#include "wimlib_internal.h"
#include <errno.h>
-#ifdef TCHAR_IS_UTF16LE
-# include <wchar.h>
-#endif
-
/* Calculates the unaligned length, in bytes, of an on-disk WIM dentry that has
* a file name and short name that take the specified numbers of bytes. This
* excludes any alternate data stream entries that may follow the dentry. */
size_t name_utf16le_nbytes;
int ret;
#if TCHAR_IS_UTF16LE
- name_utf16le_nbytes = tstrlen(name) * 2;
- name_utf16le = MALLOC(name_utf16le_nbytes + 2);
+ name_utf16le_nbytes = tstrlen(name) * sizeof(utf16lechar);
+ name_utf16le = MALLOC(name_utf16le_nbytes + sizeof(utf16lechar));
if (!name_utf16le)
return WIMLIB_ERR_NOMEM;
- memcpy(name_utf16le, name, name_utf16le_nbytes + 2);
+ memcpy(name_utf16le, name, name_utf16le_nbytes + sizeof(utf16lechar));
ret = 0;
#else
- ret = tstr_to_utf16le(name, strlen(name), &name_utf16le,
+ ret = tstr_to_utf16le(name, tstrlen(name), &name_utf16le,
&name_utf16le_nbytes);
if (ret == 0) {
if (name_utf16le_nbytes > 0xffff) {
FREE(name_utf16le);
- ERROR("Multibyte string \"%s\" is too long!", name);
+ ERROR("Multibyte string \"%"TS"\" is too long!", name);
ret = WIMLIB_ERR_INVALID_UTF8_STRING;
}
}
}
/* Calculate the full path of @dentry. The full path of its parent must have
- * already been calculated. */
-int
-calculate_dentry_full_path(struct wim_dentry *dentry, void *ignore)
+ * already been calculated, or it must be the root dentry. */
+static int
+calculate_dentry_full_path(struct wim_dentry *dentry)
{
tchar *full_path;
u32 full_path_nbytes;
+ int ret;
- wimlib_assert(dentry_is_root(dentry) ||
- dentry->parent->full_path != NULL);
+ if (dentry->_full_path)
+ return 0;
if (dentry_is_root(dentry)) {
- full_path = TMALLOC(2);
+ full_path = TSTRDUP(T("/"));
if (!full_path)
return WIMLIB_ERR_NOMEM;
- full_path[0] = T('/');
- full_path[1] = T('\0');
full_path_nbytes = 1 * sizeof(tchar);
} else {
+ struct wim_dentry *parent;
tchar *parent_full_path;
u32 parent_full_path_nbytes;
- const struct wim_dentry *parent;
size_t filename_nbytes;
- int ret;
parent = dentry->parent;
if (dentry_is_root(parent)) {
parent_full_path = T("");
parent_full_path_nbytes = 0;
} else {
- parent_full_path = parent->full_path;
+ if (!parent->_full_path) {
+ ret = calculate_dentry_full_path(parent);
+ if (ret)
+ return ret;
+ }
+ parent_full_path = parent->_full_path;
parent_full_path_nbytes = parent->full_path_nbytes;
}
+
+ /* Append this dentry's name as a tchar string to the full path
+ * of the parent followed by the path separator */
#if TCHAR_IS_UTF16LE
filename_nbytes = dentry->file_name_nbytes;
#else
- ret = utf16le_to_mbs_nbytes(dentry->file_name,
- dentry->file_name_nbytes,
- &filename_nbytes);
- if (ret)
- return ret;
+ {
+ int ret = utf16le_to_tstr_nbytes(dentry->file_name,
+ dentry->file_name_nbytes,
+ &filename_nbytes);
+ if (ret)
+ return ret;
+ }
#endif
full_path_nbytes = parent_full_path_nbytes + sizeof(tchar) +
- dentry->file_name_nbytes;
+ filename_nbytes;
full_path = MALLOC(full_path_nbytes + sizeof(tchar));
