def guessColspec(self, otpl_text, header):
self.otpl_file.write(otpl_text)
self.otpl_file.close()
expected = ColumnSpecification.from_string(header)
result = guess_colspec(self.segments)
self.assertSequenceEqual(str(expected), str(result))
self.assertEqual(expected, result)
def otpl_to_brat(configuration):
"""
For a list of `text_files` (paths), read the associated OTPL files and
write the converted brat files.
:type configuration: Configuration
:return: the error count (number of failed conversions)
"""
converter = OtplBratConverter()
converter.set_colspec(configuration.colspec)
errors = 0
if configuration.name_labels is not None:
converter.set_name_dict(configuration.name_labels)
for text_file in configuration.input_files:
otpl_file = make_path_to(text_file, configuration.otpl_suffix)
brat_file = make_path_to(text_file, configuration.brat_suffix)
if exists(otpl_file):
segments = configure_reader(otpl_file, configuration)
if segments is None:
errors += 1
continue
if configuration.colspec is None:
configuration.colspec = guess_colspec(segments)
converter.set_colspec(configuration.colspec)
if not converter.convert(segments, text_file, brat_file):
L.error('conversion for "%s" failed', text_file)
errors += 1
else:
L.error('could not locate OTPL file "%s" for "%s"',
otpl_file, text_file)
errors += 1
if not errors:
brat_config_file = join(dirname(configuration.input_files[-1]),
configuration.config)
if not exists(brat_config_file):
converter.write_config_file(brat_config_file)
if errors:
L.debug('conversion of %s file%s failed',
errors, '' if errors == 1 else 's')
return errors
