def test_missing_filename_python(self):
"""Make sure missing files raise the correct error"""
filename = self.get_temp_filename()
os.remove(filename)
import convertextract
from convertextract.exceptions import MissingFileError
with self.assertRaises(MissingFileError):
convertextract.process(filename)
def test_unsupported_extension_python(self):
"""Make sure unsupported extension raises the correct error"""
filename = self.get_temp_filename(extension="extension")
import convertextract
from convertextract.exceptions import ExtensionNotSupported
with self.assertRaises(ExtensionNotSupported):
convertextract.process(filename)
os.remove(filename)
def test_standardized_text_python(self):
"""Make sure standardized text matches from python"""
import convertextract
result = convertextract.process(self.standardized_text_filename)
if isinstance(result, bytes):
result = result.decode("utf8")
self.assertEqual(
''.join(result.split()),
self.get_standardized_text(),
"standardized text fails for %s" % self.extension,
)
def compare_python_output(self, filename, expected_filename=None, **kwargs):
if expected_filename is None:
expected_filename = self.get_expected_filename(filename, **kwargs)
import convertextract
result = convertextract.process(filename, **kwargs)
if isinstance(result, bytes):
result = result.decode("utf8")
with open(expected_filename, 'r', encoding='utf8') as stream:
result = self.clean_str(result)
expected = self.clean_str(stream.read())
self.assertEqual(result, expected)
def compare_converted_python_output(self,
filename,
expected_filename=None,
**kwargs):
# import pdb; pdb.set_trace()
if expected_filename is None:
expected_filename = self.get_expected_filename(filename, **kwargs)
# print(kwargs['language'])
import convertextract
result = convertextract.process(filename, **kwargs)
if isinstance(result, bytes):
result = result.decode("utf8")
# print(type(result))
# self.maxDiff = None
with open(expected_filename, 'r', encoding="utf8") as stream:
result = self.clean_str(result)
expected = self.clean_str(stream.read())
self.assertEqual(result, expected)
def gooey_main():
"""Interpret the command-line arguments, process the document and
raise errors accordingly (with traceback surpressed).
"""
parser = get_gui()
args = parser.parse_args()
try:
logging.info(f'Extracting text from "{args.filename}"')
if args.mapping:
logging.info(
f'Preparing to convert from mapping file provided at "{args.mapping}"')
else:
logging.info(
f'Preparing to convert from "{args.input_language}" to "{args.output_language}"')
output = process(**vars(args))
except CommandLineError as ex:
logging.error(
"Whoops. Something went wrong. Please check your input file and your mapping file (if using a custom one)")
sys.stderr.write(red(ex) + '\n')
sys.exit(1)
else:
fn, ext = os.path.splitext(args.filename)
output_path = fn + '_converted' + ext
logging.info(f'Converted file available at "{output_path}"')