if (!full_path)
return WIMLIB_ERR_NOMEM;
memcpy(full_path, parent_full_path, parent_full_path_nbytes);
- full_path[parent_full_path_nbytes] = T('/');
+ full_path[parent_full_path_nbytes / sizeof(tchar)] = T('/');
#if TCHAR_IS_UTF16LE
- memcpy(&full_path[parent_full_path_nbytes + 1],
+ memcpy(&full_path[parent_full_path_nbytes / sizeof(tchar) + 1],
dentry->file_name,
- dentry->file_name_nbytes + sizeof(tchar));
+ filename_nbytes + sizeof(tchar));
#else
- utf16le_to_mbs_buf(dentry->file_name,
- dentry->file_name_nbytes,
- &full_path[parent_full_path_nbytes + 1]);
+ utf16le_to_tstr_buf(dentry->file_name,
+ dentry->file_name_nbytes,
+ &full_path[parent_full_path_nbytes /
+ sizeof(tchar) + 1]);
#endif
}
- FREE(dentry->full_path);
- dentry->full_path = full_path;
+ dentry->_full_path = full_path;
dentry->full_path_nbytes= full_path_nbytes;
return 0;
}
+static int
+do_calculate_dentry_full_path(struct wim_dentry *dentry, void *_ignore)
+{
+ return calculate_dentry_full_path(dentry);
+}
+
+int
+calculate_dentry_tree_full_paths(struct wim_dentry *root)
+{
+ return for_dentry_in_tree(root, do_calculate_dentry_full_path, NULL);
+}
+
+tchar *
+dentry_full_path(struct wim_dentry *dentry)
+{
+ calculate_dentry_full_path(dentry);
+ return dentry->_full_path;
+}
+
static int
increment_subdir_offset(struct wim_dentry *dentry, void *subdir_offset_p)
{
int ret;
struct wim_dentry *child;
- ret = tstr_to_utf16le(name, strlen(name),
+ ret = tstr_to_utf16le(name, tstrlen(name) * sizeof(tchar),
&utf16le_name, &utf16le_name_nbytes);
if (ret) {
child = NULL;
struct wim_dentry *cur_dentry, *parent_dentry;
const utf16lechar *p, *pp;
- parent_dentry = wim_root_dentry(w);
+ cur_dentry = parent_dentry = wim_root_dentry(w);
p = path;
while (1) {
while (*p == cpu_to_le16('/'))
p++;
- cur_dentry = parent_dentry;
if (*p == '\0')
break;
pp = p;
return NULL;
dentry = get_dentry_utf16le(w, path_utf16le, path_utf16le_nbytes);
FREE(path_utf16le);
- return cur_dentry;
+ return dentry;
#endif
}
return NULL;
}
+/* Takes in a path of length @len in @buf, and transforms it into a string for
+ * the path of its parent directory. */
+static void
+to_parent_name(tchar *buf, size_t len)
+{
+ ssize_t i = (ssize_t)len - 1;
+ while (i >= 0 && buf[i] == T('/'))
+ i--;
+ while (i >= 0 && buf[i] != T('/'))
+ i--;
+ while (i >= 0 && buf[i] == T('/'))
+ i--;
+ buf[i + 1] = T('\0');
+}
+
/* Returns the dentry that corresponds to the parent directory of @path, or NULL
* if the dentry is not found. */
struct wim_dentry *
/* Prints the full path of a dentry. */
int
-print_dentry_full_path(struct wim_dentry *dentry, void *ignore)
+print_dentry_full_path(struct wim_dentry *dentry, void *_ignore)
{
- if (dentry->full_path)
- printf("%"TS"\n", dentry->full_path);
+ int ret = calculate_dentry_full_path(dentry);
+ if (ret)
+ return ret;
+ tprintf(T("%"TS"\n"), dentry->_full_path);
return 0;
}
wim_timestamp_to_str(inode->i_last_write_time, buf, sizeof(buf));
tprintf(T("Last Write Time = %"TS"\n"), buf);
