Minimal patch for pip

diff -rU3 pip-orig/src/pip/_internal/utils/unpacking.py pip/src/pip/_internal/utils/unpacking.py
--- pip-orig/src/pip/_internal/utils/unpacking.py	2022-11-05 16:25:43.000000000 +0100
+++ pip/src/pip/_internal/utils/unpacking.py	2023-08-08 13:17:47.705613554 +0200
@@ -184,6 +184,13 @@
                     "outside target directory ({})"
                 )
                 raise InstallationError(message.format(filename, path, location))
+
+            # Call the `data` filter for its side effect (raising exception)
+            try:
+                tarfile.data_filter(member.replace(name=fn), location)
+            except tarfile.LinkOutsideDestinationError:
+                pass
+
             if member.isdir():
                 ensure_dir(path)
             elif member.issym():


Test from https://github.com/pypa/pip/pull/12214

diff -rU3 pip-orig/tests/unit/test_utils_unpacking.py pip/tests/unit/test_utils_unpacking.py
--- pip-orig/tests/unit/test_utils_unpacking.py	2022-11-05 16:25:43.000000000 +0100
+++ pip/tests/unit/test_utils_unpacking.py	2023-08-08 13:17:35.151540108 +0200
@@ -167,6 +167,23 @@
         test_tar = self.make_tar_file('test_tar.tar', files)
         untar_file(test_tar, self.tempdir)
 
+    def test_unpack_tar_filter(self) -> None:
+        """
+        Test that the tarfile.data_filter is used to disallow dangerous
+        behaviour (PEP-721)
+        """
+        test_tar = os.path.join(self.tempdir, "test_tar_filter.tar")
+        with tarfile.open(test_tar, "w") as mytar:
+            file_tarinfo = tarfile.TarInfo("bad-link")
+            file_tarinfo.type = tarfile.SYMTYPE
+            file_tarinfo.linkname = "../../../../pwn"
+            mytar.addfile(file_tarinfo, io.BytesIO(b""))
+        with pytest.raises(InstallationError) as e:
+            untar_file(test_tar, self.tempdir)
+
+        assert "is outside the destination" in str(e.value)
+
+
 
 def test_unpack_tar_unicode(tmpdir):
     test_tar = tmpdir / "test.tar"


Patch for vendored distlib from https://github.com/pypa/distlib/pull/201

diff --git a/distlib/util.py b/distlib/util.py
index e0622e4..4349d0b 100644
--- a/src/pip/_vendor/distlib/util.py
+++ b/src/pip/_vendor/distlib/util.py
@@ -1249,6 +1249,19 @@ def check_path(path):
             for tarinfo in archive.getmembers():
                 if not isinstance(tarinfo.name, text_type):
                     tarinfo.name = tarinfo.name.decode('utf-8')
+
+        # Limit extraction of dangerous items, if this Python
+        # allows it easily. If not, just trust the input.
+        # See: https://docs.python.org/3/library/tarfile.html#extraction-filters
+        def extraction_filter(member, path):
+            """Run tarfile.tar_fillter, but raise the expected ValueError"""
+            # This is only called if the current Python has tarfile filters
+            try:
+                return tarfile.tar_filter(member, path)
+            except tarfile.FilterError as exc:
+                raise ValueError(str(exc))
+        archive.extraction_filter = extraction_filter
+
         archive.extractall(dest_dir)
 
     finally: