int -do_dentry_extract_skeleton(tchar path[], struct wim_dentry *dentry, - struct apply_ctx *ctx) -{ - struct wim_inode *inode = dentry->d_inode; - int ret; - const tchar *oldpath; - - if (unlikely(is_linked_extraction(ctx))) { - struct wim_lookup_table_entry *unnamed_lte; - - unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode); - if (unnamed_lte && unnamed_lte->extracted_file) { - oldpath = unnamed_lte->extracted_file; - if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) - goto hardlink; - else - goto symlink; - } - } - - /* Create hard link if this dentry corresponds to an already-extracted - * inode. */ - if (inode->i_extracted_file) { - oldpath = inode->i_extracted_file; - goto hardlink; - } - - /* Skip symlinks unless they can be extracted as reparse points rather - * than created directly. */ - if (inode_is_symlink(inode) && !ctx->supported_features.reparse_points) - return 0; - - /* Create this file or directory unless it's the extraction root, which - * was already created if necessary. */ - if (dentry != ctx->target_dentry) { - ret = extract_inode(path, ctx, inode); - if (ret) - return ret; - } - - /* Create empty named data streams. */ - if (can_extract_named_data_streams(ctx)) { - for (u16 i = 0; i < inode->i_num_ads; i++) { - file_spec_t file_spec; - struct wim_ads_entry *entry = &inode->i_ads_entries[i]; - - if (!ads_entry_is_named_stream(entry)) - continue; - if (entry->lte) - continue; - if (ctx->ops->uses_cookies) - file_spec.cookie = inode->extract_cookie; - else - file_spec.path = path; - ret = ctx->ops->extract_named_stream(file_spec, - entry->stream_name, - entry->stream_name_nbytes / 2, - entry->lte, ctx); - if (ret) { - ERROR_WITH_ERRNO("\"%"TS"\": failed to create " - "empty named data stream", - path); - return ret; - } - } - } - - /* Set file attributes (if supported). */ - ret = extract_file_attributes(path, ctx, dentry, 0); - if (ret) - return ret; - - /* Set or remove file short name (if supported). */ - ret = extract_short_name(path, ctx, dentry); - if (ret) - return ret; +#define PWM_FOUND_WIM_HDR (-1) - /* If inode has multiple links and hard links are supported in this - * extraction mode and volume, save the path to the extracted file in - * case it's needed to create a hard link. */ - if (unlikely(is_linked_extraction(ctx))) { - struct wim_lookup_table_entry *unnamed_lte; - - unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode); - if (unnamed_lte) { - unnamed_lte->extracted_file = TSTRDUP(path); - if (!unnamed_lte->extracted_file) - return WIMLIB_ERR_NOMEM; - } - } else if (inode->i_nlink > 1 && ctx->supported_features.hard_links) { - inode->i_extracted_file = TSTRDUP(path); - if (!inode->i_extracted_file) - return WIMLIB_ERR_NOMEM; - } - return 0; - -symlink: - ret = extract_multiimage_symlink(oldpath, path, ctx, dentry); - if (ret) - return ret; - dentry->was_linked = 1; - return 0; - -hardlink: - ret = extract_hardlink(oldpath, path, ctx); - if (ret) - return ret; - dentry->was_linked = 1; - return 0; -} - -/* This is a wrapper around do_dentry_extract_skeleton() that handles building - * the path, doing short name reordering. This is also idempotent; dentries - * already processed have skeleton_extracted set and no action is taken. See - * apply_operations.requires_short_name_reordering for more details about short - * name reordering. */ +/* Read the header for a blob in a pipable WIM. If @pwm_hdr_ret is not NULL, + * also look for a pipable WIM header and return PWM_FOUND_WIM_HDR if found. */ static int -dentry_extract_skeleton(struct wim_dentry *dentry, struct apply_ctx *ctx) +read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE], + struct wim_reshdr *reshdr_ret, + struct wim_header_disk *pwm_hdr_ret) { - tchar path[ctx->ops->path_max]; - struct wim_dentry *orig_dentry; - struct wim_dentry *other_dentry; int ret; + struct pwm_blob_hdr blob_hdr; + u64 magic; - if (dentry->skeleton_extracted) - return 0; - - orig_dentry = NULL; - if (ctx->supported_features.short_names - && ctx->ops->requires_short_name_reordering - && !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 - && dentry_in_list(other_dentry)) - { - 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; - ret = do_dentry_extract_skeleton(path, dentry, ctx); - if (ret) - return ret; + ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr)); + if (unlikely(ret)) + goto read_error; - dentry->skeleton_extracted = 1; + magic = le64_to_cpu(blob_hdr.magic); - if (orig_dentry) { - dentry = orig_dentry; - orig_dentry = NULL; - goto again; + if (magic == PWM_MAGIC && pwm_hdr_ret != NULL) { + memcpy(pwm_hdr_ret, &blob_hdr, sizeof(blob_hdr)); + ret = full_read(&pwm->in_fd, + (u8 *)pwm_hdr_ret + sizeof(blob_hdr), + sizeof(*pwm_hdr_ret) - sizeof(blob_hdr)); + if (unlikely(ret)) + goto read_error; + return PWM_FOUND_WIM_HDR; } - dentry->d_inode->i_dos_name_extracted = 1; - return 0; -} - -/* Create a file or directory, then immediately extract all streams. The WIM - * may not be read sequentially by this function. */ -static int -dentry_extract(struct wim_dentry *dentry, struct apply_ctx *ctx) -{ - tchar path[ctx->ops->path_max]; - int ret; - - 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); -} - -/* Finish extracting a file, directory, or symbolic link by setting file - * security and timestamps. */ -static int -dentry_extract_final(struct wim_dentry *dentry, struct apply_ctx *ctx) -{ - int ret; - tchar path[ctx->ops->path_max]; - - if (!build_extraction_path(path, dentry, ctx)) - return 0; - - ret = extract_security(path, ctx, dentry); - if (ret) - return ret; - - if (ctx->ops->requires_final_set_attributes_pass) { - /* Set file attributes (if supported). */ - ret = extract_file_attributes(path, ctx, dentry, 1); - if (ret) - return ret; + if (unlikely(magic != PWM_BLOB_MAGIC)) { + ERROR("Data read on pipe is invalid (expected blob header)"); + return WIMLIB_ERR_INVALID_PIPABLE_WIM; } - return extract_timestamps(path, ctx, dentry); -} + copy_hash(hash_ret, blob_hdr.hash); -static int -extract_structure(struct list_head *dentry_list, struct apply_ctx *ctx) -{ - struct wim_dentry *dentry; - int ret; + reshdr_ret->size_in_wim = 0; /* Not available */ + reshdr_ret->flags = le32_to_cpu(blob_hdr.flags); + reshdr_ret->offset_in_wim = pwm->in_fd.offset; + reshdr_ret->uncompressed_size = le64_to_cpu(blob_hdr.uncompressed_size); - list_for_each_entry(dentry, dentry_list, extraction_list) { - ret = dentry_extract_skeleton(dentry, ctx); - if (ret) - return ret; + if (unlikely(reshdr_ret->uncompressed_size == 0)) { + ERROR("Data read on pipe is invalid (resource is of 0 size)"); + return WIMLIB_ERR_INVALID_PIPABLE_WIM; } - return 0; -} - -static int -extract_dir_structure(struct list_head *dentry_list, struct apply_ctx *ctx) -{ - struct wim_dentry *dentry; - int ret; - list_for_each_entry(dentry, dentry_list, extraction_list) { - if (dentry_is_directory(dentry)) { - ret = dentry_extract_skeleton(dentry, ctx); - if (ret) - return ret; - } - } return 0; -} -static int -extract_dentries(struct list_head *dentry_list, struct apply_ctx *ctx) -{ - struct wim_dentry *dentry; - int ret; - - list_for_each_entry(dentry, dentry_list, extraction_list) { - ret = dentry_extract(dentry, ctx); - if (ret) - return ret; - } - return 0; +read_error: + if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE) + ERROR("The pipe ended before all needed data was sent!"); + else + ERROR_WITH_ERRNO("Error reading pipable WIM from pipe"); + return ret; } static int -extract_final_metadata(struct list_head *dentry_list, struct apply_ctx *ctx) +read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs) { - struct wim_dentry *dentry; int ret; + u8 hash[SHA1_HASH_SIZE]; + struct wim_reshdr reshdr; + struct wim_header_disk pwm_hdr; + struct wim_resource_descriptor rdesc; + struct blob_descriptor *blob; + + copy_guid(ctx->progress.extract.guid, ctx->wim->hdr.guid); + ctx->progress.extract.part_number = ctx->wim->hdr.part_number; + ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts; + ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN); + if (ret) + return ret; + + while (ctx->num_blobs_remaining) { + + ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr); + + if (ret == PWM_FOUND_WIM_HDR) { + u16 part_number = le16_to_cpu(pwm_hdr.part_number); + u16 total_parts = le16_to_cpu(pwm_hdr.total_parts); + + if (part_number == ctx->progress.extract.part_number && + total_parts == ctx->progress.extract.total_parts && + guids_equal(pwm_hdr.guid, ctx->progress.extract.guid)) + continue; + + copy_guid(ctx->progress.extract.guid, pwm_hdr.guid); + ctx->progress.extract.part_number = part_number; + ctx->progress.extract.total_parts = total_parts; + ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN); + if (ret) + return ret; + + continue; + } - list_for_each_entry_reverse(dentry, dentry_list, extraction_list) { - ret = dentry_extract_final(dentry, ctx); if (ret) return ret; + + if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA) + && (blob = lookup_blob(ctx->wim->blob_table, hash)) + && (blob->out_refcnt)) + { + wim_reshdr_to_desc_and_blob(&reshdr, ctx->wim, &rdesc, blob); + ret = read_blob_with_sha1(blob, cbs); + blob_unset_is_located_in_wim_resource(blob); + if (ret) + return ret; + ctx->num_blobs_remaining--; + } else { + wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc); + ret = skip_wim_resource(&rdesc); + if (ret) + return ret; + } } + return 0; } @@ -1170,12 +266,12 @@ static int create_temporary_file(struct filedes *fd_ret, tchar **name_ret) { tchar *name; - int raw_fd; int open_flags; + int raw_fd; retry: name = ttempnam(NULL, T("wimlib")); - if (name == NULL) { + if (!name) { ERROR_WITH_ERRNO("Failed to create temporary filename"); return WIMLIB_ERR_NOMEM; } @@ -1191,7 +287,8 @@ retry: FREE(name); goto retry; } - ERROR_WITH_ERRNO("Failed to open temporary file \"%"TS"\"", name); + ERROR_WITH_ERRNO("Failed to create temporary file " + "\"%"TS"\"", name); FREE(name); return WIMLIB_ERR_OPEN; } @@ -1201,392 +298,154 @@ retry: return 0; } -/* Extract all instances of the stream @lte that are being extracted in this - * call of extract_tree(), but actually read the stream data from @lte_override. - */ static int -extract_stream_instances(struct wim_lookup_table_entry *lte, - struct wim_lookup_table_entry *lte_override, - struct apply_ctx *ctx) -{ - struct wim_dentry **lte_dentries; - tchar path[ctx->ops->path_max]; - size_t i; - int ret; - - if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries)) - lte_dentries = lte->inline_lte_dentries; - else - lte_dentries = lte->lte_dentries; - - for (i = 0; i < lte->out_refcnt; i++) { - struct wim_dentry *dentry = lte_dentries[i]; - - if (dentry->tmp_flag) - continue; - if (!build_extraction_path(path, dentry, ctx)) - continue; - ret = extract_streams(path, ctx, dentry, lte, lte_override); - if (ret) - goto out_clear_tmp_flags; - dentry->tmp_flag = 1; - } - ret = 0; -out_clear_tmp_flags: - for (i = 0; i < lte->out_refcnt; i++) - lte_dentries[i]->tmp_flag = 0; - return ret; -} - -/* Determine whether the specified stream needs to be extracted to a temporary - * file or not. - * - * @lte->out_refcnt specifies the number of instances of this stream that must - * be extracted. - * - * @is_partial_res is %true if this stream is just one of multiple in a single - * WIM resource being extracted. */ -static bool -need_tmpfile_to_extract(struct wim_lookup_table_entry *lte, - bool is_partial_res) +begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx) { - /* Temporary file is always required when reading a partial resource, - * since in that case we retrieve all the contained streams in one pass. - * */ - if (is_partial_res) - return true; - - /* Otherwise we don't need a temporary file if only a single instance of - * the stream is needed. */ - if (lte->out_refcnt == 1) - return false; + struct apply_ctx *ctx = _ctx; - wimlib_assert(lte->out_refcnt >= 2); + ctx->cur_blob = blob; + ctx->cur_blob_offset = 0; - /* We also don't need a temporary file if random access to the stream is - * allowed. */ - if (lte->resource_location != RESOURCE_IN_WIM || - filedes_is_seekable(<e->rspec->wim->in_fd)) - return false; + if (unlikely(blob->out_refcnt > MAX_OPEN_FILES)) + return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name); - return true; + return call_begin_blob(blob, ctx->saved_cbs); } static int -begin_extract_stream(struct wim_lookup_table_entry *lte, - u32 flags, void *_ctx) +extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx) { struct apply_ctx *ctx = _ctx; + union wimlib_progress_info *progress = &ctx->progress; int ret; - if (flags & BEGIN_STREAM_FLAG_WHOLE_STREAM) { - DEBUG("Whole stream (size=%"PRIu64") will be read into memory", - lte->size); - ctx->cur_stream = lte; - filedes_invalidate(&ctx->tmpfile_fd); - return 0; + ctx->cur_blob_offset += size; + + if (likely(ctx->supported_features.hard_links)) { + progress->extract.completed_bytes += + (u64)size * ctx->cur_blob->out_refcnt; + if (ctx->cur_blob_offset == ctx->cur_blob->size) + progress->extract.completed_streams += ctx->cur_blob->out_refcnt; + } else { + const struct blob_extraction_target *targets = + blob_extraction_targets(ctx->cur_blob); + for (u32 i = 0; i < ctx->cur_blob->out_refcnt; i++) { + const struct wim_inode *inode = targets[i].inode; + const struct wim_dentry *dentry; + + inode_for_each_extraction_alias(dentry, inode) { + progress->extract.completed_bytes += size; + if (ctx->cur_blob_offset == ctx->cur_blob->size) + progress->extract.completed_streams++; + } + } } + if (progress->extract.completed_bytes >= ctx->next_progress) { - if (!need_tmpfile_to_extract(lte, - (flags & BEGIN_STREAM_FLAG_PARTIAL_RESOURCE))) - { - DEBUG("Temporary file not needed " - "for stream (size=%"PRIu64")", lte->size); - ret = extract_stream_instances(lte, lte, ctx); + ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS); if (ret) return ret; - return BEGIN_STREAM_STATUS_SKIP_STREAM; + set_next_progress(progress->extract.completed_bytes, + progress->extract.total_bytes, + &ctx->next_progress); } - DEBUG("Temporary file needed for stream (size=%"PRIu64")", lte->size); - return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name); -} - -static int -extract_chunk(const void *chunk, size_t size, void *_ctx) -{ - struct apply_ctx *ctx = _ctx; - int ret; - - if (filedes_valid(&ctx->tmpfile_fd)) { + if (unlikely(filedes_valid(&ctx->tmpfile_fd))) { + /* Just extracting to temporary file for now. */ ret = full_write(&ctx->tmpfile_fd, chunk, size); - if (ret) - ERROR_WITH_ERRNO("Error writing to file descriptor"); - } else { - struct wim_lookup_table_entry lte_override; - - memcpy(<e_override, ctx->cur_stream, - sizeof(struct wim_lookup_table_entry)); - - lte_override.resource_location = RESOURCE_IN_ATTACHED_BUFFER; - lte_override.size = size; - lte_override.attached_buffer = (void *)chunk; - - ret = extract_stream_instances(ctx->cur_stream, <e_override, ctx); - } - return ret; -} - -static int -end_extract_stream(struct wim_lookup_table_entry *lte, - int status, void *_ctx) -{ - struct apply_ctx *ctx = _ctx; - struct wim_lookup_table_entry lte_override; - int ret; - int errno_save = errno; - - if (!filedes_valid(&ctx->tmpfile_fd)) - return status; - - ret = filedes_close(&ctx->tmpfile_fd); - - if (status) { - ret = status; - errno = errno_save; - goto out_delete_tmpfile; - } - - if (ret) { - ERROR_WITH_ERRNO("Error writing temporary file %"TS, ctx->tmpfile_name); - ret = WIMLIB_ERR_WRITE; - goto out_delete_tmpfile; + if (ret) { + ERROR_WITH_ERRNO("Error writing data to " + "temporary file \"%"TS"\"", + ctx->tmpfile_name); + } + return ret; } - /* Now that a full stream has been extracted to a temporary file, - * extract all instances of it to the actual target. */ - - memcpy(<e_override, lte, sizeof(struct wim_lookup_table_entry)); - lte_override.resource_location = RESOURCE_IN_FILE_ON_DISK; - lte_override.file_on_disk = ctx->tmpfile_name; - - ret = extract_stream_instances(lte, <e_override, ctx); - -out_delete_tmpfile: - errno_save = errno; - tunlink(ctx->tmpfile_name); - FREE(ctx->tmpfile_name); - errno = errno_save; - return ret; -} - -/* Extracts a list of streams (ctx.stream_list), assuming that the directory - * structure and empty files were already created. This relies on the - * per-`struct wim_lookup_table_entry' list of dentries that reference each - * stream that was constructed earlier. */ -static int -extract_stream_list(struct apply_ctx *ctx) -{ - struct read_stream_list_callbacks cbs = { - .begin_stream = begin_extract_stream, - .begin_stream_ctx = ctx, - .consume_chunk = extract_chunk, - .consume_chunk_ctx = ctx, - .end_stream = end_extract_stream, - .end_stream_ctx = ctx, - }; - return read_stream_list(&ctx->stream_list, - offsetof(struct wim_lookup_table_entry, extraction_list), - &cbs, VERIFY_STREAM_HASHES); + return call_consume_chunk(chunk, size, ctx->saved_cbs); } -#define PWM_ALLOW_WIM_HDR 0x00001 -#define PWM_SILENT_EOF 0x00002 - -/* Read the header from a stream in a pipable WIM. */ +/* Copy the blob's data from the temporary file to each of its targets. + * + * This is executed only in the very uncommon case that a blob is being + * extracted to more than MAX_OPEN_FILES targets! */ static int -read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte, - struct wim_resource_spec *rspec, - int flags, struct wim_header_disk *hdr_ret) +extract_from_tmpfile(const tchar *tmpfile_name, + const struct blob_descriptor *orig_blob, + const struct read_blob_callbacks *cbs) { - union { - struct pwm_stream_hdr stream_hdr; - struct wim_header_disk pwm_hdr; - } buf; - struct wim_reshdr reshdr; + struct blob_descriptor tmpfile_blob; + const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob); int ret; - ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr)); - if (ret) - goto read_error; - - if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) { - BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr)); - ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1, - sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr)); + memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor)); + tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK; + tmpfile_blob.file_on_disk = (tchar *)tmpfile_name; + tmpfile_blob.out_refcnt = 1; + for (u32 i = 0; i < orig_blob->out_refcnt; i++) { + tmpfile_blob.inline_blob_extraction_targets[0] = targets[i]; + ret = read_blob_with_cbs(&tmpfile_blob, cbs); if (ret) - goto read_error; - lte->resource_location = RESOURCE_NONEXISTENT; - memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr)); - return 0; - } - - if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) { - ERROR("Data read on pipe is invalid (expected stream header)."); - return WIMLIB_ERR_INVALID_PIPABLE_WIM; + return ret; } - - copy_hash(lte->hash, buf.stream_hdr.hash); - - reshdr.size_in_wim = 0; - reshdr.flags = le32_to_cpu(buf.stream_hdr.flags); - reshdr.offset_in_wim = pwm->in_fd.offset; - reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size); - wim_res_hdr_to_spec(&reshdr, pwm, rspec); - lte_bind_wim_resource_spec(lte, rspec); - lte->flags = rspec->flags; - lte->size = rspec->uncompressed_size; - lte->offset_in_res = 0; return 0; - -read_error: - if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF)) - ERROR_WITH_ERRNO("Error reading pipable WIM from pipe"); - return ret; } static int -extract_streams_from_pipe(struct apply_ctx *ctx) +end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx) { - struct wim_lookup_table_entry *found_lte; - struct wim_resource_spec *rspec; - struct wim_lookup_table_entry *needed_lte; - struct wim_lookup_table *lookup_table; - struct wim_header_disk pwm_hdr; - int ret; - int pwm_flags; - - ret = WIMLIB_ERR_NOMEM; - found_lte = new_lookup_table_entry(); - if (found_lte == NULL) - goto out; - - rspec = MALLOC(sizeof(struct wim_resource_spec)); - if (rspec == NULL) - goto out_free_found_lte; - - 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; - memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN); - ctx->progress.extract.part_number = ctx->wim->hdr.part_number; - ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts; - if (ctx->progress_func) - ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN, - &ctx->progress); - while (ctx->num_streams_remaining) { - if (found_lte->resource_location != RESOURCE_NONEXISTENT) - lte_unbind_wim_resource_spec(found_lte); - ret = read_pwm_stream_header(ctx->wim, found_lte, rspec, - pwm_flags, &pwm_hdr); - 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->flags & WIM_RESHDR_FLAG_METADATA) - && (needed_lte = lookup_stream(lookup_table, found_lte->hash)) - && (needed_lte->out_refcnt)) - { - tchar *tmpfile_name = NULL; - struct wim_lookup_table_entry *lte_override; - struct wim_lookup_table_entry tmpfile_lte; - - needed_lte->offset_in_res = found_lte->offset_in_res; - needed_lte->flags = found_lte->flags; - needed_lte->size = found_lte->size; - - lte_unbind_wim_resource_spec(found_lte); - lte_bind_wim_resource_spec(needed_lte, rspec); - - if (needed_lte->out_refcnt > 1) { - - struct filedes tmpfile_fd; - - /* Extract stream to temporary file. */ - ret = create_temporary_file(&tmpfile_fd, &tmpfile_name); - if (ret) { - lte_unbind_wim_resource_spec(needed_lte); - goto out_free_found_lte; - } - - ret = extract_full_stream_to_fd(needed_lte, - &tmpfile_fd); - if (ret) { - filedes_close(&tmpfile_fd); - goto delete_tmpfile; - } - - if (filedes_close(&tmpfile_fd)) { - ERROR_WITH_ERRNO("Error writing to temporary " - "file \"%"TS"\"", tmpfile_name); - ret = WIMLIB_ERR_WRITE; - goto delete_tmpfile; - } - memcpy(&tmpfile_lte, needed_lte, - sizeof(struct wim_lookup_table_entry)); - tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK; - tmpfile_lte.file_on_disk = tmpfile_name; - lte_override = &tmpfile_lte; - } else { - lte_override = needed_lte; - } + struct apply_ctx *ctx = _ctx; - ret = extract_stream_instances(needed_lte, lte_override, ctx); - delete_tmpfile: - lte_unbind_wim_resource_spec(needed_lte); - if (tmpfile_name) { - tunlink(tmpfile_name); - FREE(tmpfile_name); - } - if (ret) - goto out_free_found_lte; - ctx->num_streams_remaining--; - } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) { - ret = skip_wim_stream(found_lte); - if (ret) - goto out_free_found_lte; - } else { - u16 part_number = le16_to_cpu(pwm_hdr.part_number); - u16 total_parts = le16_to_cpu(pwm_hdr.total_parts); + if (unlikely(filedes_valid(&ctx->tmpfile_fd))) { + filedes_close(&ctx->tmpfile_fd); + if (!status) + status = extract_from_tmpfile(ctx->tmpfile_name, blob, + ctx->saved_cbs); + filedes_invalidate(&ctx->tmpfile_fd); + tunlink(ctx->tmpfile_name); + FREE(ctx->tmpfile_name); + return status; + } - if (part_number != ctx->progress.extract.part_number || - total_parts != ctx->progress.extract.total_parts || - memcmp(pwm_hdr.guid, ctx->progress.extract.guid, - WIM_GID_LEN)) - { - ctx->progress.extract.part_number = part_number; - ctx->progress.extract.total_parts = total_parts; - memcpy(ctx->progress.extract.guid, - pwm_hdr.guid, WIM_GID_LEN); - if (ctx->progress_func) { - ctx->progress_func( - WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN, - &ctx->progress); - } + return call_end_blob(blob, status, ctx->saved_cbs); +} - } - } +/* + * Read the list of blobs to extract and feed their data into the specified + * callback functions. + * + * This handles checksumming each blob. + * + * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS. + * + * This also works if the WIM is being read from a pipe. + * + * This also will split up blobs that will need to be extracted to more than + * MAX_OPEN_FILES locations, as measured by the 'out_refcnt' of each blob. + * Therefore, the apply_operations implementation need not worry about running + * out of file descriptors, unless it might open more than one file descriptor + * per 'blob_extraction_target' (e.g. Win32 currently might because the + * destination file system might not support hard links). + */ +int +extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs) +{ + struct read_blob_callbacks wrapper_cbs = { + .begin_blob = begin_extract_blob_wrapper, + .consume_chunk = extract_chunk_wrapper, + .end_blob = end_extract_blob_wrapper, + .ctx = ctx, + }; + ctx->saved_cbs = cbs; + if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) { + return read_blobs_from_pipe(ctx, &wrapper_cbs); + } else { + return read_blob_list(&ctx->blob_list, + offsetof(struct blob_descriptor, + extraction_list), + &wrapper_cbs, VERIFY_BLOB_HASHES); } - ret = 0; -out_free_found_lte: - if (found_lte->resource_location != RESOURCE_IN_WIM) - FREE(rspec); - free_lookup_table_entry(found_lte); -out: - return ret; - -resume_done: - /* TODO */ - return 0; } /* Extract a WIM dentry to standard output. @@ -1596,38 +455,39 @@ resume_done: * unnamed data stream only. */ static int extract_dentry_to_stdout(struct wim_dentry *dentry, - const struct wim_lookup_table *lookup_table) + const struct blob_table *blob_table) { struct wim_inode *inode = dentry->d_inode; - struct wim_lookup_table_entry *lte; + struct blob_descriptor *blob; struct filedes _stdout; if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT | - FILE_ATTRIBUTE_DIRECTORY)) + FILE_ATTRIBUTE_DIRECTORY | + FILE_ATTRIBUTE_ENCRYPTED)) { ERROR("\"%"TS"\" is not a regular file and therefore cannot be " "extracted to standard output", dentry_full_path(dentry)); return WIMLIB_ERR_NOT_A_REGULAR_FILE; } - lte = inode_unnamed_lte(inode, lookup_table); - if (!lte) { - const u8 *hash = inode_unnamed_stream_hash(inode); + blob = inode_get_blob_for_unnamed_data_stream(inode, blob_table); + if (!blob) { + const u8 *hash = inode_get_hash_of_unnamed_data_stream(inode); if (!is_zero_hash(hash)) - return stream_not_found_error(inode, hash); + return blob_not_found_error(inode, hash); return 0; } filedes_init(&_stdout, STDOUT_FILENO); - return extract_full_stream_to_fd(lte, &_stdout); + return extract_blob_to_fd(blob, &_stdout); } static int extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries, - const struct wim_lookup_table *lookup_table) + const struct blob_table *blob_table) { for (size_t i = 0; i < num_dentries; i++) { - int ret = extract_dentry_to_stdout(dentries[i], lookup_table); + int ret = extract_dentry_to_stdout(dentries[i], blob_table); if (ret) return ret; } @@ -1646,14 +506,14 @@ remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees) { size_t i, j = 0; for (i = 0; i < num_trees; i++) { - if (!trees[i]->tmp_flag) { + if (!trees[i]->d_tmp_flag) { /* Found distinct dentry. */ - trees[i]->tmp_flag = 1; + trees[i]->d_tmp_flag = 1; trees[j++] = trees[i]; } } for (i = 0; i < j; i++) - trees[i]->tmp_flag = 0; + trees[i]->d_tmp_flag = 0; return j; } @@ -1667,23 +527,23 @@ remove_contained_trees(struct wim_dentry **trees, size_t num_trees) { size_t i, j = 0; for (i = 0; i < num_trees; i++) - trees[i]->tmp_flag = 1; + trees[i]->d_tmp_flag = 1; for (i = 0; i < num_trees; i++) { struct wim_dentry *d = trees[i]; while (!dentry_is_root(d)) { - d = d->parent; - if (d->tmp_flag) + d = d->d_parent; + if (d->d_tmp_flag) goto tree_contained; } trees[j++] = trees[i]; continue; tree_contained: - trees[i]->tmp_flag = 0; + trees[i]->d_tmp_flag = 0; } for (i = 0; i < j; i++) - trees[i]->tmp_flag = 0; + trees[i]->d_tmp_flag = 0; return j; } @@ -1691,21 +551,23 @@ static int dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list) { struct list_head *dentry_list = _dentry_list; - list_add_tail(&dentry->extraction_list, dentry_list); + list_add_tail(&dentry->d_extraction_list_node, dentry_list); return 0; } static void dentry_reset_extraction_list_node(struct wim_dentry *dentry) { - dentry->extraction_list = (struct list_head){NULL, NULL}; + dentry->d_extraction_list_node = (struct list_head){NULL, NULL}; } static int dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore) { - list_del(&dentry->extraction_list); - dentry_reset_extraction_list_node(dentry); + if (will_extract_dentry(dentry)) { + list_del(&dentry->d_extraction_list_node); + dentry_reset_extraction_list_node(dentry); + } return 0; } @@ -1738,36 +600,50 @@ build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees, place_after = dentry_list; ancestor = dentry; do { - ancestor = ancestor->parent; - if (dentry_in_list(ancestor)) { - place_after = &ancestor->extraction_list; + ancestor = ancestor->d_parent; + if (will_extract_dentry(ancestor)) { + place_after = &ancestor->d_extraction_list_node; break; } } while (!dentry_is_root(ancestor)); ancestor = dentry; do { - ancestor = ancestor->parent; - if (dentry_in_list(ancestor)) + ancestor = ancestor->d_parent; + if (will_extract_dentry(ancestor)) break; - list_add(&ancestor->extraction_list, place_after); + list_add(&ancestor->d_extraction_list_node, place_after); } while (!dentry_is_root(ancestor)); } } } -static const struct apply_operations * -select_apply_operations(int extract_flags) +static void +destroy_dentry_list(struct list_head *dentry_list) { -#ifdef WITH_NTFS_3G - if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) - return &ntfs_3g_apply_ops; -#endif -#ifdef __WIN32__ - return &win32_apply_ops; -#else - return &unix_apply_ops; -#endif + struct wim_dentry *dentry, *tmp; + struct wim_inode *inode; + + list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) { + inode = dentry->d_inode; + dentry_reset_extraction_list_node(dentry); + inode->i_visited = 0; + inode->i_can_externally_back = 0; + if ((void *)dentry->d_extraction_name != (void *)dentry->d_name) + FREE(dentry->d_extraction_name); + dentry->d_extraction_name = NULL; + dentry->d_extraction_name_nchars = 0; + } +} + +static void +destroy_blob_list(struct list_head *blob_list) +{ + struct blob_descriptor *blob; + + list_for_each_entry(blob, blob_list, extraction_list) + if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets)) + FREE(blob->blob_extraction_targets); } #ifdef __WIN32__ @@ -1784,19 +660,20 @@ file_name_valid(utf16lechar *name, size_t num_chars, bool fix) if (num_chars == 0) return true; for (i = 0; i < num_chars; i++) { - switch (name[i]) { + switch (le16_to_cpu(name[i])) { #ifdef __WIN32__ - case cpu_to_le16('\\'): - case cpu_to_le16(':'): - case cpu_to_le16('*'): - case cpu_to_le16('?'): - case cpu_to_le16('"'): - case cpu_to_le16('<'): - case cpu_to_le16('>'): - case cpu_to_le16('|'): + case '\x01'...'\x1F': + case '\\': + case ':': + case '*': + case '?': + case '"': + case '<': + case '>': + case '|': #endif - case cpu_to_le16('/'): - case cpu_to_le16('\0'): + case '/': + case '\0': if (fix) name[i] = replacement_char; else @@ -1804,16 +681,6 @@ file_name_valid(utf16lechar *name, size_t num_chars, bool fix) } } -#ifdef __WIN32__ - if (name[num_chars - 1] == cpu_to_le16(' ') || - name[num_chars - 1] == cpu_to_le16('.')) - { - if (fix) - name[num_chars - 1] = replacement_char; - else - return false; - } -#endif return true; } @@ -1823,19 +690,24 @@ dentry_calculate_extraction_name(struct wim_dentry *dentry, { int ret; - if (dentry == ctx->target_dentry) + if (dentry_is_root(dentry)) return 0; - if (!dentry_is_supported(dentry, &ctx->supported_features)) - goto skip_dentry; +#ifdef WITH_NTFS_3G + if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) { + dentry->d_extraction_name = dentry->d_name; + dentry->d_extraction_name_nchars = dentry->d_name_nbytes / + sizeof(utf16lechar); + return 0; + } +#endif - if (!ctx->ops->supports_case_sensitive_filenames) - { + if (!ctx->supported_features.case_sensitive_filenames) { struct wim_dentry *other; list_for_each_entry(other, &dentry->d_ci_conflict_list, d_ci_conflict_list) { - if (dentry_in_list(other)) { + if (will_extract_dentry(other)) { if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) { WARNING("\"%"TS"\" has the same " @@ -1857,17 +729,14 @@ dentry_calculate_extraction_name(struct wim_dentry *dentry, } } - if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) { -#if TCHAR_IS_UTF16LE - dentry->extraction_name = dentry->file_name; - dentry->extraction_name_nchars = dentry->file_name_nbytes / 2; - return 0; -#else - return utf16le_to_tstr(dentry->file_name, - dentry->file_name_nbytes, - &dentry->extraction_name, - &dentry->extraction_name_nchars); -#endif + if (file_name_valid(dentry->d_name, dentry->d_name_nbytes / 2, false)) { + size_t nbytes = 0; + ret = utf16le_get_tstr(dentry->d_name, + dentry->d_name_nbytes, + (const tchar **)&dentry->d_extraction_name, + &nbytes); + dentry->d_extraction_name_nchars = nbytes / sizeof(tchar); + return ret; } else { if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES) { @@ -1886,23 +755,22 @@ dentry_calculate_extraction_name(struct wim_dentry *dentry, out_replace: { - utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2]; + utf16lechar utf16_name_copy[dentry->d_name_nbytes / 2]; - memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes); - file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true); + memcpy(utf16_name_copy, dentry->d_name, dentry->d_name_nbytes); + file_name_valid(utf16_name_copy, dentry->d_name_nbytes / 2, true); - tchar *tchar_name; + const tchar *tchar_name; size_t tchar_nchars; - #if TCHAR_IS_UTF16LE - tchar_name = utf16_name_copy; - tchar_nchars = dentry->file_name_nbytes / 2; - #else - ret = utf16le_to_tstr(utf16_name_copy, - dentry->file_name_nbytes, - &tchar_name, &tchar_nchars); + + ret = utf16le_get_tstr(utf16_name_copy, + dentry->d_name_nbytes, + &tchar_name, &tchar_nchars); if (ret) return ret; - #endif + + tchar_nchars /= sizeof(tchar); + size_t fixed_name_num_chars = tchar_nchars; tchar fixed_name[tchar_nchars + 50]; @@ -1910,14 +778,13 @@ out_replace: fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars, T(" (invalid filename #%lu)"), ++ctx->invalid_sequence); - #if !TCHAR_IS_UTF16LE - FREE(tchar_name); - #endif - dentry->extraction_name = memdup(fixed_name, - 2 * fixed_name_num_chars + 2); - if (!dentry->extraction_name) + + utf16le_put_tstr(tchar_name); + + dentry->d_extraction_name = TSTRDUP(fixed_name); + if (!dentry->d_extraction_name) return WIMLIB_ERR_NOMEM; - dentry->extraction_name_nchars = fixed_name_num_chars; + dentry->d_extraction_name_nchars = fixed_name_num_chars; } return 0; @@ -1931,9 +798,7 @@ skip_dentry: * extracted, with special handling for dentries that are unsupported by the * extraction backend or have invalid names. * - * Note: this has a dependency on start_extract() being called because - * ctx.supported_features must be filled in in order to determine whether each - * dentry is supported. + * ctx->supported_features must be filled in. * * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING */ @@ -1956,7 +821,7 @@ dentry_list_calculate_extraction_names(struct list_head *dentry_list, if (cur == dentry_list) break; - dentry = list_entry(cur, struct wim_dentry, extraction_list); + dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node); ret = dentry_calculate_extraction_name(dentry, ctx); if (ret) @@ -1973,34 +838,35 @@ dentry_list_calculate_extraction_names(struct list_head *dentry_list, static int dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags, - struct wim_lookup_table *lookup_table) + struct blob_table *blob_table) { struct wim_inode *inode = dentry->d_inode; - struct wim_lookup_table_entry *lte; + struct blob_descriptor *blob; int ret; bool force = false; - /* Special case: when extracting from a pipe, the WIM lookup table is + /* Special case: when extracting from a pipe, the WIM blob table is * initially empty, so "resolving" an inode's streams is initially not - * possible. However, we still need to keep track of which streams, - * identified by SHA1 message digests, need to be extracted, so we - * "resolve" the inode's streams anyway by allocating new entries. */ + * possible. However, we still need to keep track of which blobs, + * identified by SHA-1 message digests, need to be extracted, so we + * "resolve" the inode's streams anyway by allocating a 'struct + * blob_descriptor' for each one. */ if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) force = true; - ret = inode_resolve_streams(inode, lookup_table, force); + ret = inode_resolve_streams(inode, blob_table, force); if (ret) return ret; - for (u32 i = 0; i <= inode->i_num_ads; i++) { - lte = inode_stream_lte_resolved(inode, i); - if (lte) - lte->out_refcnt = 0; + for (unsigned i = 0; i < inode->i_num_streams; i++) { + blob = stream_blob_resolved(&inode->i_streams[i]); + if (blob) + blob->out_refcnt = 0; } return 0; } /* * For each dentry to be extracted, resolve all streams in the corresponding - * inode and set 'out_refcnt' in each to 0. + * inode and set 'out_refcnt' in all referenced blob_descriptors to 0. * * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM. */ @@ -2011,10 +877,10 @@ dentry_list_resolve_streams(struct list_head *dentry_list, struct wim_dentry *dentry; int ret; - list_for_each_entry(dentry, dentry_list, extraction_list) { + list_for_each_entry(dentry, dentry_list, d_extraction_list_node) { ret = dentry_resolve_streams(dentry, ctx->extract_flags, - ctx->wim->lookup_table); + ctx->wim->blob_table); if (ret) return ret; } @@ -2022,125 +888,156 @@ dentry_list_resolve_streams(struct list_head *dentry_list, } static int -ref_stream(struct wim_lookup_table_entry *lte, - struct wim_dentry *dentry, struct apply_ctx *ctx) +ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry, + struct apply_ctx *ctx) { - if (!lte) + struct wim_inode *inode = dentry->d_inode; + struct blob_descriptor *blob = stream_blob_resolved(strm); + struct blob_extraction_target *targets; + + if (!blob) return 0; - /* Tally the size only for each extraction of the stream (not hard - * links). */ - if (!(dentry->d_inode->i_visited && - ctx->supported_features.hard_links) && - (!is_linked_extraction(ctx) || (lte->out_refcnt == 0 && - lte->extracted_file == NULL))) - { - ctx->progress.extract.total_bytes += lte->size; - ctx->progress.extract.num_streams++; - } + /* Tally the size only for each actual extraction of the stream (not + * additional hard links to the inode). */ + if (inode->i_visited && ctx->supported_features.hard_links) + return 0; + + ctx->progress.extract.total_bytes += blob->size; + ctx->progress.extract.total_streams++; + + if (inode->i_visited) + return 0; - /* Add stream to the dentry_list only one time, even if it's going - * to be extracted to multiple locations. */ - if (lte->out_refcnt == 0) { - list_add_tail(<e->extraction_list, &ctx->stream_list); - ctx->num_streams_remaining++; + /* Add each blob to 'ctx->blob_list' only one time, regardless of how + * many extraction targets it will have. */ + if (blob->out_refcnt == 0) { + list_add_tail(&blob->extraction_list, &ctx->blob_list); + ctx->num_blobs_remaining++; } - if (!(ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) { - struct wim_dentry **lte_dentries; + /* Set this stream as an extraction target of 'blob'. */ + + if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) { + targets = blob->inline_blob_extraction_targets; + } else { + struct blob_extraction_target *prev_targets; + size_t alloc_blob_extraction_targets; - /* Append dentry to this stream's array of dentries referencing - * it. Use inline array to avoid memory allocation until the - * number of dentries becomes too large. */ - if (lte->out_refcnt < ARRAY_LEN(lte->inline_lte_dentries)) { - lte_dentries = lte->inline_lte_dentries; + if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) { + prev_targets = NULL; + alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets); } else { - struct wim_dentry **prev_lte_dentries; - size_t alloc_lte_dentries; + prev_targets = blob->blob_extraction_targets; + alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets; + } - if (lte->out_refcnt == ARRAY_LEN(lte->inline_lte_dentries)) { - prev_lte_dentries = NULL; - alloc_lte_dentries = ARRAY_LEN(lte->inline_lte_dentries); - } else { - prev_lte_dentries = lte->lte_dentries; - alloc_lte_dentries = lte->alloc_lte_dentries; + if (blob->out_refcnt == alloc_blob_extraction_targets) { + alloc_blob_extraction_targets *= 2; + targets = REALLOC(prev_targets, + alloc_blob_extraction_targets * + sizeof(targets[0])); + if (!targets) + return WIMLIB_ERR_NOMEM; + if (!prev_targets) { + memcpy(targets, + blob->inline_blob_extraction_targets, + sizeof(blob->inline_blob_extraction_targets)); } + blob->blob_extraction_targets = targets; + blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets; + } + targets = blob->blob_extraction_targets; + } + targets[blob->out_refcnt].inode = inode; + targets[blob->out_refcnt].stream = strm; + blob->out_refcnt++; + return 0; +} - if (lte->out_refcnt == alloc_lte_dentries) { - alloc_lte_dentries *= 2; - lte_dentries = REALLOC(prev_lte_dentries, - alloc_lte_dentries * - sizeof(lte_dentries[0])); - if (lte_dentries == NULL) - return WIMLIB_ERR_NOMEM; - if (prev_lte_dentries == NULL) { - memcpy(lte_dentries, - lte->inline_lte_dentries, - sizeof(lte->inline_lte_dentries)); - } - lte->lte_dentries = lte_dentries; - lte->alloc_lte_dentries = alloc_lte_dentries; +static int +ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode, + struct wim_inode_stream *strm, struct apply_ctx *ctx) +{ + bool need_stream = false; + switch (strm->stream_type) { + case STREAM_TYPE_DATA: + if (stream_is_named(strm)) { + /* Named data stream */ + if (ctx->supported_features.named_data_streams) + need_stream = true; + } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY | + FILE_ATTRIBUTE_ENCRYPTED)) + && !(inode_is_symlink(inode) + && !ctx->supported_features.reparse_points + && ctx->supported_features.symlink_reparse_points)) + { + /* + * Unnamed data stream. Skip if any of the following is true: + * + * - file is a directory + * - file is encrypted + * - backend needs to create the file as UNIX symlink + * - backend will extract the stream as externally + * backed from the WIM archive itself + */ + if (ctx->apply_ops->will_back_from_wim) { + int ret = (*ctx->apply_ops->will_back_from_wim)(dentry, ctx); + if (ret > 0) /* Error? */ + return ret; + if (ret < 0) /* Won't externally back? */ + need_stream = true; + } else { + need_stream = true; } - lte_dentries = lte->lte_dentries; } - lte_dentries[lte->out_refcnt] = dentry; + break; + case STREAM_TYPE_REPARSE_POINT: + wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT); + if (ctx->supported_features.reparse_points || + (inode_is_symlink(inode) && + ctx->supported_features.symlink_reparse_points)) + need_stream = true; + break; + case STREAM_TYPE_EFSRPC_RAW_DATA: + wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED); + if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) { + if (ctx->supported_features.encrypted_directories) + need_stream = true; + } else { + if (ctx->supported_features.encrypted_files) + need_stream = true; + } + break; } - lte->out_refcnt++; + if (need_stream) + return ref_stream(strm, dentry, ctx); return 0; } static int dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx) -{ - struct wim_inode *inode = dentry->d_inode; - int ret; - - /* The unnamed data stream will always be extracted, except in an - * unlikely case. */ - if (!inode_is_encrypted_directory(inode)) { - ret = ref_stream(inode_unnamed_lte_resolved(inode), - dentry, ctx); - if (ret) - return ret; - } - - /* Named data streams will be extracted only if supported in the current - * extraction mode and volume, and to avoid complications, if not doing - * a linked extraction. */ - if (can_extract_named_data_streams(ctx)) { - for (u16 i = 0; i < inode->i_num_ads; i++) { - if (!ads_entry_is_named_stream(&inode->i_ads_entries[i])) - continue; - ret = ref_stream(inode->i_ads_entries[i].lte, - dentry, ctx); - if (ret) - return ret; - } +{ + struct wim_inode *inode = dentry->d_inode; + for (unsigned i = 0; i < inode->i_num_streams; i++) { + int ret = ref_stream_if_needed(dentry, inode, + &inode->i_streams[i], ctx); + if (ret) + return ret; } inode->i_visited = 1; return 0; } /* - * For each dentry to be extracted, iterate through the data streams of the - * corresponding inode. For each such stream that is not to be ignored due to - * the supported features or extraction flags, add it to the list of streams to - * be extracted (ctx->stream_list) if not already done so. + * Given a list of dentries to be extracted, build the list of blobs that need + * to be extracted, and for each blob determine the streams to which that blob + * will be extracted. * - * Also, if doing a sequential extraction, build a mapping from each stream to - * the dentries referencing it. - * - * This also initializes the extract progress info with byte and stream + * This also initializes the extract progress info with byte and blob * information. * - * Note: This has a dependency on start_extract being called because - * ctx.supported_features must be filled in in order to determine whether named - * data streams are supported. - * - * Note: this uses the i_visited member of the inodes (assumed to be 0 - * initially), but does not reset it. - * - * Possible error codes: WIMLIB_ERR_NOMEM. + * ctx->supported_features must be filled in. */ static int dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx) @@ -2148,20 +1045,34 @@ dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx) struct wim_dentry *dentry; int ret; - list_for_each_entry(dentry, dentry_list, extraction_list) { + list_for_each_entry(dentry, dentry_list, d_extraction_list_node) { ret = dentry_ref_streams(dentry, ctx); if (ret) return ret; } + list_for_each_entry(dentry, dentry_list, d_extraction_list_node) + dentry->d_inode->i_visited = 0; return 0; } -/* Tally features necessary to extract a dentry and the corresponding inode. */ static void -dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features) +dentry_list_build_inode_alias_lists(struct list_head *dentry_list) { - struct wim_inode *inode = dentry->d_inode; + struct wim_dentry *dentry; + + list_for_each_entry(dentry, dentry_list, d_extraction_list_node) + dentry->d_inode->i_first_extraction_alias = NULL; + + list_for_each_entry(dentry, dentry_list, d_extraction_list_node) { + dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias; + dentry->d_inode->i_first_extraction_alias = dentry; + } +} +static void +inode_tally_features(const struct wim_inode *inode, + struct wim_features *features) +{ if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE) features->archive_files++; if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN) @@ -2180,10 +1091,8 @@ dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features) features->not_context_indexed_files++; if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE) features->sparse_files++; - if (inode_has_named_stream(inode)) + if (inode_has_named_data_stream(inode)) features->named_data_streams++; - if (inode->i_visited) - features->hard_links++; if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) { features->reparse_points++; if (inode_is_symlink(inode)) @@ -2191,13 +1100,27 @@ dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features) else features->other_reparse_points++; } - if (inode->i_security_id != -1) + if (inode_has_security_descriptor(inode)) features->security_descriptors++; - if (dentry->short_name_nbytes) - features->short_names++; if (inode_has_unix_data(inode)) features->unix_data++; - inode->i_visited = 1; +} + +/* Tally features necessary to extract a dentry and the corresponding inode. */ +static void +dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features) +{ + struct wim_inode *inode = dentry->d_inode; + + if (dentry_has_short_name(dentry)) + features->short_names++; + + if (inode->i_visited) { + features->hard_links++; + } else { + inode_tally_features(inode, features); + inode->i_visited = 1; + } } /* Tally the features necessary to extract the specified dentries. */ @@ -2207,49 +1130,30 @@ dentry_list_get_features(struct list_head *dentry_list, { struct wim_dentry *dentry; - memset(features, 0, sizeof(struct wim_features)); - - list_for_each_entry(dentry, dentry_list, extraction_list) + list_for_each_entry(dentry, dentry_list, d_extraction_list_node) dentry_tally_features(dentry, features); - list_for_each_entry(dentry, dentry_list, extraction_list) + list_for_each_entry(dentry, dentry_list, d_extraction_list_node) dentry->d_inode->i_visited = 0; } -static u32 -compute_supported_attributes_mask(const struct wim_features *supported_features) -{ - u32 mask = (u32)~0UL; - - if (!supported_features->archive_files) - mask &= ~FILE_ATTRIBUTE_ARCHIVE; - - if (!supported_features->hidden_files) - mask &= ~FILE_ATTRIBUTE_HIDDEN; - - if (!supported_features->system_files) - mask &= ~FILE_ATTRIBUTE_SYSTEM; - - if (!supported_features->not_context_indexed_files) - mask &= ~FILE_ATTRIBUTE_NOT_CONTENT_INDEXED; - - if (!supported_features->compressed_files) - mask &= ~FILE_ATTRIBUTE_COMPRESSED; - - if (!supported_features->sparse_files) - mask &= ~FILE_ATTRIBUTE_SPARSE_FILE; - - if (!supported_features->reparse_points) - mask &= ~FILE_ATTRIBUTE_REPARSE_POINT; - - return mask; -} - 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) + int extract_flags) { + /* Encrypted files. */ + if (required_features->encrypted_files && + !supported_features->encrypted_files) + WARNING("Ignoring EFS-encrypted data of %lu files", + required_features->encrypted_files); + + /* Named data streams. */ + if (required_features->named_data_streams && + !supported_features->named_data_streams) + WARNING("Ignoring named data streams of %lu files", + required_features->named_data_streams); + /* File attributes. */ if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) { /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE. @@ -2287,38 +1191,12 @@ do_feature_check(const struct wim_features *required_features, required_features->encrypted_directories); } - /* Encrypted files. */ - if (required_features->encrypted_files && - !supported_features->encrypted_files) - WARNING("Ignoring %lu encrypted files", - required_features->encrypted_files); - - /* Named data streams. */ - if (required_features->named_data_streams && - (!supported_features->named_data_streams || - (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | - WIMLIB_EXTRACT_FLAG_HARDLINK)))) - WARNING("Ignoring named data streams of %lu files", - required_features->named_data_streams); - /* Hard links. */ - if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) && - !supported_features->hard_links) - { - ERROR("Extraction backend does not support hard links!"); - return WIMLIB_ERR_UNSUPPORTED; - } if (required_features->hard_links && !supported_features->hard_links) WARNING("Extracting %lu hard links as independent files", required_features->hard_links); /* Symbolic links and reparse points. */ - if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) && - !supported_features->symlink_reparse_points) - { - ERROR("Extraction backend does not support symbolic links!"); - return WIMLIB_ERR_UNSUPPORTED; - } if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) && required_features->symlink_reparse_points && !supported_features->symlink_reparse_points && @@ -2332,12 +1210,11 @@ do_feature_check(const struct wim_features *required_features, { if (supported_features->symlink_reparse_points) { if (required_features->other_reparse_points) { - WARNING("Ignoring %lu non-symlink/junction " - "reparse point files", + WARNING("Ignoring reparse data of %lu non-symlink/junction files", required_features->other_reparse_points); } } else { - WARNING("Ignoring %lu reparse point files", + WARNING("Ignoring reparse data of %lu files", required_features->reparse_points); } } @@ -2366,6 +1243,13 @@ do_feature_check(const struct wim_features *required_features, return WIMLIB_ERR_UNSUPPORTED; } + if (required_features->unix_data && + !