- tprintf(T("Reparse Tag = 0x%"PRIx32"\n"), inode->i_reparse_tag);
+ if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
+ tprintf(T("Reparse Tag = 0x%"PRIx32"\n"), inode->i_reparse_tag);
+ tprintf(T("Reparse Point Flags = 0x%"PRIx16"\n"),
+ inode->i_not_rpfixed);
+ tprintf(T("Reparse Point Unknown 2 = 0x%"PRIx32"\n"),
+ inode->i_rp_unknown_2);
+ }
+ tprintf(T("Reparse Point Unknown 1 = 0x%"PRIx32"\n"),
+ inode->i_rp_unknown_1);
tprintf(T("Hard Link Group = 0x%"PRIx64"\n"), inode->i_ino);
tprintf(T("Hard Link Group Size = %"PRIu32"\n"), inode->i_nlink);
tprintf(T("Number of Alternate Data Streams = %hu\n"), inode->i_num_ads);
wimlib_printf(T("Filename = \"%"WS"\"\n"), dentry->file_name);
if (dentry_has_short_name(dentry))
wimlib_printf(T("Short Name \"%"WS"\"\n"), dentry->short_name);
- if (dentry->full_path)
- tprintf(T("Full Path = \"%"WS"\"\n"), dentry->full_path);
+ if (dentry->_full_path)
+ tprintf(T("Full Path = \"%"TS"\"\n"), dentry->_full_path);
lte = inode_stream_lte(dentry->d_inode, 0, lookup_table);
if (lte) {
hash = inode_stream_hash(inode, 0);
if (hash) {
tprintf(T("Hash = 0x"));
- print_hash(hash);
+ print_hash(hash, stdout);
tputchar(T('\n'));
tputchar(T('\n'));
}
hash = inode_stream_hash(inode, i + 1);
if (hash) {
tprintf(T("Hash = 0x"));
- print_hash(hash);
+ print_hash(hash, stdout);
tputchar(T('\n'));
}
print_lookup_table_entry(inode_stream_lte(inode, i + 1, lookup_table),
dentry_common_init(struct wim_dentry *dentry)
{
memset(dentry, 0, sizeof(struct wim_dentry));
- dentry->refcnt = 1;
}
-static struct wim_inode *
+struct wim_inode *
new_timeless_inode()
{
struct wim_inode *inode = CALLOC(1, sizeof(struct wim_inode));
if (inode) {
inode->i_security_id = -1;
inode->i_nlink = 1;
- #ifdef WITH_FUSE
inode->i_next_stream_id = 1;
+ inode->i_not_rpfixed = 1;
+ INIT_LIST_HEAD(&inode->i_list);
+ #ifdef WITH_FUSE
if (pthread_mutex_init(&inode->i_mutex, NULL) != 0) {
ERROR_WITH_ERRNO("Error initializing mutex");
FREE(inode);
}
/* Creates an unlinked directory entry. */
-int new_dentry(const tchar *name, struct wim_dentry **dentry_ret)
+int
+new_dentry(const tchar *name, struct wim_dentry **dentry_ret)
{
struct wim_dentry *dentry;
int ret;
ads_entry->stream_name = p;
ads_entry->stream_name_nbytes = name_nbytes;
} else {
- if (name && *(const tchar*)name == T('\0')) {
+ if (name && *(const tchar*)name != T('\0')) {
ret = get_utf16le_name(name, &ads_entry->stream_name,
&ads_entry->stream_name_nbytes);
}
}
/* Frees an inode. */
-void free_inode(struct wim_inode *inode)
+void
+free_inode(struct wim_inode *inode)
{
if (inode) {
if (inode->i_ads_entries) {
wimlib_assert(inode->i_num_opened_fds == 0);
FREE(inode->i_fds);
pthread_mutex_destroy(&inode->i_mutex);
- if (inode->i_hlist.pprev)
- hlist_del(&inode->i_hlist);
#endif
+ /* HACK: This may instead delete the inode from i_list, but the
+ * hlist_del() behaves the same as list_del(). */
+ hlist_del(&inode->i_hlist);
FREE(inode->i_extracted_file);
FREE(inode);
}
/* Decrements link count on an inode and frees it if the link count reaches 0.
* */
-static void put_inode(struct wim_inode *inode)
+static void
+put_inode(struct wim_inode *inode)
{
wimlib_assert(inode->i_nlink != 0);
if (--inode->i_nlink == 0) {
* The corresponding inode (if any) is freed only if its link count is
* decremented to 0.
*/
-void free_dentry(struct wim_dentry *dentry)
+void
+free_dentry(struct wim_dentry *dentry)
{
FREE(dentry->file_name);
FREE(dentry->short_name);
- FREE(dentry->full_path);
+ FREE(dentry->_full_path);
if (dentry->d_inode)
put_inode(dentry->d_inode);
FREE(dentry);
}
-void put_dentry(struct wim_dentry *dentry)
-{
- wimlib_assert(dentry->refcnt != 0);
- if (--dentry->refcnt == 0)
- free_dentry(dentry);
-}
-
/* This function is passed as an argument to for_dentry_in_tree_depth() in order
* to free a directory tree. */
-static int do_free_dentry(struct wim_dentry *dentry, void *__lookup_table)
+static int
+do_free_dentry(struct wim_dentry *dentry, void *__lookup_table)
{
struct wim_lookup_table *lookup_table = __lookup_table;
unsigned i;
lte_decrement_refcnt(lte, lookup_table);
}
}
-
- put_dentry(dentry);
+ free_dentry(dentry);
return 0;
}
* table entries corresponding to the dentries will be
* decremented.
*/
-void free_dentry_tree(struct wim_dentry *root, struct wim_lookup_table *lookup_table)
+void
+free_dentry_tree(struct wim_dentry *root, struct wim_lookup_table *lookup_table)
{
if (root)
for_dentry_in_tree_depth(root, do_free_dentry, lookup_table);
}
-int increment_dentry_refcnt(struct wim_dentry *dentry, void *ignore)
-{
- dentry->refcnt++;
- return 0;
-}
-
/*
* Links a dentry into the directory tree.
*
* @parent: The dentry that will be the parent of @dentry.
* @dentry: The dentry to link.
*/
-bool
+struct wim_dentry *
dentry_add_child(struct wim_dentry * restrict parent,
struct wim_dentry * restrict child)
{
else if (result > 0)
new = &((*new)->rb_right);
else
- return false;
+ return this;
}
child->parent = parent;
rb_link_node(&child->rb_node, rb_parent, new);
rb_insert_color(&child->rb_node, root);
- return true;
+ return NULL;
}
/* Unlink a WIM dentry from the directory entry tree. */
if (inode->i_num_ads == 0) {
return NULL;
} else {
- int ret;
size_t stream_name_utf16le_nbytes;
u16 i;
struct wim_ads_entry *result;
#else
utf16lechar *stream_name_utf16le;
- ret = tstr_to_utf16le(stream_name, tstrlen(stream_name) * sizeof(tchar),
- &stream_name_utf16le,
- &stream_name_utf16le_nbytes);
- if (ret)
- return NULL;
+ {
+ int ret = tstr_to_utf16le(stream_name,
+ tstrlen(stream_name) *
+ sizeof(tchar),
+ &stream_name_utf16le,
