fix: Fix the permission errors in Windows and add back the CVE-2007-4559 Patch changes

.gitignore

@@ -163,6 +163,11 @@ typings/
 # Output of 'npm pack'
 *.tgz
 
+# Except test file
+!tests/functional/testdata/lib/utils/test.tgz
+!tests/functional/testdata/lib/utils/path_reversal_uxix.tgz
+!tests/functional/testdata/lib/utils/path_reversal_win.tgz
+
 # Yarn Integrity file
 .yarn-integrity
 

samcli/lib/utils/tar.py

@@ -2,7 +2,9 @@
 Tarball Archive utility
 """
 
+import os
 import tarfile
+from typing import Union, IO, Optional
 from tempfile import TemporaryFile
 from contextlib import contextmanager
 
@@ -39,3 +41,44 @@ def create_tarball(tar_paths, tar_filter=None, mode="w"):
         yield tarballfile
     finally:
         tarballfile.close()
+
+
+def _is_within_directory(directory: Union[str, os.PathLike], target: Union[str, os.PathLike]) -> bool:
+    """Checks if target is located under directory"""
+    abs_directory = os.path.abspath(directory)
+    abs_target = os.path.abspath(target)
+
+    prefix = os.path.commonprefix([abs_directory, abs_target])
+
+    return bool(prefix == abs_directory)
+
+
+def extract_tarfile(
+    tarfile_path: Union[str, os.PathLike] = "",
+    file_obj: Optional[IO[bytes]] = None,
+    unpack_dir: Union[str, os.PathLike] = "",
+) -> None:
+    """
+    Extracts a tarfile using the provided parameters. If file_obj is specified,
+    it is used instead of the file_obj opened for tarfile_path.
+
+    Parameters
+    ----------
+    tarfile_path Union[str, os.PathLike]
+        Key representing a full path to the file or directory and the Value representing the path within the tarball
+
+    file_obj Optional[IO[bytes]]
+        Object for the tarfile that will be extracted
+
+    unpack_dir Union[str, os.PathLike]
+        The directory where the tarfile members will be extracted.
+    """
+    with tarfile.open(name=tarfile_path, fileobj=file_obj, mode="r") as tar:
+        # Makes sure the tar file is sanitized and is free of directory traversal vulnerability
+        # See: https://github.com/advisories/GHSA-gw9q-c7gh-j9vm
+        for member in tar.getmembers():
+            member_path = os.path.join(unpack_dir, member.name)
+            if not _is_within_directory(unpack_dir, member_path):
+                raise tarfile.ExtractError("Attempted Path Traversal in Tar File")
+
+        tar.extractall(unpack_dir)

samcli/local/docker/container.py

@@ -3,7 +3,6 @@
 """
 import os
 import logging
-import tarfile
 import tempfile
 import threading
 import socket
@@ -14,6 +13,7 @@
 
 from docker.errors import NotFound as DockerNetworkNotFound
 from samcli.lib.utils.retry import retry
+from samcli.lib.utils.tar import extract_tarfile
 from .exceptions import ContainerNotStartableException
 
 from .utils import to_posix_path, find_free_port, NoFreePortsError
@@ -362,7 +362,8 @@ def _can_connect_to_socket(self) -> bool:
         a_socket.close()
         return connection_succeeded
 
-    def copy(self, from_container_path, to_host_path):
+    def copy(self, from_container_path, to_host_path) -> None:
+        """Copies a path from container into host path"""
 
         if not self.is_created():
             raise RuntimeError("Container does not exist. Cannot get logs for this container")
@@ -378,8 +379,7 @@ def copy(self, from_container_path, to_host_path):
             # Seek the handle back to start of file for tarfile to use
             fp.seek(0)
 
-            with tarfile.open(fileobj=fp, mode="r") as tar:
-                tar.extractall(path=to_host_path)
+            extract_tarfile(file_obj=fp, unpack_dir=to_host_path)
 
