def testDefault(self):
logging.getLogger().addHandler(logging.StreamHandler()) # might spam your console...
self.interceptLogs('otplc.extractor')
self.otpl_file.write(
"This DT 6 nsubj B-NP NULL\n"
"is VBZ 6 cop B-VP NULL\n"
"Florian NNP 6 nn B-NP NULL\n"
"ʼs POS 3 pos I-NP db:id\n"
"weird JJ 6 amod I-NP NULL\n"
"test NN 0 root I-NP NULL\n"
". DOT 6 punct O NULL\n\n"
"And DT 6 nsubj B-NP NULL\n"
"another VBZ 6 cop B-VP NULL\n"
"one NN 0 root I-NP NULL\n"
". DOT 6 punct O NULL\n\n"
)
self.otpl_file.close()
expected = 'This is Florian ʼs weird test .\nAnd another one .\n'
self.assertEqual(0, otpl_to_text(Configuration([self.otpl_file.name])))
result = open(make_path_to(self.otpl_file.name, Configuration.TEXT_SUFFIX)).read()
self.assertEqual(expected, result)
def otpl_to_text(configuration):
"""
Extract the text using the tokens of the OTPL files and store the results
into separate plain-text files.
:param configuration: a :class:`otplc.settings.Configuration` object
:return: The number of failed conversion for the input files.
"""
errors = 0
for otpl_file in configuration.input_files:
text_file = make_path_to(otpl_file, configuration.text_suffix)
msg = "output text file and input OTPL file have the same path " \
"(ensure the OTPL file does not use the extension '{}')"
assert otpl_file != text_file, msg.format(configuration.text_suffix)
segments = configure_reader(otpl_file, configuration)
if segments is None:
errors += 1
continue
if configuration.colspec is None:
configuration.colspec = guess_colspec(segments)
token = configuration.colspec.token
try:
with open(text_file,
encoding=configuration.encoding,
mode='wt') as out_stream:
for seg in segments:
print(*[row[token] for row in seg], file=out_stream)
except IOError as e:
L.error('I/O error while extracting %s to %s: %s',
otpl_file, text_file, str(e))
errors += 1
return errors