+ &stream_name_utf16le_nbytes);
+ if (ret)
+ return NULL;
+ }
#endif
i = 0;
result = NULL;
new_entry = &inode->i_ads_entries[num_ads - 1];
if (init_ads_entry(new_entry, stream_name, stream_name_nbytes, is_utf16le))
return NULL;
-#ifdef WITH_FUSE
new_entry->stream_id = inode->i_next_stream_id++;
-#endif
inode->i_num_ads = num_ads;
return new_entry;
}
return ret;
}
+/* Set the unnamed stream of a WIM inode, given a data buffer containing the
+ * stream contents. */
+int
+inode_set_unnamed_stream(struct wim_inode *inode, const void *data, size_t len,
+ struct wim_lookup_table *lookup_table)
+{
+ struct wim_lookup_table_entry *lte, *existing_lte;
+ u8 hash[SHA1_HASH_SIZE];
+ void *buf;
+
+ sha1_buffer(data, len, hash);
+ existing_lte = __lookup_resource(lookup_table, hash);
+ if (existing_lte) {
+ wimlib_assert(wim_resource_size(existing_lte) == len);
+ lte = existing_lte;
+ lte->refcnt++;
+ } else {
+ lte = new_lookup_table_entry();
+ if (!lte)
+ return WIMLIB_ERR_NOMEM;
+ buf = MALLOC(len);
+ if (!buf) {
+ free_lookup_table_entry(lte);
+ return WIMLIB_ERR_NOMEM;
+ }
+ memcpy(buf, data, len);
+ lte->resource_location = RESOURCE_IN_ATTACHED_BUFFER;
+ lte->attached_buffer = buf;
+ lte->resource_entry.original_size = len;
+ copy_hash(lte->hash, hash);
+ lookup_table_insert(lookup_table, lte);
+ }
+ inode->i_lte = lte;
+ inode->i_resolved = 1;
+ return 0;
+}
+
/* Remove an alternate data stream from a WIM inode */
void
inode_remove_ads(struct wim_inode *inode, u16 idx,
if (size != sizeof(struct wimlib_unix_data))
return BAD_UNIX_DATA;
- ret = read_full_wim_resource(lte, unix_data, 0);
+ ret = read_full_resource_into_buf(lte, unix_data);
if (ret)
return ret;
}
#endif /* !__WIN32__ */
+/* Replace weird characters in filenames and alternate data stream names.
+ *
+ * In particular we do not want the path separator to appear in any names, as
+ * that would make it possible for a "malicious" WIM to extract itself to any
+ * location it wanted to. */
+static void
+replace_forbidden_characters(utf16lechar *name)
+{
+ utf16lechar *p;
+
+ for (p = name; *p; p++) {
+ #ifdef __WIN32__
+ if (wcschr(L"<>:\"/\\|?*", (wchar_t)*p))
+ #else
+ if (*p == cpu_to_le16('/'))
+ #endif
+ {
+ #ifdef __WIN32__
+ *p = cpu_to_le16(0xfffd);
+ #else
+ *p = cpu_to_le16('?');
+ #endif
+ if (name) {
+ WARNING("File, directory, or stream name \"%"WS"\"\n"
+ " contains forbidden characters; "
+ "substituting replacement characters.",
+ name);
+ name = NULL;
+ }
+ }
+ }
+}
+
/*
* Reads the alternate data stream entries of a WIM dentry.
