Address review: drop _ prefixes on test helpers, gate on cpython_only, assert against _EXTHEADER_READ_CHUNK, fix _safe_read docstring

Author: iamsharduld

Lib/tarfile.py

@@ -268,7 +268,7 @@ def _safe_read(fileobj, size):
     """Read up to *size* bytes from *fileobj* in bounded chunks.
 
     Returns the same bytes as ``fileobj.read(size)`` would (including a short
-    result at end of file), but never pre-allocates *size* bytes, so an
+    result at end of file), but limits pre-allocation, so an
     oversized size field in a crafted header cannot force a huge allocation.
     """
     if size <= _EXTHEADER_READ_CHUNK:

Lib/test/test_tarfile.py

@@ -550,7 +550,7 @@ def test_extractfile_attrs(self):
             self.assertIs(fobj.seekable(), True)
 
 
-class _ReadSizeRecorder(io.BytesIO):
+class ReadSizeRecorder(io.BytesIO):
     # Records the largest size ever passed to read(), so a test can check
     # that tarfile does not request far more data than the archive holds
     # (which on a real file would pre-allocate it).
@@ -564,36 +564,37 @@ def read(self, size=-1):
         return super().read(size)
 
 
+@support.cpython_only
 class ExtendedHeaderMemoryTest(unittest.TestCase):
     # gh-151497: the size of a GNU long name/link or a pax extended header is
     # read from the archive and is untrusted.  A crafted header can claim a
     # size far larger than the file actually contains; opening such an archive
     # must not try to read (and so pre-allocate) the claimed size in one go.
 
-    def _crafted_archive(self, hdrtype):
+    def crafted_archive(self, hdrtype):
         tarinfo = tarfile.TarInfo("A")
         tarinfo.type = hdrtype
         tarinfo.size = 0xFFFFFFFF  # ~4 GiB claimed in a 512-byte header
         return tarinfo.tobuf(format=tarfile.GNU_FORMAT)
 
-    def _check(self, hdrtype):
-        fobj = _ReadSizeRecorder(self._crafted_archive(hdrtype))
+    def check(self, hdrtype):
+        fobj = ReadSizeRecorder(self.crafted_archive(hdrtype))
         try:
             with tarfile.open(fileobj=fobj, mode="r:") as tar:
                 tar.getmembers()
         except tarfile.ReadError:
             pass  # a truncated header is fine; we only check the allocation
         # The bogus ~4 GiB size must never reach a single read() call.
-        self.assertLess(fobj.max_read_size, 10 * 1024 * 1024)
+        self.assertLessEqual(fobj.max_read_size, tarfile._EXTHEADER_READ_CHUNK)
 
     def test_gnu_longname_oversized_size(self):
-        self._check(tarfile.GNUTYPE_LONGNAME)
+        self.check(tarfile.GNUTYPE_LONGNAME)
 
     def test_gnu_longlink_oversized_size(self):
-        self._check(tarfile.GNUTYPE_LONGLINK)
+        self.check(tarfile.GNUTYPE_LONGLINK)
 
     def test_pax_header_oversized_size(self):
-        self._check(tarfile.XHDTYPE)
+        self.check(tarfile.XHDTYPE)
 
 
 class MiscReadTestBase(CommonReadTest):