     @staticmethod
     def _write_container_output(output_itr, stdout=None, stderr=None):

tests/functional/lib/utils/test_tar.py

@@ -0,0 +1,58 @@
+import os
+import tempfile
+import shutil
+import platform
+from pathlib import Path
+from tarfile import ExtractError
+
+from unittest import TestCase
+
+from samcli.lib.utils.tar import extract_tarfile
+
+
+class TestExtractTarFile(TestCase):
+    def test_extract_tarfile_arg_path_unpacks_a_tar(self):
+        test_tar = Path(__file__).resolve().parents[3].joinpath("functional", "testdata", "lib", "utils", "test.tgz")
+        test_dir = tempfile.mkdtemp()
+        extract_tarfile(tarfile_path=test_tar, unpack_dir=test_dir)
+        output_files = set(os.listdir(test_dir))
+        shutil.rmtree(test_dir)
+        self.assertEqual({"test_utils.py"}, output_files)
+
+    def test_raise_exception_for_unsafe_tarfile_with_path_arg(self):
+        tar_filename = "path_reversal_win.tgz" if platform.system().lower() == "windows" else "path_reversal_uxix.tgz"
+        test_tar = Path(__file__).resolve().parents[3].joinpath("functional", "testdata", "lib", "utils", tar_filename)
+        test_dir = tempfile.mkdtemp()
+        self.assertRaisesRegex(
+            ExtractError,
+            "Attempted Path Traversal in Tar File",
+            extract_tarfile,
+            tarfile_path=test_tar,
+            unpack_dir=test_dir,
+        )
+        shutil.rmtree(test_dir)
+
+    def test_extract_tarfile_arg_fileobj_unpacks_a_tar(self):
+        test_tar = Path(__file__).resolve().parents[3].joinpath("functional", "testdata", "lib", "utils", "test.tgz")
+        test_dir = tempfile.mkdtemp()
+        with open(test_tar, mode="rb") as tar:
+            before_extract_output_files = set(os.listdir(test_dir))
+            self.assertEqual(set(), before_extract_output_files)
+            extract_tarfile(file_obj=tar, unpack_dir=test_dir)
+            after_extract_output_files = set(os.listdir(test_dir))
+            self.assertEqual({"test_utils.py"}, after_extract_output_files)
+        shutil.rmtree(test_dir)
+
+    def test_raise_exception_for_unsafe_tarfile_with_flieobj_arg(self):
+        tar_filename = "path_reversal_win.tgz" if platform.system().lower() == "windows" else "path_reversal_uxix.tgz"
+        test_tar = Path(__file__).resolve().parents[3].joinpath("functional", "testdata", "lib", "utils", tar_filename)
+        test_dir = tempfile.mkdtemp()
+        with open(test_tar, mode="rb") as tar:
+            self.assertRaisesRegex(
+                ExtractError,
+                "Attempted Path Traversal in Tar File",
+                extract_tarfile,
+                file_obj=tar,
+                unpack_dir=test_dir,
+            )
+        shutil.rmtree(test_dir)

tests/unit/lib/utils/test_tar.py

@@ -1,7 +1,9 @@
+import io
 from unittest import TestCase
+import tarfile
 from unittest.mock import Mock, patch, call
 
-from samcli.lib.utils.tar import create_tarball
+from samcli.lib.utils.tar import extract_tarfile, create_tarball, _is_within_directory
 
 
 class TestTar(TestCase):
@@ -88,3 +90,72 @@ def tar_filter(tar_info):
         temp_file_mock.seek.assert_called_once_with(0)
         temp_file_mock.close.assert_called_once()
         tarfile_open_patch.assert_called_once_with(fileobj=temp_file_mock, mode="w")
+
+    @patch("samcli.lib.utils.tar.tarfile.open")
+    @patch("samcli.lib.utils.tar._is_within_directory")
+    def test_extract_tarfile_file_name(self, is_within_directory_patch, tarfile_open_patch):
+        tarfile_path = "/test_tarfile_path/"
+        unpack_dir = "/test_unpack_dir/"
+        is_within_directory_patch.return_value = True
+
+        tarfile_file_mock = Mock()
+        tar_file_obj_mock = Mock()
+        tar_file_obj_mock.name = "obj_name"
+        tarfile_file_mock.getmembers.return_value = [tar_file_obj_mock]
+        tarfile_open_patch.return_value.__enter__.return_value = tarfile_file_mock
+
+        extract_tarfile(tarfile_path=tarfile_path, unpack_dir=unpack_dir)
+
+        is_within_directory_patch.assert_called_once()
+        tarfile_file_mock.getmembers.assert_called_once()
+        tarfile_file_mock.extractall.assert_called_once_with(unpack_dir)
+
+    @patch("samcli.lib.utils.tar.tarfile.open")
+    @patch("samcli.lib.utils.tar._is_within_directory")
+    def test_extract_tarfile_fileobj(self, is_within_directory_patch, tarfile_open_patch):
+        stream_str = io.BytesIO(b"Hello World!")
+        unpack_dir = "/test_unpack_dir/"
+        is_within_directory_patch.return_value = True
+
+        tarfile_file_mock = Mock()  # Mock tarfile
+        tar_file_obj_mock = Mock()  # Mock member inside tarfile
+        tar_file_obj_mock.name = "obj_name"
+        tarfile_file_mock.getmembers.return_value = [tar_file_obj_mock]
+        tarfile_open_patch.return_value.__enter__.return_value = tarfile_file_mock
+
+        extract_tarfile(file_obj=stream_str, unpack_dir=unpack_dir)
+
+        is_within_directory_patch.assert_called_once()
+        tarfile_file_mock.getmembers.assert_called_once()
+        tarfile_file_mock.extractall.assert_called_once_with(unpack_dir)
+
+    @patch("samcli.lib.utils.tar.tarfile.open")
+    @patch("samcli.lib.utils.tar._is_within_directory")
+    def test_extract_tarfile_obj_not_within_dir(self, is_within_directory_patch, tarfile_open_patch):
+        tarfile_path = "/test_tarfile_path/"
+        unpack_dir = "/test_unpack_dir/"
+        is_within_directory_patch.return_value = False
+
+        tarfile_file_mock = Mock()
+        tar_file_obj_mock = Mock()
+        tar_file_obj_mock.name = "obj_name"
+        tarfile_file_mock.getmembers.return_value = [tar_file_obj_mock]
+        tarfile_open_patch.return_value.__enter__.return_value = tarfile_file_mock
+
+        with self.assertRaises(tarfile.ExtractError):
+            extract_tarfile(tarfile_path=tarfile_path, unpack_dir=unpack_dir)
+
+        is_within_directory_patch.assert_called_once()
+        tarfile_file_mock.getmembers.assert_called_once()
+
+    def test_tarfile_obj_is_within_dir(self):
+        directory = "/my/path"
+        target = "/my/path/file"
+
+        self.assertTrue(_is_within_directory(directory, target))
+
+    def test_tarfile_obj_is_not_within_dir(self):
+        directory = "/my/path"
+        target = "/another/path/file"
+
+        self.assertFalse(_is_within_directory(directory, target))

tests/unit/local/docker/test_container.py

@@ -839,8 +839,8 @@ def setUp(self):
         self.container.id = "containerid"
 
     @patch("samcli.local.docker.container.tempfile")
-    @patch("samcli.local.docker.container.tarfile")
-    def test_must_copy_files_from_container(self, tarfile_mock, tempfile_mock):
+    @patch("samcli.local.docker.container.extract_tarfile")
+    def test_must_copy_files_from_container(self, extract_tarfile_mock, tempfile_mock):
         source = "source"
         dest = "dest"
 
@@ -853,19 +853,14 @@ def test_must_copy_files_from_container(self, tarfile_mock, tempfile_mock):
         tempfile_ctxmgr.__enter__ = Mock(return_value=fp_mock)
         tempfile_ctxmgr.__exit__ = Mock()
 
-        tarfile_ctxmgr = tarfile_mock.open.return_value = Mock()
-        tar_mock = Mock()
-        tarfile_ctxmgr.return_value.__enter__ = Mock(return_value=tar_mock)
-        tarfile_ctxmgr.return_value.__exit__ = Mock()
-
         self.container.copy(source, dest)
 
+        extract_tarfile_mock.assert_called_with(file_obj=fp_mock, unpack_dir=dest)
+
         # Make sure archive data is written to the file
         fp_mock.write.assert_has_calls([call(x) for x in tar_stream], any_order=False)
 
         # Make sure we open the tarfile right and extract to right location
-        tarfile_mock.open.assert_called_with(fileobj=fp_mock, mode="r")
-        tar_mock.extractall(path=dest)
 
     def test_raise_if_container_is_not_created(self):
         source = "source"