*
}
p = get_u64(p, &length);
- p += 8; /* Skip the reserved field */
+ p = get_u64(p, &cur_entry->unused);
p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash);
p = get_u16(p, &cur_entry->stream_name_nbytes);
get_bytes(p, cur_entry->stream_name_nbytes,
cur_entry->stream_name);
cur_entry->stream_name[cur_entry->stream_name_nbytes / 2] = 0;
+ replace_forbidden_characters(cur_entry->stream_name);
}
/* It's expected that the size of every ADS entry is a multiple
* of 8. However, to be safe, I'm allowing the possibility of
p = get_u32(p, (u32*)&inode->i_security_id);
p = get_u64(p, &dentry->subdir_offset);
- /* 2 unused fields */
- p += 2 * sizeof(u64);
- /*p = get_u64(p, &dentry->unused1);*/
- /*p = get_u64(p, &dentry->unused2);*/
+ p = get_u64(p, &inode->i_unused_1);
+ p = get_u64(p, &inode->i_unused_2);
p = get_u64(p, &inode->i_creation_time);
p = get_u64(p, &inode->i_last_access_time);
p = get_bytes(p, SHA1_HASH_SIZE, inode->i_hash);
- /*
- * I don't know what's going on here. It seems like M$ screwed up the
+ /* I don't know what's going on here. It seems like M$ screwed up the
* reparse points, then put the fields in the same place and didn't
- * document it. The WIM_HDR_FLAG_RP_FIX flag in the WIM header might
- * have something to do with this, but it's not documented.
- */
+ * document it. */
if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
- /* ??? */
- p += 4;
+ p = get_u32(p, &inode->i_rp_unknown_1);
p = get_u32(p, &inode->i_reparse_tag);
- p += 4;
+ p = get_u16(p, &inode->i_rp_unknown_2);
+ p = get_u16(p, &inode->i_not_rpfixed);
} else {
- p = get_u32(p, &inode->i_reparse_tag);
+ p = get_u32(p, &inode->i_rp_unknown_1);
p = get_u64(p, &inode->i_ino);
}
WARNING("File name in WIM dentry \"%"WS"\" is not "
"null-terminated!", file_name);
}
+ replace_forbidden_characters(file_name);
}
/* Align the calculated size */
* u64 reserved1; (always 0)
* u64 reserved2; (always 0)
* };*/
- /*DEBUG("Dentry for file or directory `%s' has %"PRIu64" extra "*/
- /*"bytes of data",*/
- /*file_name_utf8, dentry->length - calculated_size);*/
+ /*DEBUG("Dentry for file or directory `%"WS"' has %"PRIu64" "*/
+ /*"extra bytes of data", file_name,*/
+ /*dentry->length - calculated_size);*/
}
/* Read the short filename if present. Note: if there is no short
WARNING("Short name in WIM dentry \"%"WS"\" is not "
"null-terminated!", file_name);
}
+ replace_forbidden_characters(short_name);
}
/*
p = put_u32(p, inode->i_attributes);
p = put_u32(p, inode->i_security_id);
p = put_u64(p, dentry->subdir_offset);
- p = put_u64(p, 0); /* unused1 */
- p = put_u64(p, 0); /* unused2 */
+ p = put_u64(p, inode->i_unused_1);
+ p = put_u64(p, inode->i_unused_2);
p = put_u64(p, inode->i_creation_time);
p = put_u64(p, inode->i_last_access_time);
p = put_u64(p, inode->i_last_write_time);
hash = inode_stream_hash(inode, 0);
p = put_bytes(p, SHA1_HASH_SIZE, hash);
if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
- p = put_zeroes(p, 4);
+ p = put_u32(p, inode->i_rp_unknown_1);
p = put_u32(p, inode->i_reparse_tag);
- p = put_zeroes(p, 4);
+ p = put_u16(p, inode->i_rp_unknown_2);
+ p = put_u16(p, inode->i_not_rpfixed);
} else {
u64 link_group_id;
- p = put_u32(p, 0);
+ p = put_u32(p, inode->i_rp_unknown_1);
if (inode->i_nlink == 1)
link_group_id = 0;
else
* alternate data stream entries. */
for (u16 i = 0; i < inode->i_num_ads; i++) {
p = put_u64(p, ads_entry_total_length(&inode->i_ads_entries[i]));
- p = put_u64(p, 0); /* Unused */
+ p = put_u64(p, inode->i_ads_entries[i].unused);
hash = inode_stream_hash(inode, i + 1);
p = put_bytes(p, SHA1_HASH_SIZE, hash);
p = put_u16(p, inode->i_ads_entries[i].stream_name_nbytes);