class Bot(object):
def __init__(self, config, client):
assert (config is not None)
assert (client is not None)
self._configuration = config
self._client = client
self._brain_factory = BrainFactory(self)
self._question_depth = 0
self._question_start_time = None
self._spell_checker = None
self.initiate_spellchecker()
self._sentence_splitter = None
self.initiate_sentence_splitter()
self._sentence_joiner = None
self.initiate_sentence_joiner()
self._from_translator = None
self._to_translator = None
self.initiate_translator()
self._sentiment_analyser = None
self._sentiment_scores = None
self.initiate_sentiment_analyser()
self._conversation_mgr = ConversationManager(config.conversations)
self._conversation_mgr.initialise(self._client.storage_factory)
def ylogger_type(self):
return "bot"
@property
def id(self):
return self._configuration.section_name
@property
def client(self):
return self._client
@property
def configuration(self):
return self._configuration
@property
def brain_factory(self):
return self._brain_factory
@property
def spell_checker(self):
return self._spell_checker
def initiate_spellchecker(self):
if self.configuration is not None:
if self.configuration.spelling is not None:
self._spell_checker = SpellingChecker.initiate_spellchecker(
self.configuration.spelling, self.client.storage_factory)
@property
def sentence_splitter(self):
return self._sentence_splitter
def initiate_sentence_splitter(self):
if self.configuration is not None:
if self.configuration.splitter is not None:
self._sentence_splitter = SentenceSplitter.initiate_sentence_splitter(
self.configuration.splitter)
@property
def sentence_joiner(self):
return self._sentence_joiner
def initiate_sentence_joiner(self):
if self.configuration is not None:
if self.configuration.joiner is not None:
self._sentence_joiner = SentenceJoiner.initiate_sentence_joiner(
self.configuration.joiner)
@property
def from_translator(self):
return self._from_translator
@property
def to_translator(self):
return self._to_translator
def initiate_translator(self):
if self.configuration is not None:
if self.configuration.from_translator is not None:
self._from_translator = BaseTranslator.initiate_translator(
self.configuration.from_translator)
if self.configuration.to_translator is not None:
self._to_translator = BaseTranslator.initiate_translator(
self.configuration.to_translator)
@property
def sentiment_analyser(self):
return self._sentiment_analyser
@property
def sentiment_scores(self):
return self._sentiment_scores
def initiate_sentiment_analyser(self):
if self.configuration is not None:
if self.configuration.sentiment_analyser is not None:
self._sentiment_analyser, self._sentiment_scores = BaseSentimentAnalyser.initiate_sentiment_analyser(
self.configuration.sentiment_analyser)
@property
def brain(self):
return self._brain_factory.select_brain()
@property
def conversations(self):
return self._conversation_mgr
@property
def default_response(self):
if self.configuration is not None:
return self.configuration.default_response
return BotConfiguration.DEFAULT_RESPONSE
@property
def default_response_srai(self):
if self.configuration is not None:
return self.configuration.default_response_srai
return None
@property
def exit_response(self):
if self.configuration is not None:
return self.configuration.exit_response
return BotConfiguration.DEFAULT_EXIT_RESPONSE
@property
def exit_response_srai(self):
if self.configuration is not None:
return self.configuration.exit_response_srai
return BotConfiguration.DEFAULT_EXIT_RESPONSE_SRAI
@property
def initial_question(self):
if self.configuration is not None:
return self.configuration.initial_question
return BotConfiguration.DEFAULT_INITIAL_QUESTION
@property
def initial_question_srai(self):
if self.configuration is not None:
return self.configuration.initial_question_srai
return BotConfiguration.DEFAULT_INITIAL_QUESTION_SRAI
@property
def override_properties(self):
if self.configuration is not None:
return self.configuration.override_properties
return False
def get_version_string(self, client_context):
assert (client_context is not None)
if client_context.brain.properties.has_property("version"):
# The old version of returning the version string, did not distinquish
# between App and Grammar version
return "%s, v%s, initiated %s" % (
client_context.brain.properties.property("name"),
client_context.brain.properties.property("version"),
client_context.brain.properties.property("birthdate"))
else:
# This version now does
return "%s, App: v%s Grammar v%s, initiated %s" % (
client_context.brain.properties.property("name"),
client_context.brain.properties.property("app_version"),
client_context.brain.properties.property("grammar_version"),
client_context.brain.properties.property("birthdate"))
def has_conversation(self, client_context):
assert (self._conversation_mgr is not None)
return self._conversation_mgr.has_conversation(client_context)
def get_conversation(self, client_context):
assert (self._conversation_mgr is not None)
return self._conversation_mgr.get_conversation(client_context)
def save_conversation(self, client_context):
assert (self._conversation_mgr is not None)
self._conversation_mgr.save_conversation(client_context)
def check_spelling_before(self, client_context, each_sentence):
