X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fextract.c;h=c1a5adb62aa095417b3b74874385c96456dfa86e;hp=5b1c1cc609de5edf5fbc565d6ed808a1a791603e;hb=6a84b7d59aa7cc0509ac870e5cbf122c5d755781;hpb=3de1ec66f778edda19865482d685bc6f4e17faf7 diff --git a/src/extract.c b/src/extract.c index 5b1c1cc6..c1a5adb6 100644 --- a/src/extract.c +++ b/src/extract.c @@ -54,16 +54,14 @@ #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" #include "wimlib/unix_data.h" -#ifdef __WIN32__ -# include "wimlib/win32.h" /* for realpath() equivalent */ -#endif -#include "wimlib/xml.h" -#include "wimlib/wildcard.h" #include "wimlib/wim.h" +#include "wimlib/win32.h" /* for realpath() equivalent */ +#include "wimlib/xml.h" #define WIMLIB_EXTRACT_FLAG_FROM_PIPE 0x80000000 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE 0x40000000 @@ -86,7 +84,12 @@ WIMLIB_EXTRACT_FLAG_STRICT_GLOB | \ WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES | \ WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE | \ - WIMLIB_EXTRACT_FLAG_WIMBOOT) + WIMLIB_EXTRACT_FLAG_WIMBOOT | \ + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K | \ + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K | \ + WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K | \ + WIMLIB_EXTRACT_FLAG_COMPACT_LZX \ + ) /* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA. */ @@ -98,7 +101,7 @@ do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg) } static int -start_file_phase(struct apply_ctx *ctx, uint64_t end_file_count, enum wimlib_progress_msg msg) +start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg) { ctx->progress.extract.current_file_count = 0; ctx->progress.extract.end_file_count = end_file_count; @@ -106,13 +109,13 @@ start_file_phase(struct apply_ctx *ctx, uint64_t end_file_count, enum wimlib_pro } int -start_file_structure_phase(struct apply_ctx *ctx, uint64_t end_file_count) +start_file_structure_phase(struct apply_ctx *ctx, u64 end_file_count) { return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE); } int -start_file_metadata_phase(struct apply_ctx *ctx, uint64_t end_file_count) +start_file_metadata_phase(struct apply_ctx *ctx, u64 end_file_count) { return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA); } @@ -136,158 +139,124 @@ end_file_metadata_phase(struct apply_ctx *ctx) return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA); } -#define PWM_ALLOW_WIM_HDR 0x00001 +#define PWM_FOUND_WIM_HDR (-1) -/* Read the header for a blob in a pipable WIM. */ +/* 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 -read_pwm_blob_header(WIMStruct *pwm, struct blob_descriptor *blob, - struct wim_resource_descriptor *rdesc, - int flags, struct wim_header_disk *hdr_ret) +read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE], + struct wim_reshdr *reshdr_ret, + struct wim_header_disk *pwm_hdr_ret) { - union { - struct pwm_blob_hdr blob_hdr; - struct wim_header_disk pwm_hdr; - } buf; - struct wim_reshdr reshdr; int ret; + struct pwm_blob_hdr blob_hdr; + u64 magic; - ret = full_read(&pwm->in_fd, &buf.blob_hdr, sizeof(buf.blob_hdr)); - if (ret) + ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr)); + if (unlikely(ret)) goto read_error; - if ((flags & PWM_ALLOW_WIM_HDR) && - le64_to_cpu(buf.blob_hdr.magic) == PWM_MAGIC) - { - BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.blob_hdr)); - ret = full_read(&pwm->in_fd, &buf.blob_hdr + 1, - sizeof(buf.pwm_hdr) - sizeof(buf.blob_hdr)); + magic = le64_to_cpu(blob_hdr.magic); - if (ret) + 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; - blob->blob_location = BLOB_NONEXISTENT; - memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr)); - return 0; + return PWM_FOUND_WIM_HDR; } - if (le64_to_cpu(buf.blob_hdr.magic) != PWM_BLOB_MAGIC) { - ERROR("Data read on pipe is invalid (expected blob header)."); + if (unlikely(magic != PWM_BLOB_MAGIC)) { + ERROR("Data read on pipe is invalid (expected blob header)"); return WIMLIB_ERR_INVALID_PIPABLE_WIM; } - copy_hash(blob->hash, buf.blob_hdr.hash); + copy_hash(hash_ret, blob_hdr.hash); + + 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); + + 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; + } - reshdr.size_in_wim = 0; - reshdr.flags = le32_to_cpu(buf.blob_hdr.flags); - reshdr.offset_in_wim = pwm->in_fd.offset; - reshdr.uncompressed_size = le64_to_cpu(buf.blob_hdr.uncompressed_size); - wim_res_hdr_to_desc(&reshdr, pwm, rdesc); - blob_set_is_located_in_wim_resource(blob, rdesc); - blob->flags = rdesc->flags; - blob->size = rdesc->uncompressed_size; - blob->offset_in_res = 