def test_builtinmapper_invalid(self):
"""Test that an invalid builtin scheme name raises a
BuiltinCharMapNotFound exception.
"""
with pytest.raises(BuiltinCharMapNotFoundError):
CharMapper.builtin_mapper('hello')
def main(): # pragma: no cover
try:
version = ('CAMeL Tools v{}'.format(__version__))
arguments = docopt(__doc__, version=version)
# Open files (or just use stdin and stdout)
fin, fout = _open_files(arguments['FILE'], arguments['--output'])
try:
mapper = CharMapper.builtin_mapper('arclean')
_arclean(mapper, fin, fout)
# If everything worked so far, this shouldn't happen
except Exception:
sys.stderr.write('Error: An error occured during cleaning.\n')
fin.close()
fout.close()
sys.exit(1)
# Cleanup
if arguments['FILE'] is not None:
fin.close()
if arguments['--output'] is not None:
fout.close()
sys.exit(0)
except KeyboardInterrupt:
sys.stderr.write('Exiting...\n')
sys.exit(1)
except Exception:
sys.stderr.write('Error: An unknown error occurred.\n')
sys.exit(1)
def test_builtinmapper_arclean(self):
"""Test that the builtin 'arclean' scheme is loaded without errors.
"""
assert CharMapper.builtin_mapper('arclean')
def test_builtinmapper_xmlbw2hsb(self):
"""Test that the builtin 'xmlbw2hsb' scheme is loaded without
errors.
"""
assert CharMapper.builtin_mapper('xmlbw2hsb')
def test_builtinmapper_hsb2xmlbw(self):
"""Test that the builtin 'hsb2xmlbw' scheme is loaded without errors.
"""
assert CharMapper.builtin_mapper('hsb2xmlbw')
def test_builtinmapper_safebw2bw(self):
"""Test that the builtin 'safebw2bw' scheme is loaded without errors.
"""
assert CharMapper.builtin_mapper('safebw2bw')
def test_builtinmapper_bw2ar(self):
"""Test that the builtin 'bw2ar' scheme is loaded without errors.
"""
assert CharMapper.builtin_mapper('bw2ar')
def main(): # pragma: no cover
try:
version = ('CAMeL Tools v{}'.format(__version__))
arguments = docopt(__doc__, version=version)
if arguments['--list']:
for scheme in _BUILTIN_SCHEMES:
print("{} {}".format(scheme[0].ljust(20), scheme[1]))
sys.exit(0)
if arguments['--scheme'] is not None:
if arguments['--scheme'] not in [s[0] for s in _BUILTIN_SCHEMES]:
sys.stderr.write('Error: {} is not a valid scheme.\n'
'Run `camel_transliterate -l` to see the list'
' of available schemes.'
'\n'.format(repr(arguments['--scheme'])))
sys.exit(1)
if arguments['--marker'] is None:
marker = '@@IGNORE@@'
else:
marker = arguments['--marker']
ignore_markers = arguments['--ignore-markers']
strip_markers = arguments['--strip-markers']
# Open files (or just use stdin and stdout)
fin, fout = _open_files(arguments['FILE'], arguments['--output'])
# Load the CharMapper and initialize a Transliterator with it
try:
mapper = CharMapper.builtin_mapper(arguments['--scheme'])
trans = Transliterator(mapper, marker)
except Exception: # pylint: disable=W0703
sys.stderr.write('Error: Could not load builtin scheme'
' {}.\n'.format(repr(arguments['--scheme'])))
sys.exit(1)
# Transliterate lines
try:
for line in fin:
line = force_unicode(line)
if six.PY3:
fout.write(
trans.transliterate(line, strip_markers,
ignore_markers))
else:
fout.write(
force_encoding(
trans.transliterate(line, strip_markers,
ignore_markers)))
fout.flush()
# If everything worked so far, this shouldn't happen
except Exception: # pylint: disable=W0703
sys.stderr.write('Error: An unkown error occured during '
'transliteration.\n')
sys.exit(1)
# Cleanup
if arguments['FILE'] is not None:
fin.close()
if arguments['--output'] is not None:
fout.close()
sys.exit(0)
except KeyboardInterrupt:
sys.stderr.write('Exiting...\n')
sys.exit(1)
except Exception:
sys.stderr.write('Error: An unknown error occurred.\n')
sys.exit(1)
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
import subprocess
import sys
import os
import argparse
import time
from helpers.preprocess import preprocess
from helpers.tag import tag
from ai.tests.mle import train_mle, predict_mle
from camel_tools.utils.charmap import CharMapper
ar2bw = CharMapper.builtin_mapper('ar2bw')
def is_bool(s):
return str(s) != 'False'
parser = argparse.ArgumentParser(
description=
'This program rewrites (transliterates) from one language script to another'
)
# --model_name can take values "mle", "word2word", "line2line", or "hybrid"
parser.add_argument('--model_name',
action="store",
dest='model_name',
def __init__(self) -> None:
super().__init__()
self.sec_cleaner = None
self.clean_mapper = CharMapper.builtin_mapper('arclean')
self.ar2bw_mapper = CharMapper.builtin_mapper('ar2bw')
self.bw2ar_mapper = CharMapper.builtin_mapper('bw2ar')