if self.spell_checker is not None:
self.spell_checker.check_spelling_before(client_context,
each_sentence)
def check_spelling_and_retry(self, client_context, each_sentence):
if self.spell_checker is not None:
return self.spell_checker.check_spelling_and_retry(
client_context, each_sentence)
return None
def get_default_response(self, client_context):
assert (client_context is not None)
if self.default_response_srai is not None:
sentence = Sentence(client_context.brain.tokenizer,
self.default_response_srai)
default_response = client_context.brain.ask_question(
client_context, sentence)
if default_response is None or not default_response:
default_response = self.default_response
return default_response
else:
return self.default_response
def get_initial_question(self, client_context):
assert (client_context is not None)
if self.initial_question_srai is not None:
sentence = Sentence(client_context.brain.tokenizer,
self.initial_question_srai)
initial_question = client_context.brain.ask_question(
client_context, sentence)
if initial_question is None or not initial_question:
initial_question = self.initial_question
return initial_question
else:
return self.initial_question
def get_exit_response(self, client_context):
assert (client_context is not None)
if self.exit_response_srai is not None:
sentence = Sentence(client_context.brain.tokenizer,
self.exit_response_srai)
exit_response = client_context.brain.ask_question(
client_context, sentence)
if exit_response is None or not exit_response:
exit_response = self.exit_response
return exit_response
else:
return self.exit_response
def pre_process_text(self, client_context, text, srai):
assert (client_context is not None)
assert (client_context.brain is not None)
if srai is False:
pre_processed = client_context.brain.pre_process_question(
client_context, text)
YLogger.debug(client_context, "Pre Processed (%s): %s",
client_context.userid, pre_processed)
else:
pre_processed = text
if pre_processed is None or pre_processed == "":
assert (self.configuration is not None)
pre_processed = self.configuration.empty_string
return pre_processed
def get_question(self, client_context, pre_processed, srai):
if srai is False:
return Question.create_from_text(client_context,
pre_processed,
srai=srai)
else:
return Question.create_from_text(client_context,
pre_processed,
split=False,
srai=srai)
def combine_answers(self, answers, srai):
assert (answers is not None)
assert (self._sentence_joiner is not None)
return self._sentence_joiner.combine_answers(answers, srai)
def post_process_response(self, client_context, response, srai):
if srai is False:
assert (client_context is not None)
answer = client_context.brain.post_process_response(
client_context, response).strip()
if not answer:
answer = self.get_default_response(client_context)
else:
answer = response
return answer
def log_answer(self, client_context, text, answer, responselogger):
YLogger.debug(client_context, "Processed Response (%s): %s",
client_context.userid, answer)
if responselogger is not None:
responselogger.log_response(text, answer)
def ask_question(self,
client_context,
text,
srai=False,
responselogger=None):
assert (client_context is not None)
if srai is False:
client_context.bot = self
client_context.brain = client_context.bot.brain
assert (client_context.bot is not None)
assert (client_context.brain is not None)
client_context.mark_question_start(text)
pre_processed = self.pre_process_text(client_context, text, srai)
question = self.get_question(client_context, pre_processed, srai)
conversation = self.get_conversation(client_context)
if len(conversation.questions) == 0:
if self.client.trigger_manager is not None:
self.client.trigger_manager.trigger(
SystemTriggers.CONVERSATION_START, client_context)
assert (conversation is not None)
conversation.record_dialog(question)
answers = self.process_sentences(client_context, question, srai,
responselogger)
client_context.reset_question()
if srai is True:
conversation.pop_dialog()
self.save_conversation(client_context)
conversation.save_sentiment()
if self.client.trigger_manager is not None and srai is False:
self.client.trigger_manager.trigger(SystemTriggers.QUESTION_ASKED,
client_context)
return self.combine_answers(answers, srai)
def process_sentences(self, client_context, question, srai,
responselogger):
answers = []
sentence_no = 0
for sentence in question.sentences:
question.set_current_sentence_no(sentence_no)
answer = self.process_sentence(client_context, sentence, srai,
responselogger)
answers.append(answer)
sentence_no += 1
return answers
def process_sentence(self, client_context, sentence, srai, responselogger):
assert (client_context is not None)
assert (client_context.brain is not None)
client_context.check_max_recursion()
client_context.check_max_timeout()
if srai is False:
self.check_spelling_before(client_context, sentence)
response = client_context.brain.ask_question(client_context, sentence,
srai)
if response is None and srai is False:
response = self.check_spelling_and_retry(client_context, sentence)
if response is not None:
return self.handle_response(client_context, sentence, response,
srai, responselogger)
else:
return self.handle_none_response(client_context, sentence,
responselogger)
def handle_response(self, client_context, sentence, response, srai,
responselogger):
assert (sentence is not None)
YLogger.debug(client_context, "Raw Response (%s): %s",
client_context.userid, response)
sentence.response = response
sentence.calculate_sentinment_score(client_context)
answer = self.post_process_response(client_context, response, srai)
self.log_answer(client_context, sentence.text, answer, responselogger)
return answer
def handle_none_response(self, client_context, sentence, responselogger):
assert (sentence is not None)
sentence.response = self.get_default_response(client_context)
sentence.calculate_sentinment_score(client_context)
if responselogger is not None:
responselogger.log_unknown_response(sentence)
return sentence.response
class Bot(object):
def __init__(self, config, client):
assert (config is not None)
assert (client is not None)
self._configuration = config
self._client = client
self._brain_factory = BrainFactory(self)
self._question_depth = 0
self._question_start_time = None
self._spell_checker = None
self.initiate_spellchecker()
self._sentence_splitter = None
self.initiate_sentence_splitter()
self._sentence_joiner = None
self.initiate_sentence_joiner()
self._from_translator = None
self._to_translator = None
self.initiate_translator()
self._sentiment_analyser = None
self._sentiment_scores = None
self.initiate_sentiment_analyser()
self._conversation_mgr = ConversationManager(config.conversations)
self._conversation_mgr.initialise(self._client.storage_factory)
self._utterance = None
self._is_error = False
self._srai_count = 0
def ylogger_type(self):
return "bot"
@property
def id(self):
return self._configuration.section_name
@property
def client(self):
return self._client
@property
def configuration(self):
return self._configuration
@property
def brain_factory(self):
return self._brain_factory
@property
def spell_checker(self):
return self._spell_checker
def initiate_spellchecker(self):
if self.configuration.spelling is not None:
self._spell_checker = SpellingChecker.initiate_spellchecker(
self.configuration.spelling, self.client.storage_factory)
@property
def sentence_splitter(self):
return self._sentence_splitter
def initiate_sentence_splitter(self):
if self.configuration.splitter is not None:
self._sentence_splitter = SentenceSplitter.initiate_sentence_splitter(
self.configuration.splitter)
if self._sentence_splitter is not None:
if self.brain.properties.has_property(
"splitter_split_chars") is True:
self._sentence_splitter.set_configuration_split_chars(
self.brain.properties.property("splitter_split_chars"))
if self.brain.properties.has_property(
"punctuation_chars") is True:
self._sentence_splitter.set_configuration_punctuation_chars(
self.brain.properties.property("punctuation_chars"))
@property
def sentence_joiner(self):
return self._sentence_joiner
def initiate_sentence_joiner(self):
if self.configuration.joiner is not None:
self._sentence_joiner = SentenceJoiner.initiate_sentence_joiner(
self.configuration.joiner)
if self._sentence_joiner is not None:
if self.brain.properties.has_property(
"joiner_join_chars") is True:
self._sentence_joiner.set_configuration_join_chars(
self.brain.properties.property("joiner_join_chars"))
if self.brain.properties.has_property(
"joiner_terminator") is True:
self._sentence_joiner.set_configuration_terminator(
self.brain.properties.property("joiner_terminator"))
@property
def from_translator(self):
return self._from_translator
@property
def to_translator(self):
return self._to_translator
def initiate_translator(self):
if self.configuration.from_translator is not None:
self._from_translator = BaseTranslator.initiate_translator(
self.configuration.from_translator)
if self.configuration.to_translator is not None:
self._to_translator = BaseTranslator.initiate_translator(
self.configuration.to_translator)
@property
def sentiment_analyser(self):
return self._sentiment_analyser
@property
def sentiment_scores(self):
return self._sentiment_scores
def initiate_sentiment_analyser(self):
if self.configuration.sentiment_analyser is not None:
self._sentiment_analyser, self._sentiment_scores = BaseSentimentAnalyser.initiate_sentiment_analyser(
self.configuration.sentiment_analyser)
@property
def brain(self):
return self._brain_factory.select_brain()
@property
def conversations(self):
return self._conversation_mgr
@property
def default_response(self):
return self.configuration.default_response
@property
def default_response_srai(self):
return self.configuration.default_response_srai
@property
def exit_response(self):
return self.configuration.exit_response
@property
def exit_response_srai(self):
return self.configuration.exit_response_srai
@property
def exception_response(self):
return self.configuration.exception_response
@property
def initial_question(self):
return self.configuration.initial_question
@property
def initial_question_srai(self):
return self.configuration.initial_question_srai
@property
def override_properties(self):
return self.configuration.override_properties
@property
def utterance(self):
return self._utterance
def get_version_string(self, client_context):
assert (client_context is not None)
if client_context.brain.properties.has_property("version"):
