# Reproductions/tests for crashes/read errors in TiffDecode.c # When run in Python, all of these images should fail for # one reason or another, either as a buffer overrun, # unrecognized datastream, or truncated image file. # There shouldn't be any segfaults. # # if run like # `valgrind --tool=memcheck pytest test_tiff_crashes.py 2>&1 | grep TiffDecode.c` # the output should be empty. There may be Python issues # in the valgrind especially if run in a debug Python # version. import pytest from PIL import Image from .helper import on_ci @pytest.mark.parametrize( "test_file", [ "Tests/images/crash_1.tif", "Tests/images/crash_2.tif", "Tests/images/crash-2020-10-test.tif", "Tests/images/crash-1152ec2d1a1a71395b6f2ce6721c38924d025bf3.tif", "Tests/images/crash-0e16d3bfb83be87356d026d66919deaefca44dac.tif", ], ) @pytest.mark.filterwarnings("ignore:Possibly corrupt EXIF data") @pytest.mark.filterwarnings("ignore:Metadata warning") def test_tiff_crashes(test_file): try: with Image.open(test_file) as im: im.load() except FileNotFoundError: if not on_ci(): pytest.skip("test image not found") return raise except OSError: pass