Issue #1159051: GzipFile now raises EOFError when reading a corrupted file
with truncated header or footer.
Added tests for reading truncated gzip and bzip2 files.
diff --git a/Lib/test/test_bz2.py b/Lib/test/test_bz2.py
index 977d10b..fb104d7 100644
--- a/Lib/test/test_bz2.py
+++ b/Lib/test/test_bz2.py
@@ -295,6 +295,24 @@
self.assertRaises(ValueError, f.readline)
self.assertRaises(ValueError, f.readlines)
+ def test_read_truncated(self):
+ # Drop the eos_magic field (6 bytes) and CRC (4 bytes).
+ truncated = self.DATA[:-10]
+ with open(self.filename, 'wb') as f:
+ f.write(truncated)
+ with BZ2File(self.filename) as f:
+ self.assertRaises(EOFError, f.read)
+ with BZ2File(self.filename) as f:
+ self.assertEqual(f.read(len(self.TEXT)), self.TEXT)
+ self.assertRaises(EOFError, f.read, 1)
+ # Incomplete 4-byte file header, and block header of at least 146 bits.
+ for i in range(22):
+ with open(self.filename, 'wb') as f:
+ f.write(truncated[:i])
+ with BZ2File(self.filename) as f:
+ self.assertRaises(EOFError, f.read, 1)
+
+
class BZ2CompressorTest(BaseTest):
def testCompress(self):
# "Test BZ2Compressor.compress()/flush()"
diff --git a/Lib/test/test_gzip.py b/Lib/test/test_gzip.py
index ced226f..ba9d7da 100644
--- a/Lib/test/test_gzip.py
+++ b/Lib/test/test_gzip.py
@@ -365,6 +365,21 @@
datac = gzip.compress(data)
self.assertEqual(gzip.decompress(datac), data)
+ def test_read_truncated(self):
+ data = data1*50
+ # Drop the CRC (4 bytes) and file size (4 bytes).
+ truncated = gzip.compress(data)[:-8]
+ with gzip.GzipFile(fileobj=io.BytesIO(truncated)) as f:
+ self.assertRaises(EOFError, f.read)
+ with gzip.GzipFile(fileobj=io.BytesIO(truncated)) as f:
+ self.assertEqual(f.read(len(data)), data)
+ self.assertRaises(EOFError, f.read, 1)
+ # Incomplete 10-byte header.
+ for i in range(2, 10):
+ with gzip.GzipFile(fileobj=io.BytesIO(truncated[:i])) as f:
+ self.assertRaises(EOFError, f.read, 1)
+
+
def test_main(verbose=None):
support.run_unittest(TestGzip)