gh-130819: Update tarfile.py#_create_gnu_long_header to align with GNU Tar

Doc/whatsnew/3.14.rst

@@ -1175,6 +1175,15 @@ sysconfig
   (Contributed by Xuehai Pan in :gh:`131799`.)
 
 
+tarfile
+-------
+
+* Emit ``mode``, ``uname`` and ``gname`` fields for long paths in
+  :mod:`tarfile` archives, providing better bit-for-bit compatibility with GNU
+  ``tar(1)``.
+  (Contributed by Dahan Gong in :gh:`130820`.)
+
+
 threading
 ---------
 

Lib/tarfile.py

@@ -258,6 +258,29 @@ def copyfileobj(src, dst, length=None, exception=OSError, bufsize=None):
         dst.write(buf)
     return
 
+def _get_user_group_names(uid, gid, unames_cache, gnames_cache):
+    # Calls to pwd.getpwuid() and grp.getgrgid() tend to be expensive. To speed
+    # things up, cache the resolved usernames and group names.
+    if pwd:
+        if uid not in unames_cache:
+            try:
+                unames_cache[uid] = pwd.getpwuid(uid)[0]
+            except KeyError:
+                unames_cache[uid] = ''
+        uname = unames_cache[uid]
+    else:
+        uname = None
+    if grp:
+        if gid not in gnames_cache:
+            try:
+                gnames_cache[gid] = grp.getgrgid(gid)[0]
+            except KeyError:
+                gnames_cache[gid] = ''
+        gname = gnames_cache[gid]
+    else:
+        gname = None
+    return uname, gname
+
 def _safe_print(s):
     encoding = getattr(sys.stdout, 'encoding', None)
     if encoding is not None:
@@ -883,6 +906,9 @@ class TarInfo(object):
         _link_target = None,
         )
 
+    _unames = {}    # Cached mappings of uid=0 -> uname
+    _gnames = {}    # Cached mappings of gid=0 -> gname
+
     def __init__(self, name=""):
         """Construct a TarInfo object. name is the optional name
            of the member.
@@ -1190,6 +1216,10 @@ def _create_gnu_long_header(cls, name, type, encoding, errors):
         info["type"] = type
         info["size"] = len(name)
         info["magic"] = GNU_MAGIC
+        info["mode"] = 0o100644
+        uname, gname = _get_user_group_names(0, 0, cls._unames, cls._gnames)
+        info["uname"] = uname or ""
+        info["gname"] = gname or ""
 
         # create extended header + name blocks.
         return cls._create_header(info, USTAR_FORMAT, encoding, errors) + \
@@ -2141,22 +2171,12 @@ def gettarinfo(self, name=None, arcname=None, fileobj=None):
         tarinfo.type = type
         tarinfo.linkname = linkname
 
-        # Calls to pwd.getpwuid() and grp.getgrgid() tend to be expensive. To
-        # speed things up, cache the resolved usernames and group names.
-        if pwd:
-            if tarinfo.uid not in self._unames:
-                try:
-                    self._unames[tarinfo.uid] = pwd.getpwuid(tarinfo.uid)[0]
-                except KeyError:
-                    self._unames[tarinfo.uid] = ''
-            tarinfo.uname = self._unames[tarinfo.uid]
-        if grp:
-            if tarinfo.gid not in self._gnames:
-                try:
-                    self._gnames[tarinfo.gid] = grp.getgrgid(tarinfo.gid)[0]
-                except KeyError:
-                    self._gnames[tarinfo.gid] = ''
-            tarinfo.gname = self._gnames[tarinfo.gid]
+        uname, gname = _get_user_group_names(tarinfo.uid, tarinfo.gid,
+                                             self._unames, self._gnames)
+        if uname != None:
+            tarinfo.uname = uname
+        if gname != None:
+            tarinfo.gname = gname
 
         if type in (CHRTYPE, BLKTYPE):
             if hasattr(os, "major") and hasattr(os, "minor"):

Lib/test/test_tarfile.py

@@ -1878,6 +1878,40 @@ def test_longnamelink_1025(self):
         self._test(("longnam/" * 127) + "longname_",
                    ("longlnk/" * 127) + "longlink_")
 
+    def test_hidden_header_for_longname(self):
+        # Regression test for gh-130819.
+        memory_file = io.BytesIO()
+        tar = tarfile.open(mode="w", fileobj=memory_file, format=tarfile.GNU_FORMAT)
+        tar_info = tarfile.TarInfo("abcdef" * 20)
+        tar_info.type = tarfile.DIRTYPE
+        tar.addfile(tar_info, None)
+        tar.close()
+
+        class RawTabInfo(tarfile.TarInfo):
+
+            def _proc_member(self, tar_file):
+                if self.type in (tarfile.GNUTYPE_LONGNAME, tarfile.GNUTYPE_LONGLINK):
+                    tester.assertEqual(self.mode, 0o644)
+                    unames_cache = RawTabInfo._unames
+                    gnames_cache = RawTabInfo._gnames
+                    if unames_cache:
+                        tester.assertIn(0, unames_cache)
+                    if gnames_cache:
+                        tester.assertIn(0, gnames_cache)
+                    tester.assertEqual(self.uname, unames_cache.get(0, ""))
+                    tester.assertEqual(self.gname, gnames_cache.get(0, ""))
+                return super()._proc_member(tar_file)  # type: ignore
+
+        tester = self
+        memory_file.seek(0)
+        tar = tarfile.open(fileobj=memory_file, mode="r", tarinfo=RawTabInfo)
+        try:
+            members = tar.getmembers()
+            self.assertEqual(len(members), 1)
+        finally:
+            tar.close()
+            memory_file.close()
+
 
 class DeviceHeaderTest(WriteTestBase, unittest.TestCase):
 

Misc/ACKS

@@ -665,6 +665,7 @@ Mikhail Golubev
 Marta Gómez Macías
 Guilherme Gonçalves
 Tiago Gonçalves
+Dahan Gong
 Chris Gonnerman
 Shelley Gooch
 David Goodger

Misc/NEWS.d/next/Library/2025-03-04-03-14-44.gh-issue-130819.Dphgb6.rst

@@ -0,0 +1,3 @@
+Emit ``mode``, ``uname`` and ``gname`` fields for long paths in
+:mod:`tarfile` archives, providing better bit-for-bit compatibility with GNU
+``tar(1)``.