def test_corrupted_ocr(self):
fs = FilesystemSource(os.path.join(test_data_path, "corrupted"))
with SourceManager() as sm:
for h in fs.handles(sm):
resource = h.follow(sm)
self.assertEqual(convert(resource, OutputType.Text), None,
"{0}: error handling failed".format(h))
def test_ocr_conversions(self):
fs = FilesystemSource(os.path.join(test_data_path, "good"))
with SourceManager() as sm:
for h in fs.handles(sm):
resource = h.follow(sm)
self.assertEqual(
convert(resource, OutputType.Text).value, expected_result,
"{0}: content failed".format(h))
def test_size_computation(self):
fs = FilesystemSource(test_data_path)
with SourceManager() as sm:
for h in fs.handles(sm):
resource = h.follow(sm)
size = convert(resource, OutputType.ImageDimensions)
if not size:
if "rgba32" in h.relative_path:
self.skipTest("Pillow RGBA bug detected -- skipping")
else:
size = size.value
self.assertEqual(size, expected_size, "{0}: size failed")
def test_eml_files(self):
fs = FilesystemSource(test_data_path)
with SourceManager() as sm:
for h in fs.handles(sm):
mail_source = Source.from_handle(h)
self.assertIsInstance(
mail_source,
MailSource,
"conversion of {0} to MailSource failed".format(h))
for h in mail_source.handles(sm):
self.assertIsInstance(
h,
MailPartHandle)