gh-107398: Fix tarfile stream mode exception when process the file with the gzip extra field (GH-126304)

(cherry picked from commit 65f9932)

Co-authored-by: Nadeshiko Manju <me@manjusaka.me>
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>

Author: 2 people

Lib/tarfile.py

@@ -489,7 +489,7 @@ def _init_read_gz(self):
 
         if flag & 4:
             xlen = ord(self.__read(1)) + 256 * ord(self.__read(1))
-            self.read(xlen)
+            self.__read(xlen)
         if flag & 8:
             while True:
                 s = self.__read(1)

Lib/test/test_tarfile.py

@@ -884,10 +884,39 @@ def test_extractall_hardlink_on_symlink(self):
                 self._assert_on_file_content(hardlink_filepath, sha256_regtype)
 
 
+class GzipReadTestBase:
+
+    def test_read_with_extra_field(self):
+        with open(self.tarname, 'rb') as f:
+            data = bytearray(f.read())
+        flags = data[3]
+        self.assertEqual(flags, 8)
+        data[3] = flags | 4
+        data[10:10] = b'\x05\x00extra'
+        with open(tmpname, 'wb') as f:
+            f.write(data)
+        print(self.mode)
+        with tarfile.open(tmpname, mode=self.mode):
+            pass
+
+    def test_read_with_file_comment(self):
+        with open(self.tarname, 'rb') as f:
+            data = bytearray(f.read())
+        flags = data[3]
+        self.assertEqual(flags, 8)
+        data[3] = flags | 16
+        i = data.index(0, 10) + 1
+        data[i:i] = b'comment\x00'
+        with open(tmpname, 'wb') as f:
+            f.write(data)
+        with tarfile.open(tmpname, mode=self.mode):
+            pass
+
+
 class MiscReadTest(MiscReadTestBase, unittest.TestCase):
     test_fail_comp = None
 
-class GzipMiscReadTest(GzipTest, MiscReadTestBase, unittest.TestCase):
+class GzipMiscReadTest(GzipTest, GzipReadTestBase, MiscReadTestBase, unittest.TestCase):
     pass
 
 class Bz2MiscReadTest(Bz2Test, MiscReadTestBase, unittest.TestCase):
@@ -959,7 +988,7 @@ def test_compare_members(self):
         finally:
             tar1.close()
 
-class GzipStreamReadTest(GzipTest, StreamReadTest):
+class GzipStreamReadTest(GzipTest, GzipReadTestBase, StreamReadTest):
     pass
 
 class Bz2StreamReadTest(Bz2Test, StreamReadTest):

Misc/NEWS.d/next/Library/2024-11-02-02-02-31.gh-issue-107398.uUtA6Q.rst

@@ -0,0 +1 @@
+Fix :mod:`tarfile` stream mode exception when process the file with the gzip extra field.