From 613d3a774674f0033e55721935d2af915fc418a2 Mon Sep 17 00:00:00 2001
From: Petr Viktorin <encukou@gmail.com>
Date: Wed, 26 Jun 2024 14:59:49 +0200
Subject: [PATCH] untar_file: remove common leading directory before unpacking

Fixes: #12781
---
 news/12781.bugfix.rst                |  1 +
 src/pip/_internal/utils/unpacking.py | 14 +++++++--
 tests/unit/test_utils_unpacking.py   | 43 ++++++++++++++++++++++++++++
 3 files changed, 56 insertions(+), 2 deletions(-)
 create mode 100644 news/12781.bugfix.rst

diff --git a/news/12781.bugfix.rst b/news/12781.bugfix.rst
new file mode 100644
index 00000000000..6bd43d347db
--- /dev/null
+++ b/news/12781.bugfix.rst
@@ -0,0 +1 @@
+Fix finding hardlink targets in tar files with an ignored top-level directory.
diff --git a/src/pip/_internal/utils/unpacking.py b/src/pip/_internal/utils/unpacking.py
index 341269550ce..875e30e13ab 100644
--- a/src/pip/_internal/utils/unpacking.py
+++ b/src/pip/_internal/utils/unpacking.py
@@ -190,9 +190,19 @@ def untar_file(filename: str, location: str) -> None:
         else:
             default_mode_plus_executable = _get_default_mode_plus_executable()
 
+            if leading:
+                # Strip the leading directory from all files in the archive,
+                # including hardlink targets (which are relative to the
+                # unpack location).
+                for member in tar.getmembers():
+                    name_lead, name_rest = split_leading_dir(member.name)
+                    member.name = name_rest
+                    if member.islnk():
+                        lnk_lead, lnk_rest = split_leading_dir(member.linkname)
+                        if lnk_lead == name_lead:
+                            member.linkname = lnk_rest
+
             def pip_filter(member: tarfile.TarInfo, path: str) -> tarfile.TarInfo:
-                if leading:
-                    member.name = split_leading_dir(member.name)[1]
                 orig_mode = member.mode
                 try:
                     try:
diff --git a/tests/unit/test_utils_unpacking.py b/tests/unit/test_utils_unpacking.py
index 3fdd822e739..50500868061 100644
--- a/tests/unit/test_utils_unpacking.py
+++ b/tests/unit/test_utils_unpacking.py
@@ -197,6 +197,49 @@ def test_unpack_tar_filter(self) -> None:
 
         assert "is outside the destination" in str(e.value)
 
+    @pytest.mark.parametrize(
+        ("input_prefix", "unpack_prefix"),
+        [
+            ("", ""),
+            ("dir/", ""),  # pip ignores a common leading directory
+            ("dir/sub/", "sub/"),  # pip ignores *one* common leading directory
+        ],
+    )
+    def test_unpack_tar_links(self, input_prefix: str, unpack_prefix: str) -> None:
+        """
+        Test unpacking a *.tar with file containing hard & soft links
+        """
+        test_tar = os.path.join(self.tempdir, "test_tar_links.tar")
+        content = b"file content"
+        with tarfile.open(test_tar, "w") as mytar:
+            file_tarinfo = tarfile.TarInfo(input_prefix + "regular_file.txt")
+            file_tarinfo.size = len(content)
+            mytar.addfile(file_tarinfo, io.BytesIO(content))
+
+            hardlink_tarinfo = tarfile.TarInfo(input_prefix + "hardlink.txt")
+            hardlink_tarinfo.type = tarfile.LNKTYPE
+            hardlink_tarinfo.linkname = input_prefix + "regular_file.txt"
+            mytar.addfile(hardlink_tarinfo)
+
+            symlink_tarinfo = tarfile.TarInfo(input_prefix + "symlink.txt")
+            symlink_tarinfo.type = tarfile.SYMTYPE
+            symlink_tarinfo.linkname = "regular_file.txt"
+            mytar.addfile(symlink_tarinfo)
+
+        untar_file(test_tar, self.tempdir)
+
+        os.system(f"ls -alR {self.tempdir}")
+
+        unpack_dir = os.path.join(self.tempdir, unpack_prefix)
+        with open(os.path.join(unpack_dir, "regular_file.txt"), "rb") as f:
+            assert f.read() == content
+
+        with open(os.path.join(unpack_dir, "hardlink.txt"), "rb") as f:
+            assert f.read() == content
+
+        with open(os.path.join(unpack_dir, "symlink.txt"), "rb") as f:
+            assert f.read() == content
+
 
 def test_unpack_tar_unicode(tmpdir: Path) -> None:
     test_tar = tmpdir / "test.tar"