Fix gnutar creation with unicode hardlink names on Windows (libarchiv…

…e#2227)

The code currently uses `archive_entry_hardlink` to determine if an
entry is a hardlink, however on Windows, this call will fail if the path
cannot be represented in the current locale. This instead checks to see
if any entry in the `archive_mstring` is set.

libarchive/archive_entry.c

@@ -526,6 +526,12 @@ archive_entry_hardlink_w(struct archive_entry *entry)
 	return (NULL);
 }
 
+int
+archive_entry_hardlink_is_set(struct archive_entry *entry)
+{
+	return (entry->ae_set & AE_SET_HARDLINK) != 0;
+}
+
 int
 _archive_entry_hardlink_l(struct archive_entry *entry,
     const char **p, size_t *len, struct archive_string_conv *sc)

libarchive/archive_entry.h

@@ -263,6 +263,7 @@ __LA_DECL void		 archive_entry_set_link_to_hardlink(struct archive_entry *);
 __LA_DECL const char	*archive_entry_hardlink(struct archive_entry *);
 __LA_DECL const char	*archive_entry_hardlink_utf8(struct archive_entry *);
 __LA_DECL const wchar_t	*archive_entry_hardlink_w(struct archive_entry *);
+__LA_DECL int		 archive_entry_hardlink_is_set(struct archive_entry *);
 __LA_DECL la_int64_t	 archive_entry_ino(struct archive_entry *);
 __LA_DECL la_int64_t	 archive_entry_ino64(struct archive_entry *);
 __LA_DECL int		 archive_entry_ino_is_set(struct archive_entry *);

libarchive/archive_write_set_format_gnutar.c

@@ -523,7 +523,7 @@ archive_write_gnutar_header(struct archive_write *a,
 			goto exit_write_header;
 	}
 
-	if (archive_entry_hardlink(entry) != NULL) {
+	if (archive_entry_hardlink_is_set(entry)) {
 		tartype = '1';
 	} else
 		switch (archive_entry_filetype(entry)) {

libarchive/test/test_gnutar_filename_encoding.c

@@ -389,3 +389,105 @@ DEFINE_TEST(test_gnutar_filename_encoding_CP932_UTF8)
 	assertEqualMem(buff, "\xE8\xA1\xA8.txt", 7);
 }
 
+DEFINE_TEST(test_gnutar_filename_encoding_UTF16_win)
+{
+#if !defined(_WIN32) || defined(__CYGWIN__)
+	skipping("This test is meant to verify unicode string handling"
+		" on Windows with UTF-16 names");
+	return;
+#else
+	struct archive *a;
+	struct archive_entry *entry;
+	char buff[4096];
+	size_t used;
+
+	/*
+	 * Don't call setlocale because we're verifying that the '_w' functions
+	 * work as expected when 'hdrcharset' is UTF-8
+	 */
+
+	/* Part 1: file */
+	a = archive_write_new();
+	assertEqualInt(ARCHIVE_OK, archive_write_set_format_gnutar(a));
+	if (archive_write_set_options(a, "hdrcharset=UTF-8") != ARCHIVE_OK) {
+		skipping("This system cannot convert character-set"
+		    " from UTF-16 to UTF-8.");
+		archive_write_free(a);
+		return;
+	}
+	assertEqualInt(ARCHIVE_OK,
+	    archive_write_open_memory(a, buff, sizeof(buff), &used));
+
+	entry = archive_entry_new2(a);
+	/* Set the filename using a UTF-16 string */
+	archive_entry_copy_pathname_w(entry, L"\u8868.txt");
+	archive_entry_set_filetype(entry, AE_IFREG);
+	archive_entry_set_size(entry, 0);
+	assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
+	archive_entry_free(entry);
+	assertEqualInt(ARCHIVE_OK, archive_write_free(a));
+
+	/* Check UTF-8 version. */
+	assertEqualMem(buff, "\xE8\xA1\xA8.txt", 7);
+
+	/* Part 2: directory */
+	a = archive_write_new();
+	assertEqualInt(ARCHIVE_OK, archive_write_set_format_gnutar(a));
+	assertEqualInt(ARCHIVE_OK, archive_write_set_options(a, "hdrcharset=UTF-8"));
+	assertEqualInt(ARCHIVE_OK,
+	    archive_write_open_memory(a, buff, sizeof(buff), &used));
+
+	entry = archive_entry_new2(a);
+	/* Set the directory name using a UTF-16 string */
+	/* NOTE: Explicitly not adding trailing slash to test that code path */
+	archive_entry_copy_pathname_w(entry, L"\u8868");
+	archive_entry_set_filetype(entry, AE_IFDIR);
+	archive_entry_set_size(entry, 0);
+	assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
+	archive_entry_free(entry);
+	assertEqualInt(ARCHIVE_OK, archive_write_free(a));
+
+	/* Check UTF-8 version. */
+	assertEqualMem(buff, "\xE8\xA1\xA8/", 4);
+
+	/* Part 3: symlink */
+	a = archive_write_new();
+	assertEqualInt(ARCHIVE_OK, archive_write_set_format_gnutar(a));
+	assertEqualInt(ARCHIVE_OK, archive_write_set_options(a, "hdrcharset=UTF-8"));
+	assertEqualInt(ARCHIVE_OK,
+	    archive_write_open_memory(a, buff, sizeof(buff), &used));
+
+	entry = archive_entry_new2(a);
+	/* Set the symlink target using a UTF-16 string */
+	archive_entry_set_pathname(entry, "link.txt");
+	archive_entry_copy_symlink_w(entry, L"\u8868.txt");
+	archive_entry_set_filetype(entry, AE_IFLNK);
+	archive_entry_set_symlink_type(entry, AE_SYMLINK_TYPE_FILE);
+	archive_entry_set_size(entry, 0);
+	assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
+	archive_entry_free(entry);
+	assertEqualInt(ARCHIVE_OK, archive_write_free(a));
+
+	/* Check UTF-8 version. */
+	assertEqualMem(buff + 157, "\xE8\xA1\xA8.txt", 7);
+
+	/* Part 4: hardlink */
+	a = archive_write_new();
+	assertEqualInt(ARCHIVE_OK, archive_write_set_format_gnutar(a));
+	assertEqualInt(ARCHIVE_OK, archive_write_set_options(a, "hdrcharset=UTF-8"));
+	assertEqualInt(ARCHIVE_OK,
+	    archive_write_open_memory(a, buff, sizeof(buff), &used));
+
+	entry = archive_entry_new2(a);
+	/* Set the symlink target using a UTF-16 string */
+	archive_entry_set_pathname(entry, "link.txt");
+	archive_entry_copy_hardlink_w(entry, L"\u8868.txt");
+	archive_entry_set_size(entry, 0);
+	assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
+	archive_entry_free(entry);
+	assertEqualInt(ARCHIVE_OK, archive_write_free(a));
+
+	/* Check UTF-8 version. */
+	assertEqualMem(buff + 157, "\xE8\xA1\xA8.txt", 7);
+#endif
+}