0; return 0; read_error: - ERROR_WITH_ERRNO("Error reading pipable WIM from pipe"); + 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 -read_blobs_from_pipe(struct apply_ctx *ctx, - const struct read_blob_list_callbacks *cbs) +read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs) { - struct blob_descriptor *found_blob = NULL; - struct wim_resource_descriptor *rdesc = NULL; - struct blob_table *blob_table; 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; - ret = WIMLIB_ERR_NOMEM; - found_blob = new_blob_descriptor(); - if (!found_blob) - goto out; - - rdesc = MALLOC(sizeof(struct wim_resource_descriptor)); - if (!rdesc) - goto out; - - blob_table = ctx->wim->blob_table; - memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GUID_LEN); + 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) - goto out; + return ret; while (ctx->num_blobs_remaining) { - struct wim_header_disk pwm_hdr; - struct blob_descriptor *needed_blob; - if (found_blob->blob_location != BLOB_NONEXISTENT) - blob_unset_is_located_in_wim_resource(found_blob); - ret = read_pwm_blob_header(ctx->wim, found_blob, rdesc, - PWM_ALLOW_WIM_HDR, &pwm_hdr); - if (ret) - goto out; + ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr); - if ((found_blob->blob_location != BLOB_NONEXISTENT) - && !(found_blob->flags & WIM_RESHDR_FLAG_METADATA) - && (needed_blob = lookup_blob(blob_table, found_blob->hash)) - && (needed_blob->out_refcnt)) - { - needed_blob->offset_in_res = found_blob->offset_in_res; - needed_blob->flags = found_blob->flags; - needed_blob->size = found_blob->size; + 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; - blob_unset_is_located_in_wim_resource(found_blob); - blob_set_is_located_in_wim_resource(needed_blob, rdesc); + 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; - ret = (*cbs->begin_blob)(needed_blob, - cbs->begin_blob_ctx); - if (ret) { - blob_unset_is_located_in_wim_resource(needed_blob); - goto out; - } + continue; + } - ret = extract_blob(needed_blob, needed_blob->size, - cbs->consume_chunk, - cbs->consume_chunk_ctx); + if (ret) + return ret; - ret = (*cbs->end_blob)(needed_blob, ret, - cbs->end_blob_ctx); - blob_unset_is_located_in_wim_resource(needed_blob); + 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) - goto out; + return ret; ctx->num_blobs_remaining--; - } else if (found_blob->blob_location != BLOB_NONEXISTENT) { - ret = skip_wim_resource(found_blob->rdesc); - if (ret) - goto out; } else { - 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 || - memcmp(pwm_hdr.guid, ctx->progress.extract.guid, - WIM_GUID_LEN)) - { - ctx->progress.extract.part_number = part_number; - ctx->progress.extract.total_parts = total_parts; - memcpy(ctx->progress.extract.guid, - pwm_hdr.guid, WIM_GUID_LEN); - ret = extract_progress(ctx, - WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN); - if (ret) - goto out; - } + wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc); + ret = skip_wim_resource(&rdesc); + if (ret) + return ret; } } - ret = 0; -out: - if (found_blob && found_blob->blob_location != BLOB_IN_WIM) - FREE(rdesc); - free_blob_descriptor(found_blob); - return ret; + + return 0; } /* Creates a temporary file opened for writing. The open file descriptor is @@ -339,8 +308,8 @@ begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx) if (unlikely(blob->out_refcnt > MAX_OPEN_FILES)) return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name); - else - return (*ctx->saved_cbs->begin_blob)(blob, ctx->saved_cbs->begin_blob_ctx); + + return call_begin_blob(blob, ctx->saved_cbs); } static int @@ -364,10 +333,7 @@ extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx) const struct wim_inode *inode = targets[i].inode; const struct wim_dentry *dentry; - list_for_each_entry(dentry, - &inode->i_extraction_aliases, - d_extraction_alias_node) - { + 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++; @@ -380,30 +346,9 @@ extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx) if (ret) return ret; - if (progress->extract.completed_bytes >= - progress->extract.total_bytes) - { - ctx->next_progress = UINT64_MAX; - } else { - /* Send new message as soon as another 1/128 of the - * total has been extracted. (Arbitrary number.) */ - ctx->next_progress = - progress->extract.completed_bytes + - progress->extract.total_bytes / 128; - - /* ... Unless that would be more than 5000000 bytes, in - * which case send the next after the next 5000000 - * bytes. (Another arbitrary number.) */ - if (progress->extract.completed_bytes + 5000000 < - ctx->next_progress) - ctx->next_progress = - progress->extract.completed_bytes + 5000000; - - /* ... But always send a message as soon as we're - * completely done. */ - if (progress->extract.total_bytes < ctx->next_progress) - ctx->next_progress = progress->extract.total_bytes; - } + set_next_progress(progress->extract.completed_bytes, + progress->extract.total_bytes, + &ctx->next_progress); } if (unlikely(filedes_valid(&ctx->tmpfile_fd))) { @@ -415,67 +360,36 @@ extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx) ctx->tmpfile_name); } return ret; - } else { - return (*ctx->saved_cbs->consume_chunk)(chunk, size, - ctx->saved_cbs->consume_chunk_ctx); } + + return call_consume_chunk(chunk, size, ctx->saved_cbs); } +/* 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 -extract_from_tmpfile(const tchar *tmpfile_name, struct apply_ctx *ctx) +extract_from_tmpfile(const tchar *tmpfile_name, + const struct blob_descriptor *orig_blob, + const struct read_blob_callbacks *cbs) { struct blob_descriptor tmpfile_blob; - struct blob_descriptor *orig_blob = ctx->cur_blob; - const struct read_blob_list_callbacks *cbs = ctx->saved_cbs; + const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob); int ret; - const u32 orig_refcnt = orig_blob->out_refcnt; - - BUILD_BUG_ON(MAX_OPEN_FILES < - ARRAY_LEN(orig_blob->inline_blob_extraction_targets)); - - struct blob_extraction_target *targets = orig_blob->blob_extraction_targets; - - /* 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! */ memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor)); tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK; - tmpfile_blob.file_on_disk = ctx->tmpfile_name; - ret = 0; - for (u32 i = 0; i < orig_refcnt; i++) { - - /* Note: it usually doesn't matter whether we pass the original - * blob descriptor to callbacks provided by the extraction - * backend as opposed to the tmpfile blob descriptor, since they - * shouldn't actually read data from the blob other than through - * the read_blob_prefix() call below. But for - * WIMLIB_EXTRACT_FLAG_WIMBOOT mode on Windows it does matter - * because it needs access to the original WIM resource - * descriptor in order to create the external backing reference. - */ - - orig_blob->out_refcnt = 1; - orig_blob->inline_blob_extraction_targets[0] = targets[i]; - - ret = (*cbs->begin_blob)(orig_blob, cbs->begin_blob_ctx); - if (ret) - break; - - /* Extra SHA-1 isn't necessary here, but it shouldn't hurt as - * this case is very rare anyway. */ - ret = extract_blob(&tmpfile_blob, tmpfile_blob.size, - cbs->consume_chunk, - cbs->consume_chunk_ctx); + tmpfile_blob.file_on_disk = (tchar *)tmpfile_name; + tmpfile_blob.out_refcnt = 1; - ret = (*cbs->end_blob)(orig_blob, ret, cbs->end_blob_ctx); + 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) - break; + return ret; } - FREE(targets); - orig_blob->out_refcnt = 0; - return ret; + return 0; } static int @@ -486,15 +400,15 @@ end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx) if (unlikely(filedes_valid(&ctx->tmpfile_fd))) { filedes_close(&ctx->tmpfile_fd); if (!status) - status = extract_from_tmpfile(ctx->tmpfile_name, ctx); + 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; - } else { - return (*ctx->saved_cbs->end_blob)(blob, status, - ctx->saved_cbs->end_blob_ctx); } + + return call_end_blob(blob, status, ctx->saved_cbs); } /* @@ -505,27 +419,23 @@ end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx) * * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS. * - * This also works if the WIM is being read from a pipe, whereas attempting to - * read blobs directly (e.g. with read_full_blob_into_buf()) will not. + * 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 nominal destination (e.g. Win32 currently might because the destination - * file system might not support hard links). + * 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_list_callbacks *cbs) +extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs) { - struct read_blob_list_callbacks wrapper_cbs = { - .begin_blob = begin_extract_blob_wrapper, - .begin_blob_ctx = ctx, - .consume_chunk = extract_chunk_wrapper, - .consume_chunk_ctx = ctx, - .end_blob = end_extract_blob_wrapper, - .end_blob_ctx = ctx, + 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) { @@ -569,7 +479,7 @@ extract_dentry_to_stdout(struct wim_dentry *dentry, } filedes_init(&_stdout, STDOUT_FILENO); - return extract_full_blob_to_fd(blob, &_stdout); + return extract_blob_to_fd(blob, &_stdout); } static int @@ -596,14 +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; } @@ -617,23 +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->d_parent; - if (d->tmp_flag) + 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; } @@ -654,8 +564,10 @@ dentry_reset_extraction_list_node(struct wim_dentry *dentry) static int dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore) { - list_del(&dentry->d_extraction_list_node); - 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; } @@ -717,7 +629,7 @@ destroy_dentry_list(struct list_head *dentry_list) dentry_reset_extraction_list_node(dentry); inode->i_visited = 0; inode->i_can_externally_back = 0; - if ((void *)dentry->d_extraction_name != (void *)dentry->file_name) + 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; @@ -748,19 +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 @@ -768,16 +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; } @@ -792,8 +695,8 @@ dentry_calculate_extraction_name(struct wim_dentry *dentry, #ifdef WITH_NTFS_3G if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) { - dentry->d_extraction_name = dentry->file_name; - dentry->d_extraction_name_nchars = dentry->file_name_nbytes / + dentry->d_extraction_name = dentry->d_name; + dentry->d_extraction_name_nchars = dentry->d_name_nbytes / sizeof(utf16lechar); return 0; } @@ -826,12 +729,13 @@ dentry_calculate_extraction_name(struct wim_dentry *dentry, } } - if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) { - ret = utf16le_get_tstr(dentry->file_name, - dentry->file_name_nbytes, + 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, - &dentry->d_extraction_name_nchars); - dentry->d_extraction_name_nchars /= sizeof(tchar); + &nbytes); + dentry->d_extraction_name_nchars = nbytes / sizeof(tchar); return ret; } else { if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES) @@ -851,16 +755,16 @@ 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); const tchar *tchar_name; size_t tchar_nchars; ret = utf16le_get_tstr(utf16_name_copy, - dentry->file_name_nbytes, + dentry->d_name_nbytes, &tchar_name, &tchar_nchars); if (ret) return ret; @@ -877,8 +781,7 @@ out_replace: utf16le_put_tstr(tchar_name); - dentry->d_extraction_name = memdup(fixed_name, - 2 * fixed_name_num_chars + 2); + dentry->d_extraction_name = TSTRDUP(fixed_name); if (!dentry->d_extraction_name) return WIMLIB_ERR_NOMEM; dentry->d_extraction_name_nchars = fixed_name_num_chars; @@ -1075,10 +978,11 @@ ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode, * - file is a directory * - file is encrypted * - backend needs to create the file as UNIX symlink - * - backend will extract the stream as externally backed + * - backend will extract the stream as externally + * backed from the WIM archive itself */ - if (ctx->apply_ops->will_externally_back) { - int ret = (*ctx->apply_ops->will_externally_back)(dentry, ctx); + 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? */ @@ -1155,30 +1059,28 @@ static void dentry_list_build_inode_alias_lists(struct list_head *dentry_list) { struct wim_dentry *dentry; - struct wim_inode *inode; + + 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) { - inode = dentry->d_inode; - if (!inode->i_visited) - INIT_LIST_HEAD(&inode->i_extraction_aliases); - list_add_tail(&dentry->d_extraction_alias_node, - &inode->i_extraction_aliases); - inode->i_visited = 1; + dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias; + dentry->d_inode->i_first_extraction_alias = dentry; } - list_for_each_entry(dentry, dentry_list, d_extraction_list_node) - dentry->d_inode->i_visited = 0; } 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_READONLY) + features->readonly_files++; if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN) features->hidden_files++; if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM) features->system_files++; + if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE) + features->archive_files++; if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED) features->compressed_files++; if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) { @@ -1200,7 +1102,7 @@ inode_tally_features(const struct wim_inode *inode, else features->other_reparse_points++; } - if (inode->i_security_id != -1) + if (inode_has_security_descriptor(inode)) features->security_descriptors++; if (inode_has_unix_data(inode)) features->unix_data++; @@ -1256,9 +1158,11 @@ do_feature_check(const struct wim_features *required_features, /* File attributes. */ if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) { - /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE. - * We're an archive program, so theoretically