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA)) + { + WARNING("Ignoring UNIX metadata of %lu files", + required_features->unix_data); + } + /* DOS Names. */ if (required_features->short_names && !supported_features->short_names) @@ -2380,7 +1264,7 @@ do_feature_check(const struct wim_features *required_features, /* Timestamps. */ if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) && - !ops->set_timestamps) + !supported_features->timestamps) { ERROR("Extraction backend does not support timestamps!"); return WIMLIB_ERR_UNSUPPORTED; @@ -2389,362 +1273,178 @@ do_feature_check(const struct wim_features *required_features, return 0; } -static void -do_extract_warnings(struct apply_ctx *ctx) +static const struct apply_operations * +select_apply_operations(int extract_flags) { - if (ctx->partial_security_descriptors == 0 && - ctx->no_security_descriptors == 0) - return; - - WARNING("Extraction to \"%"TS"\" complete, but with one or more warnings:", - ctx->target); - if (ctx->partial_security_descriptors != 0) { - WARNING("- Could only partially set the security descriptor\n" - " on %lu files or directories.", - ctx->partial_security_descriptors); - } - if (ctx->no_security_descriptors != 0) { - WARNING("- Could not set security descriptor at all\n" - " on %lu files or directories.", - ctx->no_security_descriptors); - } +#ifdef WITH_NTFS_3G + if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) + return &ntfs_3g_apply_ops; +#endif #ifdef __WIN32__ - WARNING("To fully restore all security descriptors, run the program\n" - " with Administrator rights."); + return &win32_apply_ops; +#else + return &unix_apply_ops; #endif } -static void -destroy_dentry_list(struct list_head *dentry_list) -{ - struct wim_dentry *dentry, *tmp; - struct wim_inode *inode; - - list_for_each_entry_safe(dentry, tmp, dentry_list, extraction_list) { - inode = dentry->d_inode; - dentry_reset_extraction_list_node(dentry); - dentry->was_linked = 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; - } -} - -static void -destroy_stream_list(struct list_head *stream_list) -{ - struct wim_lookup_table_entry *lte; - - list_for_each_entry(lte, stream_list, extraction_list) - if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries)) - FREE(lte->lte_dentries); -} - static int extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees, - const tchar *target, int extract_flags, - wimlib_progress_func_t progress_func) + const tchar *target, int extract_flags) { + const struct apply_operations *ops; + struct apply_ctx *ctx; int ret; - struct apply_ctx ctx; - struct list_head dentry_list; - struct wim_features required_features; - - /* Handle stdout extraction as a separate case. */ - if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) - return extract_dentries_to_stdout(trees, num_trees, - wim->lookup_table); - - /* Start initializing the apply_ctx. */ - memset(&ctx, 0, sizeof(struct apply_ctx)); - ctx.wim = wim; - ctx.extract_flags = extract_flags; - ctx.target = target; - ctx.target_nchars = tstrlen(target); - ctx.progress_func = progress_func; - if (progress_func) { - ctx.progress.extract.wimfile_name = wim->filename; - ctx.progress.extract.image = wim->current_image; - ctx.progress.extract.extract_flags = (extract_flags & - WIMLIB_EXTRACT_MASK_PUBLIC); - ctx.progress.extract.image_name = wimlib_get_image_name(wim, - wim->current_image); - ctx.progress.extract.target = target; - } - - ctx.target_dentry = wim_root_dentry(wim); - /* Note: ctx.target_dentry represents the dentry that gets extracted to - * @target. There may be none, in which case it gets set to the image - * root and never matches any of the dentries actually being extracted. - */ + LIST_HEAD(dentry_list); + + if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) { + ret = extract_dentries_to_stdout(trees, num_trees, + wim->blob_table); + goto out; + } num_trees = remove_duplicate_trees(trees, num_trees); + num_trees = remove_contained_trees(trees, num_trees); - /* All trees are now distinct. */ + ops = select_apply_operations(extract_flags); - num_trees = remove_contained_trees(trees, num_trees); + if (num_trees > 1 && ops->single_tree_only) { + ERROR("Extracting multiple directory trees " + "at once is not supported in %s extraction mode!", + ops->name); + ret = WIMLIB_ERR_UNSUPPORTED; + goto out; + } - /* All trees are now distinct and non-overlapping. */ + ctx = CALLOC(1, ops->context_size); + if (!ctx) { + ret = WIMLIB_ERR_NOMEM; + goto out; + } + + ctx->wim = wim; + ctx->target = target; + ctx->target_nchars = tstrlen(target); + ctx->extract_flags = extract_flags; + if (ctx->wim->progfunc) { + ctx->progfunc = ctx->wim->progfunc; + ctx->progctx = ctx->wim->progctx; + ctx->progress.extract.image = wim->current_image; + ctx->progress.extract.extract_flags = (extract_flags & + WIMLIB_EXTRACT_MASK_PUBLIC); + ctx->progress.extract.wimfile_name = wim->filename; + ctx->progress.extract.image_name = wimlib_get_image_name(wim, + wim->current_image); + ctx->progress.extract.target = target; + } + INIT_LIST_HEAD(&ctx->blob_list); + filedes_invalidate(&ctx->tmpfile_fd); + ctx->apply_ops = ops; + + ret = (*ops->get_supported_features)(target, &ctx->supported_features); + if (ret) + goto out_cleanup; - /* Build list of dentries to be extracted. */ build_dentry_list(&dentry_list, trees, num_trees, - !(extract_flags & WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)); - - /* Select the appropriate apply_operations based on the platform and - * extract_flags. */ - ctx.ops = select_apply_operations(extract_flags); - - /* Figure out whether the root dentry is being extracted to the root of - * a volume and therefore needs to be treated "specially", for example - * not being explicitly created and not having attributes set. */ - if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special) - ctx.root_dentry_is_special = ctx.ops->target_is_root(target); - - /* Call the start_extract() callback. This gives the apply_operations - * implementation a chance to do any setup needed to access the volume. - * Furthermore, start_extract() is expected to set the supported - * features of this extraction mode (ctx.supported_features), which are - * determined at runtime as they may vary depending on the actual - * volume. */ - ret = ctx.ops->start_extract(target, &ctx); + !(extract_flags & + WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)); + + dentry_list_get_features(&dentry_list, &ctx->required_features); + + ret = do_feature_check(&ctx->required_features, &ctx->supported_features, + ctx->extract_flags); if (ret) - goto out_destroy_dentry_list; + goto out_cleanup; - /* Get and check the features required to extract the dentries. */ - dentry_list_get_features(&dentry_list, &required_features); - ret = do_feature_check(&required_features, &ctx.supported_features, - extract_flags, ctx.ops); + ret = dentry_list_calculate_extraction_names(&dentry_list, ctx); if (ret) - goto out_finish_or_abort_extract; + goto out_cleanup; - ctx.supported_attributes_mask = - compute_supported_attributes_mask(&ctx.supported_features); + if (unlikely(list_empty(&dentry_list))) { + WARNING("There is nothing to extract!"); + goto out_cleanup; + } - /* Calculate extraction name for each dentry and remove subtrees that - * can't be extracted due to naming problems. */ - ret = dentry_list_calculate_extraction_names(&dentry_list, &ctx); + ret = dentry_list_resolve_streams(&dentry_list, ctx); if (ret) - goto out_finish_or_abort_extract; + goto out_cleanup; - /* Build list of streams to extract. */ - ret = dentry_list_resolve_streams(&dentry_list, &ctx); - if (ret) - goto out_finish_or_abort_extract; - INIT_LIST_HEAD(&ctx.stream_list); - ret = dentry_list_ref_streams(&dentry_list, &ctx); + dentry_list_build_inode_alias_lists(&dentry_list); + + ret = dentry_list_ref_streams(&dentry_list, ctx); if (ret) - goto out_destroy_stream_list; + goto out_cleanup; if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) { /* When extracting from a pipe, the number of bytes of data to * extract can't be determined in the normal way (examining the - * lookup table), since at this point all we have is a set of - * SHA1 message digests of streams that need to be extracted. + * blob table), since at this point all we have is a set of + * SHA-1 message digests of blobs that need to be extracted. * However, we can get a reasonably accurate estimate by taking * from the corresponding in the WIM XML * data. This does assume that a full image is being extracted, * but currently there is no API for doing otherwise. (Also, * subtract from this if hard links are * supported by the extraction mode.) */ - ctx.progress.extract.total_bytes = + ctx->progress.extract.total_bytes = wim_info_get_image_total_bytes(wim->wim_info, wim->current_image); - if (ctx.supported_features.hard_links) { - ctx.progress.extract.total_bytes -= + if (ctx->supported_features.hard_links) { + ctx->progress.extract.total_bytes -= wim_info_get_image_hard_link_bytes(wim->wim_info, wim->current_image); } } - if (ctx.ops->realpath_works_on_nonexisting_files && - ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) || - ctx.ops->requires_realtarget_in_paths)) - { - ctx.realtarget = realpath(target, NULL); - if (!ctx.realtarget) { - ret = WIMLIB_ERR_NOMEM; - goto out_destroy_stream_list; - } - ctx.realtarget_nchars = tstrlen(ctx.realtarget); - #ifdef __WIN32__ - /* Strip trailing slashes. If we don't do this, we may create a - * path with multiple consecutive backslashes, which for some - * reason causes Windows to report that the file cannot be found. - */ - while (ctx.realtarget_nchars >= 2 - && ctx.realtarget[ctx.realtarget_nchars - 1] == L'\\' - && ctx.realtarget[ctx.realtarget_nchars - 2] != L':') - { - ctx.realtarget[--ctx.realtarget_nchars] = L'\0'; - } - #endif - } - - if (progress_func) { - int msg; - if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) - msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN; - else - msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN; - progress_func(msg, &ctx.progress); - } + ret = extract_progress(ctx, + ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ? + WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN : + WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN)); + if (ret) + goto out_cleanup; - if (!ctx.root_dentry_is_special) { - tchar path[ctx.ops->path_max]; - if (build_extraction_path(path, ctx.target_dentry, &ctx)) { - ret = extract_inode(path, &ctx, ctx.target_dentry->d_inode); - if (ret) - goto out_free_realtarget; - } - } + ret = (*ops->extract)(&dentry_list, ctx); + if (ret) + goto out_cleanup; - /* If we need to fix up the targets of absolute symbolic links - * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to - * be absolute, use realpath() (or its replacement on Windows) to get - * the absolute path to the extraction target. Note that this requires - * the target directory to exist, unless - * realpath_works_on_nonexisting_files is set in the apply_operations. - * */ - if (!ctx.realtarget && - (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) && - required_features.symlink_reparse_points) || - ctx.ops->requires_realtarget_in_paths)) + if (ctx->progress.extract.completed_bytes < + ctx->progress.extract.total_bytes) { - ctx.realtarget = realpath(target, NULL); - if (!ctx.realtarget) { - ret = WIMLIB_ERR_NOMEM; - goto out_free_realtarget; - } - ctx.realtarget_nchars = tstrlen(ctx.realtarget); - } - - if (ctx.ops->requires_short_name_reordering) { - if (progress_func) - progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN, - &ctx.progress); - ret = extract_dir_structure(&dentry_list, &ctx); - if (ret) - goto out_free_realtarget; - } - - /* Finally, the important part: extract the tree of files. */ - if (!(extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) { - /* Sequential extraction requested, so two passes are needed - * (one for file structure, one for streams.) */ - if (progress_func && !ctx.ops->requires_short_name_reordering) - progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN, - &ctx.progress); - - if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) { - ret = extract_structure(&dentry_list, &ctx); - if (ret) - goto out_free_realtarget; - } - if (progress_func) - progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END, - &ctx.progress); - if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) - ret = extract_streams_from_pipe(&ctx); - else - ret = extract_stream_list(&ctx); - if (ret) - goto out_free_realtarget; - } else { - /* Sequential extraction was not requested, so we can make do - * with one pass where we both create the files and extract - * streams. */ - if (progress_func && !ctx.ops->requires_short_name_reordering) - progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN, - &ctx.progress); - ret = extract_dentries(&dentry_list, &ctx); + ctx->progress.extract.completed_bytes = + ctx->progress.extract.total_bytes; + ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS); if (ret) - goto out_free_realtarget; - if (progress_func) - progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END, - &ctx.progress); + goto out_cleanup; } - /* If the total number of bytes to extract was miscalculated, just jump - * to the calculated number in order to avoid confusing the progress - * function. This should only occur when extracting from a pipe. */ - if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes) - { - DEBUG("Calculated %"PRIu64" bytes to extract, but actually " - "extracted %"PRIu64, - ctx.progress.extract.total_bytes, - ctx.progress.extract.completed_bytes); - } - if (progress_func && - ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes) - { - ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes; - progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress); - } - - /* Apply security descriptors and timestamps. This is done at the end, - * and in a depth-first manner, to prevent timestamps from getting - * changed by subsequent extract operations and to minimize the chance - * of the restored security descriptors getting in our way. */ - if (progress_func) - progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS, - &ctx.progress); - ret = extract_final_metadata(&dentry_list, &ctx); - if (ret) - goto out_free_realtarget; - - if (progress_func) { - int msg; - if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) - msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END; - else - msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END; - progress_func(msg, &ctx.progress); - } - do_extract_warnings(&ctx); - ret = 0; -out_free_realtarget: - FREE(ctx.realtarget); -out_destroy_stream_list: - if (!(ctx.extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER)) - destroy_stream_list(&ctx.stream_list); -out_finish_or_abort_extract: - if (ret) { - if (ctx.ops->abort_extract) - ctx.ops->abort_extract(&ctx); - } else { - if (ctx.ops->finish_extract) - ret = ctx.ops->finish_extract(&ctx); - } -out_destroy_dentry_list: + ret = extract_progress(ctx, + ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ? + WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END : + WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END)); +out_cleanup: + destroy_blob_list(&ctx->blob_list); destroy_dentry_list(&dentry_list); + FREE(ctx); +out: return ret; } static int mkdir_if_needed(const tchar *target) { - struct stat stbuf; - if (tstat(target, &stbuf)) { - if (errno == ENOENT) { - if (tmkdir(target, 0755)) { - ERROR_WITH_ERRNO("Failed to create directory " - "\"%"TS"\"", target); - return WIMLIB_ERR_MKDIR; - } - } else { - ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target); - return WIMLIB_ERR_STAT; - } - } else if (!S_ISDIR(stbuf.st_mode)) { - ERROR("\"%"TS"\" is not a directory", target); - return WIMLIB_ERR_NOTDIR; - } - return 0; + if (!tmkdir(target, 0755)) + return 0; + + if (errno == EEXIST) + return 0; + +#ifdef __WIN32__ + /* _wmkdir() fails with EACCES if called on a drive root directory. */ + if (errno == EACCES) + return 0; +#endif + + ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target); + return WIMLIB_ERR_MKDIR; } /* Make sure the extraction flags make sense, and update them if needed. */ @@ -2754,11 +1454,6 @@ check_extract_flags(const WIMStruct *wim, int *extract_flags_p) int extract_flags = *extract_flags_p; /* Check for invalid flag combinations */ - if ((extract_flags & - (WIMLIB_EXTRACT_FLAG_SYMLINK | - WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK | - WIMLIB_EXTRACT_FLAG_HARDLINK)) - return WIMLIB_ERR_INVALID_PARAM; if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NO_ACLS | @@ -2772,25 +1467,52 @@ check_extract_flags(const WIMStruct *wim, int *extract_flags_p) 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; - #ifndef WITH_NTFS_3G if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) { ERROR("wimlib was compiled without support for NTFS-3g, so\n" - " it cannot apply a WIM image directly to a NTFS volume."); + " it cannot apply a WIM image directly to an NTFS volume."); return WIMLIB_ERR_UNSUPPORTED; } #endif -#ifndef __WIN32__ if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) { +#ifdef __WIN32__ + if (!wim->filename) + return WIMLIB_ERR_NO_FILENAME; +#else ERROR("WIMBoot extraction is only supported on Windows!"); return WIMLIB_ERR_UNSUPPORTED; - } #endif + } + + if (extract_flags & (WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K | + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K | + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K | + WIMLIB_EXTRACT_FLAG_COMPACT_LZX)) + { + #ifdef __WIN32__ + int count = 0; + count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K) != 0); + count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K) != 0); + count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K) != 0); + count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_LZX) != 0); + if (count != 1) { + ERROR("Only one compression format can be specified " + "for compact-mode extraction!"); + return WIMLIB_ERR_INVALID_PARAM; + } + if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) { + ERROR("Compact-mode extraction and WIMBoot-mode " + "extraction are mutually exclusive!"); + return WIMLIB_ERR_INVALID_PARAM; + } + #else + ERROR("Compact-mode extraction (System Compression) " + "is only supported on Windows!"); + return WIMLIB_ERR_UNSUPPORTED; + #endif + } + if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX | WIMLIB_EXTRACT_FLAG_NORPFIX | @@ -2803,46 +1525,10 @@ check_extract_flags(const WIMStruct *wim, int *extract_flags_p) extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX; } - /* TODO: Since UNIX data entries are stored in the file resources, in a - * completely sequential extraction they may come up before the - * corresponding file or symbolic link data. This needs to be handled - * better. */ - if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA | - WIMLIB_EXTRACT_FLAG_FILE_ORDER)) - == WIMLIB_EXTRACT_FLAG_UNIX_DATA) - { - if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) { - WARNING("Setting UNIX file/owner group may " - "be impossible on some\n" - " symbolic links " - "when applying from a pipe."); - } else { - extract_flags |= WIMLIB_EXTRACT_FLAG_FILE_ORDER; - WARNING("Disabling sequential extraction for " - "UNIX data mode"); - } - } - *extract_flags_p = extract_flags; return 0; } -static u32 -get_wildcard_flags(int extract_flags) -{ - u32 wildcard_flags = 0; - - if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB) - wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH; - else - wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH; - - if (default_ignore_case) - wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE; - - return wildcard_flags; -} - struct append_dentry_ctx { struct wim_dentry **dentries; size_t num_dentries; @@ -2871,10 +1557,35 @@ append_dentry_cb(struct wim_dentry *dentry, void *_ctx) return 0; } +/* Append dentries matched by a path which can contain wildcard characters. */ +static int +append_matched_dentries(WIMStruct *wim, const tchar *orig_pattern, + int extract_flags, struct append_dentry_ctx *ctx) +{ + const size_t count_before = ctx->num_dentries; + tchar *pattern; + int ret; + + pattern = canonicalize_wim_path(orig_pattern); + if (!pattern) + return WIMLIB_ERR_NOMEM; + ret = expand_path_pattern(wim_get_current_root_dentry(wim), pattern, + append_dentry_cb, ctx); + FREE(pattern); + if (ret || ctx->num_dentries > count_before) + return ret; + if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB) { + ERROR("No matches for path pattern \"%"TS"\"", orig_pattern); + return