def get_ucca_parser():
ucca_dir = RESOURCES_DIR / 'ucca'
os.chdir(str(ucca_dir))
model_path = ucca_dir / 'models/ucca-bilstm'
vocab_path = ucca_dir / 'vocab'
argv = ['script_name', '-m', str(model_path), '--vocab', str(vocab_path)]
with unittest.mock.patch('sys.argv', argv):
Config.reload()
args = Config().args
model_files = [
base + '' + ext
for base, ext in map(os.path.splitext, args.models or (
args.classifier, ))
]
return Parser(model_files=model_files, config=Config(), beam=1)
def get_ucca_passage(sentence):
source_path = Path(tempfile.mkdtemp()) / 'source.txt'
with source_path.open('w') as f:
f.write(sentence + '\n')
argv = ['script_name', str(source_path)]
with unittest.mock.patch('sys.argv', argv):
Config.reload()
args = Config().args
train_passages, dev_passages, test_passages = [
read_passages(args, arg)
for arg in (args.train, args.dev, args.passages)
]
ucca_passages = [
ucca_passage for (ucca_passage, ) in PARSER.parse(
test_passages, evaluate=[], display=False, write=False)
]
assert len(ucca_passages) == 1
return ucca_passages[0]