we can do what we - * want with it. */ + + if (required_features->readonly_files && + !supported_features->readonly_files) + WARNING("Ignoring FILE_ATTRIBUTE_READONLY of %lu files", + required_features->readonly_files); if (required_features->hidden_files && !supported_features->hidden_files) @@ -1270,6 +1174,10 @@ do_feature_check(const struct wim_features *required_features, WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files", required_features->system_files); + /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE. + * We're an archive program, so theoretically we can do what we + * want with it. */ + if (required_features->compressed_files && !supported_features->compressed_files) WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files", @@ -1486,12 +1394,12 @@ extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees, * subtract from this if hard links are * supported by the extraction mode.) */ ctx->progress.extract.total_bytes = - wim_info_get_image_total_bytes(wim->wim_info, - wim->current_image); + xml_get_image_total_bytes(wim->xml_info, + wim->current_image); if (ctx->supported_features.hard_links) { ctx->progress.extract.total_bytes -= - wim_info_get_image_hard_link_bytes(wim->wim_info, - wim->current_image); + xml_get_image_hard_link_bytes(wim->xml_info, + wim->current_image); } } @@ -1585,6 +1493,34 @@ check_extract_flags(const WIMStruct *wim, int *extract_flags_p) #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 | @@ -1601,22 +1537,6 @@ check_extract_flags(const WIMStruct *wim, int *extract_flags_p) 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; @@ -1645,6 +1565,31 @@ 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, @@ -1687,20 +1632,10 @@ 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; @@ -1772,7 +1707,8 @@ image_name_ok_as_dir(const tchar *image_name) return image_name && *image_name && !tstrpbrk(image_name, filename_forbidden_chars) && tstrcmp(image_name, T(".")) && - tstrcmp(image_name, T("..")); + tstrcmp(image_name, T("..")) && + tstrlen(image_name) <= 128; } /* Extracts all images from the WIM to the directory @target, with the images @@ -1780,9 +1716,8 @@ image_name_ok_as_dir(const tchar *image_name) static int 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); - tchar buf[output_path_len + 1 + image_name_max_len + 1]; + tchar buf[output_path_len + 1 + 128 + 1]; int ret; int image; const tchar *image_name; @@ -1921,26 +1856,24 @@ wimlib_extract_image_from_pipe_with_progress(int pipe_fd, * write_pipable_wim() for more details about the format of pipable * WIMs.) */ { - struct blob_descriptor xml_blob; - struct wim_resource_descriptor xml_rdesc; - ret = read_pwm_blob_header(pwm, &xml_blob, &xml_rdesc, 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_blob.flags & WIM_RESHDR_FLAG_METADATA)) - { + 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_desc_to_hdr(&xml_rdesc, &pwm->hdr.xml_data_reshdr); - ret = read_wim_xml_data(pwm); if (ret) goto out_wimlib_free; - if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) { + if (xml_get_image_count(pwm->xml_info) != pwm->hdr.image_count) { ERROR("Image count in XML data is not the same as in WIM header."); ret = WIMLIB_ERR_IMAGE_COUNT; goto out_wimlib_free; @@ -1973,44 +1906,45 @@ wimlib_extract_image_from_pipe_with_progress(int pipe_fd, /* Load the needed metadata resource. */ for (i = 1; i <= pwm->hdr.image_count; i++) { - struct blob_descriptor *metadata_blob; struct wim_image_metadata *imd; + struct wim_reshdr reshdr; struct wim_resource_descriptor *metadata_rdesc; - metadata_blob = new_blob_descriptor(); - if (metadata_blob == NULL) { - ret = WIMLIB_ERR_NOMEM; - goto out_wimlib_free; - } - metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor)); - if (metadata_rdesc == NULL) { - ret = WIMLIB_ERR_NOMEM; - free_blob_descriptor(metadata_blob); + 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_blob_header(pwm, metadata_blob, metadata_rdesc, 0, NULL); - imd = pwm->image_metadata[i - 1]; - imd->metadata_blob = metadata_blob; - if (ret) { - FREE(metadata_rdesc); + 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_blob->flags & WIM_RESHDR_FLAG_METADATA)) { + if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) { ERROR("Expected metadata resource, but found " - "non-metadata resource."); + "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); + pwm->refcnt++; + if (i == image) { /* Metadata resource is for the image being extracted. * Parse it and save the metadata in memory. */ 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. */