WIMLIB_ERR_PATH_DOES_NOT_EXIST; + } + WARNING("No matches for path pattern \"%"TS"\"", orig_pattern); + return 0; +} + static int do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target, const tchar * const *paths, size_t num_paths, - int extract_flags, wimlib_progress_func_t progress_func) + int extract_flags) { int ret; struct wim_dentry **trees; @@ -2892,7 +1603,7 @@ do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target, if (ret) return ret; - ret = wim_checksum_unhashed_streams(wim); + ret = wim_checksum_unhashed_blobs(wim); if (ret) return ret; @@ -2913,20 +1624,10 @@ do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target, .num_alloc_dentries = 0, }; - u32 wildcard_flags = get_wildcard_flags(extract_flags); - for (size_t i = 0; i < num_paths; i++) { - tchar *path = canonicalize_wim_path(paths[i]); - if (path == NULL) { - ret = WIMLIB_ERR_NOMEM; - trees = append_dentry_ctx.dentries; - goto out_free_trees; - } - ret = expand_wildcard(wim, path, - append_dentry_cb, - &append_dentry_ctx, - wildcard_flags); - FREE(path); + ret = append_matched_dentries(wim, paths[i], + extract_flags, + &append_dentry_ctx); if (ret) { trees = append_dentry_ctx.dentries; goto out_free_trees; @@ -2966,8 +1667,7 @@ do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target, goto out_free_trees; } - ret = extract_trees(wim, trees, num_trees, - target, extract_flags, progress_func); + ret = extract_trees(wim, trees, num_trees, target, extract_flags); out_free_trees: FREE(trees); return ret; @@ -2975,13 +1675,11 @@ out_free_trees: static int extract_single_image(WIMStruct *wim, int image, - const tchar *target, int extract_flags, - wimlib_progress_func_t progress_func) + const tchar *target, int extract_flags) { const tchar *path = WIMLIB_WIM_ROOT_PATH; extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE; - return do_wimlib_extract_paths(wim, image, target, &path, 1, - extract_flags, progress_func); + return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags); } static const tchar * const filename_forbidden_chars = @@ -3007,10 +1705,7 @@ 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 *wim, - const tchar *target, - int extract_flags, - wimlib_progress_func_t progress_func) +extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags) { size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20); size_t output_path_len = tstrlen(target); @@ -3019,8 +1714,6 @@ extract_all_images(WIMStruct *wim, int image; const tchar *image_name; - extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE; - if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) { ERROR("Cannot extract multiple images in NTFS extraction mode."); return WIMLIB_ERR_INVALID_PARAM; @@ -3040,46 +1733,26 @@ extract_all_images(WIMStruct *wim, * Use image number instead. */ tsprintf(buf + output_path_len + 1, T("%d"), image); } - ret = extract_single_image(wim, image, buf, extract_flags, - progress_func); + ret = extract_single_image(wim, image, buf, extract_flags); if (ret) return ret; } return 0; } -static void -clear_lte_extracted_file(WIMStruct *wim, int extract_flags) -{ - if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK | - WIMLIB_EXTRACT_FLAG_HARDLINK))) - for_lookup_table_entry(wim->lookup_table, - lte_free_extracted_file, NULL); -} - static int -do_wimlib_extract_image(WIMStruct *wim, - int image, - const tchar *target, - int extract_flags, - wimlib_progress_func_t progress_func) +do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target, + int extract_flags) { - int ret; - if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE | WIMLIB_EXTRACT_FLAG_TO_STDOUT | WIMLIB_EXTRACT_FLAG_GLOB_PATHS)) return WIMLIB_ERR_INVALID_PARAM; if (image == WIMLIB_ALL_IMAGES) - ret = extract_all_images(wim, target, extract_flags, - progress_func); + return extract_all_images(wim, target, extract_flags); else - ret = extract_single_image(wim, image, target, extract_flags, - progress_func); - - clear_lte_extracted_file(wim, extract_flags); - return ret; + return extract_single_image(wim, image, target, extract_flags); } @@ -3090,23 +1763,18 @@ do_wimlib_extract_image(WIMStruct *wim, WIMLIBAPI int wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target, const tchar * const *paths, size_t num_paths, - int extract_flags, wimlib_progress_func_t progress_func) + int extract_flags) { - int ret; - if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC) return WIMLIB_ERR_INVALID_PARAM; - ret = do_wimlib_extract_paths(wim, image, target, paths, num_paths, - extract_flags, progress_func); - clear_lte_extracted_file(wim, extract_flags); - return ret; + return do_wimlib_extract_paths(wim, image, target, paths, num_paths, + extract_flags); } WIMLIBAPI int wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target, - const tchar *path_list_file, int extract_flags, - wimlib_progress_func_t progress_func) + const tchar *path_list_file, int extract_flags) { int ret; tchar **paths; @@ -3122,16 +1790,19 @@ wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target, ret = wimlib_extract_paths(wim, image, target, (const tchar * const *)paths, num_paths, - extract_flags, progress_func); + extract_flags); FREE(paths); FREE(mem); return ret; } WIMLIBAPI int -wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, - const tchar *target, int extract_flags, - wimlib_progress_func_t progress_func) +wimlib_extract_image_from_pipe_with_progress(int pipe_fd, + const tchar *image_num_or_name, + const tchar *target, + int extract_flags, + wimlib_progress_func_t progfunc, + void *progctx) { int ret; WIMStruct *pwm; @@ -3142,16 +1813,12 @@ wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC) return WIMLIB_ERR_INVALID_PARAM; - if (extract_flags & WIMLIB_EXTRACT_FLAG_FILE_ORDER) - return WIMLIB_ERR_INVALID_PARAM; - /* Read the WIM header from the pipe and get a WIMStruct to represent * the pipable WIM. Caveats: Unlike getting a WIMStruct with - * wimlib_open_wim(), getting a WIMStruct in this way will result in - * an empty lookup table, no XML data read, and no filename set. */ - ret = open_wim_as_WIMStruct(&pipe_fd, - WIMLIB_OPEN_FLAG_FROM_PIPE, - &pwm, progress_func); + * wimlib_open_wim(), getting a WIMStruct in this way will result in an + * empty blob table, no XML data read, and no filename set. */ + ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm, + progfunc, progctx); if (ret) return ret; @@ -3181,22 +1848,19 @@ wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, * write_pipable_wim() for more details about the format of pipable * WIMs.) */ { - struct wim_lookup_table_entry xml_lte; - struct wim_resource_spec xml_rspec; - ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL); + u8 hash[SHA1_HASH_SIZE]; + + ret = read_pwm_blob_header(pwm, hash, + &pwm->hdr.xml_data_reshdr, NULL); if (ret) goto out_wimlib_free; - if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA)) - { - ERROR("Expected XML data, but found non-metadata " - "stream."); + if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) { + ERROR("Expected XML data, but found non-metadata resource."); ret = WIMLIB_ERR_INVALID_PIPABLE_WIM; goto out_wimlib_free; } - wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr); - ret = read_wim_xml_data(pwm); if (ret) goto out_wimlib_free; @@ -3234,68 +1898,80 @@ wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, /* Load the needed metadata resource. */ for (i = 1; i <= pwm->hdr.image_count; i++) { - struct wim_lookup_table_entry *metadata_lte; struct wim_image_metadata *imd; - struct wim_resource_spec *metadata_rspec; + struct wim_reshdr reshdr; + struct wim_resource_descriptor *metadata_rdesc; - metadata_lte = new_lookup_table_entry(); - if (metadata_lte == NULL) { - ret = WIMLIB_ERR_NOMEM; - goto out_wimlib_free; - } - metadata_rspec = MALLOC(sizeof(struct wim_resource_spec)); - if (metadata_rspec == NULL) { - ret = WIMLIB_ERR_NOMEM; - free_lookup_table_entry(metadata_lte); + imd = pwm->image_metadata[i - 1]; + + ret = WIMLIB_ERR_NOMEM; + imd->metadata_blob = new_blob_descriptor(); + if (!imd->metadata_blob) goto out_wimlib_free; - } - ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL); - imd = pwm->image_metadata[i - 1]; - imd->metadata_lte = metadata_lte; - if (ret) { - FREE(metadata_rspec); + imd->metadata_blob->is_metadata = 1; + + ret = read_pwm_blob_header(pwm, imd->metadata_blob->hash, + &reshdr, NULL); + if (ret) goto out_wimlib_free; - } - if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) { + if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) { ERROR("Expected metadata resource, but found " - "non-metadata stream."); + "non-metadata resource"); ret = WIMLIB_ERR_INVALID_PIPABLE_WIM; goto out_wimlib_free; } + ret = WIMLIB_ERR_NOMEM; + metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor)); + if (!metadata_rdesc) + goto out_wimlib_free; + wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc, + imd->metadata_blob); + if (i == image) { /* Metadata resource is for the image being extracted. * Parse it and save the metadata in memory. */ - ret = read_metadata_resource(pwm, imd); + ret = read_metadata_resource(imd); if (ret) goto out_wimlib_free; imd->modified = 1; } else { /* Metadata resource is not for the image being * extracted. Skip over it. */ - ret = skip_wim_stream(metadata_lte); + ret = skip_wim_resource(metadata_rdesc); if (ret) goto out_wimlib_free; } } /* Extract the image. */ extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE; - ret = do_wimlib_extract_image(pwm, image, target, - extract_flags, progress_func); + ret = do_wimlib_extract_image(pwm, image, target, extract_flags); /* Clean up and return. */ out_wimlib_free: wimlib_free(pwm); return ret; } + +WIMLIBAPI int +wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name, + const tchar *target, int extract_flags) +{ + return wimlib_extract_image_from_pipe_with_progress(pipe_fd, + image_num_or_name, + target, + extract_flags, + NULL, + NULL); +} + WIMLIBAPI int wimlib_extract_image(WIMStruct *wim, int image, const tchar *target, - int extract_flags, wimlib_progress_func_t progress_func) + int extract_flags) { if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC) return WIMLIB_ERR_INVALID_PARAM; - return do_wimlib_extract_image(wim, image, target, extract_flags, - progress_func); + return do_wimlib_extract_image(wim, image, target, extract_flags); }