]> wimlib.net Git - wimlib/blobdiff - src/extract.c
extract.c: control characters are invalid in filenames on Windows
[wimlib] / src / extract.c
index efce477e0a122cd076799ea0cb7ad3af67f79eed..4f1c54a00df805cef6ff092e009ec32bc35e603a 100644 (file)
 #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"
-#include "wimlib/wildcard.h"
 #include "wimlib/wim.h"
 #include "wimlib/win32.h" /* for realpath() equivalent */
 #include "wimlib/xml.h"
@@ -233,19 +233,18 @@ read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cb
                if (ret)
                        return ret;
 
-               wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc);
-
-               if (!(rdesc.flags & WIM_RESHDR_FLAG_METADATA)
+               if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)
                    && (blob = lookup_blob(ctx->wim->blob_table, hash))
                    && (blob->out_refcnt))
                {
-                       blob_set_is_located_in_nonsolid_wim_resource(blob, &rdesc);
+                       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;
@@ -329,10 +328,7 @@ 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++;
@@ -345,30 +341,9 @@ 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))) {
@@ -385,61 +360,31 @@ extract_chunk_wrapper(const void *chunk, size_t size, void *_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_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_with_cbs() 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 = call_begin_blob(orig_blob, cbs);
-               if (ret)
-                       break;
+       tmpfile_blob.file_on_disk = (tchar *)tmpfile_name;
+       tmpfile_blob.out_refcnt = 1;
 
-               struct read_blob_callbacks wrapper_cbs = {
-                       .consume_chunk  = cbs->consume_chunk,
-                       .ctx            = cbs->ctx,
-               };
-               ret = read_blob_with_cbs(&tmpfile_blob, &wrapper_cbs);
-
-               ret = call_end_blob(orig_blob, ret, cbs);
+       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
@@ -450,7 +395,8 @@ 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);
@@ -555,14 +501,14 @@ 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;
 }
 
@@ -576,23 +522,23 @@ 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;
 }
 
@@ -613,8 +559,10 @@ 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;
 }
 
@@ -676,7 +624,7 @@ 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;
@@ -707,19 +655,20 @@ 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
@@ -727,16 +676,6 @@ 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;
 }
 
@@ -751,8 +690,8 @@ 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;
        }
@@ -785,12 +724,13 @@ 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)
@@ -810,16 +750,16 @@ 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;
@@ -1113,18 +1053,14 @@ 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
@@ -1158,7 +1094,7 @@ 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++;
@@ -1559,22 +1495,6 @@ 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;
@@ -1603,6 +1523,31 @@ 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,
@@ -1645,20 +1590,10 @@ 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;
@@ -1958,9 +1893,8 @@ wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
                metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
                if (!metadata_rdesc)
                        goto out_wimlib_free;
-               wim_reshdr_to_desc(&reshdr, pwm, metadata_rdesc);
-               blob_set_is_located_in_nonsolid_wim_resource(imd->metadata_blob,
-                                                            metadata_rdesc);
+               wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc,
+                                           imd->metadata_blob);
 
                if (i == image) {
                        /* Metadata resource is for the image being extracted.