diff --git a/Lib/tarfile.py b/Lib/tarfile.py index d1cb4fe15ae27d..223b573f45122c 100644 --- a/Lib/tarfile.py +++ b/Lib/tarfile.py @@ -556,7 +556,9 @@ def seek(self, pos=0): if pos - self.pos >= 0: blocks, remainder = divmod(pos - self.pos, self.bufsize) for i in xrange(blocks): - self.read(self.bufsize) + data = self.read(self.bufsize) + if not data: + break self.read(remainder) else: raise StreamError("seeking backwards is not allowed") diff --git a/Lib/test/test_tarfile.py b/Lib/test/test_tarfile.py index fe66218eaa826f..8fa499f55bda64 100644 --- a/Lib/test/test_tarfile.py +++ b/Lib/test/test_tarfile.py @@ -592,6 +592,22 @@ def test_provoke_stream_error(self): f = self.tar.extractfile(tarinfos[0]) # read the first member self.assertRaises(tarfile.StreamError, f.read) + def test_getmembers_big_size(self): + # gh-151981: A loop in seek() for streaming files tried to read the + # declared number of blocks even at EOF + for format in (tarfile.GNU_FORMAT, tarfile.PAX_FORMAT): + tinfo = tarfile.TarInfo("huge-file") + tinfo.size = 1 << 64 + bio = io.BytesIO() + # Write header without data + bio.write(tinfo.tobuf(format)) + + # Reset & try to get contents + bio.seek(0) + with tarfile.open(fileobj=bio, mode="r|") as tar: + with self.assertRaises(tarfile.ReadError): + tar.getmembers() + def test_compare_members(self): tar1 = tarfile.open(tarname, encoding="iso8859-1") try: