def translate(self,
text: str,
mark_unknown: bool = False,
formatting: Optional[str] = None,
deformat: str = 'txt',
reformat: str = 'txt') -> str:
"""
Args:
text (str)
mark_unknown (bool)
formatting (Optional[str])
deformat (str)
reformat (str)
Returns:
str
"""
if '{}-{}'.format(
*map(to_alpha3_code, [self.l1, self.l2])) in apertium.pairs:
pair = map(to_alpha3_code, [self.l1, self.l2])
else:
raise apertium.ModeNotInstalled()
if pair is not None:
l1, l2 = pair
cmds = list(self._get_commands(l1, l2))
unsafe_deformat, unsafe_reformat = self._get_format(
formatting, deformat, reformat)
deformater, reformater = self._validate_formatters(
unsafe_deformat, unsafe_reformat)
deformatted = self._get_deformat(str(deformater), text)
output = execute(deformatted, cmds)
result = self._get_reformat(str(reformater), output).strip()
return result.decode() # type: ignore
def __init__(self, lang: str) -> None:
"""
Args:
lang (str)
"""
self.analyzer_cmds = {} # type: Dict[str, List[List[str]]]
self.lang = to_alpha3_code(lang) # type: str
if self.lang not in apertium.analyzers:
raise apertium.ModeNotInstalled(self.lang)
else:
self.path, self.mode = apertium.analyzers[self.lang]
def __init__(self, lang: str) -> None:
"""
Args:
lang (str)
"""
self.tagger_cmds: Dict[str, List[List[str]]] = {}
self.lang: str = to_alpha3_code(lang)
if self.lang not in apertium.taggers:
raise apertium.ModeNotInstalled(self.lang)
else:
self.path, self.mode = apertium.taggers[self.lang]
def generate(self, in_text, formatting='none'): # type: (Generator, str, str) -> Union[str, List[str]]
"""
Args:
in_text (str)
formatting (str)
Returns:
Union[str, List[str]]
"""
self.lang = to_alpha3_code(self.lang)
if self.lang in apertium.generators:
commands = list(self._get_commands())
result = execute(in_text, commands)
return result.rstrip('\x00')
else:
raise apertium.ModeNotInstalled(self.lang)
def parse_mode_file(mode_path: str) -> List[List[str]]:
"""
Args:
mode_path (str)
Returns:
List[List[str]]
"""
with open(mode_path) as mode_file:
mode_str = mode_file.read().strip()
if mode_str:
commands = []
for cmd in mode_str.strip().split('|'):
# TODO: we should make language pairs install
# modes.xml instead; this is brittle (what if a path
# has | or ' in it?)
cmd = cmd.replace('$2', '').replace('$1', '-g')
commands.append([c.strip("'") for c in cmd.split()])
return commands
else:
raise apertium.ModeNotInstalled(mode_path)
