def query(self, input_string):
# type: (text_type)->text_type
"""* What you can do
"""
try:
return self.__query(input_string=input_string)
except UnicodeDecodeError:
logger.error(msg=traceback.format_exc())
raise Exception()
def __run_subprocess_mode(self, input_string):
# type: (text_type)->Tuple[bool,text_type]
assert isinstance(self.juman, SubprocessHandler)
assert isinstance(self.knp, SubprocessHandler)
try:
juman_result = self.juman.query(input_string, '^EOS', 'juman')
knp_result = self.knp.query(juman_result, '^EOS', 'knp')
return (True, knp_result)
except UnicodeDecodeError:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False, 'error with UnicodeDecodeError traceback={}'.format(
traceback_message))
except TimeoutError:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False, 'error with TimeoutErro traceback={}'.format(
traceback_message))
except Exception:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False, 'error traceback={}'.format(traceback_message))
def __run_pexpect_mode(self, input_string):
# type: (text_type)->Tuple[bool,text_type]
"""* What you can do
- It calls Juman in UNIX process.
- It calls KNP
- with keep process running
- with launching KNP command everytime
"""
assert isinstance(self.juman, UnixProcessHandler)
try:
juman_result = self.juman.query(input_string=input_string)
except:
return (False, 'error traceback={}'.format(traceback.format_exc()))
if isinstance(self.knp, SubprocessHandler):
try:
parsed_result = self.knp.query(sentence=juman_result.strip(),
eos_pattern='^EOS')
return (True, parsed_result)
except:
traceback_message = traceback.format_exc()
return (False, 'error traceback={}'.format(traceback_message))
else:
# Delete final \n of Juman result document. This \n causes error at KNP #
echo_process = ["echo", juman_result.strip()]
try:
echo_ps = subprocess.Popen(echo_process,
stdout=subprocess.PIPE)
echo_ps.wait()
parsed_result = subprocess.check_output(self.knp,
stdin=echo_ps.stdout)
if six.PY2:
return (True, parsed_result)
else:
return (True, parsed_result.decode('utf-8'))
except subprocess.CalledProcessError:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False,
'error with CalledProcessError. traceback={}'.format(
traceback_message))
except UnicodeDecodeError:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False,
'error with UnicodeDecodeError traceback={}'.format(
traceback_message))
except Exception:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False, 'error traceback={}'.format(traceback_message))
def __run_server_model(self, input_string):
# type: (text_type)->Tuple[bool,text_type]
assert isinstance(self.juman, list)
assert isinstance(self.knp, list)
echo_process = ["echo", input_string]
try:
echo_ps = subprocess.Popen(echo_process, stdout=subprocess.PIPE)
echo_ps.wait()
juman_ps = subprocess.Popen(self.juman,
stdin=echo_ps.stdout,
stdout=subprocess.PIPE)
juman_ps.wait()
parsed_result = subprocess.check_output(self.knp,
stdin=juman_ps.stdout)
if six.PY2:
return (True, parsed_result)
else:
return (True, parsed_result.decode('utf-8'))
except subprocess.CalledProcessError:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False,
'error with CalledProcessError. traceback={}'.format(
traceback_message))
except UnicodeDecodeError:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False, 'error with UnicodeDecodeError traceback={}'.format(
traceback_message))
except Exception:
traceback_message = traceback.format_exc()
logger.error("Error with command={}".format(
traceback.format_exc()))
return (False, 'error traceback={}'.format(traceback_message))
def __init__(self,
record_id,
text,
status,
parsed_result=None,
is_success=None,
sub_id=None,
sentence_index=None,
timestamp=datetime.now(),
updated_at=datetime.now(),
document_args=None):
# type: (int,text_type,bool,Optional[text_type],bool,str,int,datetime,datetime,Dict[str, Any]) -> None
"""
:param record_id: unique id in backend DB
:param text: input text to be parsed.
:param status: boolean flag to describe status of knp parsing.
:param parsed_result: parsing result text.
:param is_success: boolean flag to describe status of knp parsing.
:param sub_id: id in the original given text.
This is used when the original input text is too long and the original text is separated.
:param sentence_index: sentence index when the original input text is separated.
:param timestamp:
:param updated_at:
:param document_args: dict object which is attribute information for input document.
"""
self.record_id = record_id
self.status = status
self.timestamp = timestamp
self.updated_at = updated_at
self.is_success = is_success
self.sentence_index = sentence_index
self.document_args = document_args
if six.PY2:
try:
if isinstance(text, str):
self.text = text.decode('utf-8')
else:
self.text = text
except UnicodeDecodeError:
logger.error(traceback.format_exc())
try:
if isinstance(sub_id, str):
self.sub_id = sub_id.decode('utf-8')
else:
self.sub_id = sub_id
except UnicodeDecodeError:
logger.error(traceback.format_exc())
try:
if isinstance(parsed_result, str):
self.parsed_result = parsed_result.decode('utf-8')
else:
self.parsed_result = parsed_result
except UnicodeDecodeError:
logger.error(traceback.format_exc())
else:
self.text = text
self.sub_id = sub_id
self.parsed_result = parsed_result