# The old version of returning the version string, did not distinquish
# between App and Grammar version
return "%s, v%s, initiated %s " % (
client_context.brain.properties.property("name"),
client_context.brain.properties.property("version"),
client_context.brain.properties.property("birthdate"))
else:
# This version now does
return "%s, App: v%s Grammar v%s, initiated %s " % (
client_context.brain.properties.property("name"),
client_context.brain.properties.property("app_version"),
client_context.brain.properties.property("grammar_version"),
client_context.brain.properties.property("birthdate"))
def has_conversation(self, client_context):
assert (self._conversation_mgr is not None)
return self._conversation_mgr.has_conversation(client_context)
def conversation(self, client_context):
return self.get_conversation(client_context)
def get_conversation(self, client_context):
assert (self._conversation_mgr is not None)
return self._conversation_mgr.get_conversation(client_context)
def save_conversation(self, client_context):
assert (self._conversation_mgr is not None)
self._conversation_mgr.save_conversation(client_context)
def check_spelling_before(self, client_context, each_sentence):
if self.spell_checker is not None:
self.spell_checker.check_spelling_before(client_context,
each_sentence)
def check_spelling_and_retry(self, client_context, each_sentence):
if self.spell_checker is not None:
return self.spell_checker.check_spelling_and_retry(
client_context, each_sentence)
return None
def get_default_response(self, client_context):
assert (client_context is not None)
if self.default_response_srai is not None and self.default_response_srai != '':
sentence = Sentence(client_context.brain.tokenizer,
self.default_response_srai)
default_response = client_context.brain.ask_question(
client_context, sentence, default_srai=True)
if default_response is None:
default_response = client_context.brain.properties.property(
"default-response")
if default_response is None:
default_response = self.default_response
return default_response
else:
default_response = client_context.brain.properties.property(
"default-response")
if default_response is None:
default_response = self.default_response
return default_response
def get_exception_response(self, client_context):
assert (client_context is not None)
exception_response = client_context.brain.properties.property(
"exception-response")
if exception_response is None:
exception_response = self.exception_response
return exception_response
def get_initial_question(self, client_context):
assert (client_context is not None)
if self.initial_question_srai is not None:
sentence = Sentence(client_context.brain.tokenizer,
self.initial_question_srai)
initial_question = client_context.brain.ask_question(
client_context, sentence)
if initial_question is None or not initial_question:
initial_question = self.initial_question
return initial_question
else:
return self.initial_question
def get_exit_response(self, client_context):
assert (client_context is not None)
if self.exit_response_srai is not None:
sentence = Sentence(client_context.brain.tokenizer,
self.exit_response_srai)
exit_response = client_context.brain.ask_question(
client_context, sentence)
if exit_response is None or not exit_response:
exit_response = self.exit_response
return exit_response
else:
return self.exit_response
def pre_process_text(self, client_context, text, srai):
assert (client_context is not None)
assert (client_context.brain is not None)
if srai is False:
pre_processed = client_context.brain.pre_process_question(
client_context, text)
YLogger.debug(client_context, "Pre Processed (%s): %s",
client_context.userid, pre_processed)
else:
pre_processed = text
if pre_processed is None or pre_processed == "":
assert (self.configuration is not None)
pre_processed = self.configuration.empty_string
return pre_processed
def get_question(self, client_context, pre_processed, srai):
if srai is False:
return Question.create_from_text(client_context,
pre_processed,
srai=srai)
else:
return Question.create_from_text(client_context,
pre_processed,
split=False,
srai=srai)
def combine_answers(self, answers, srai):
assert (answers is not None)
assert (self._sentence_joiner is not None)
return self._sentence_joiner.combine_answers(answers, srai)
def post_process_response(self, client_context, response, srai):
if srai is False:
assert (client_context is not None)
answer = client_context.brain.post_process_response(
client_context, response).strip()
if not answer:
answer = self.get_default_response(client_context)
else:
answer = response
return answer
def log_answer(self, client_context, text, answer, responselogger, srai):
YLogger.debug(client_context, "Processed Response (%s) [srai:%s] : %s",
client_context.userid, srai, answer)
if responselogger is not None:
responselogger.log_response(text, answer)
def ask_question(self,
client_context,
text,
srai=False,
responselogger=None):
assert (client_context is not None)
if srai is False:
client_context.bot = self
client_context.brain = client_context.bot.brain
self._is_error = False
else:
if self._is_error is True:
return ''
assert (client_context.bot is not None)
assert (client_context.brain is not None)
client_context.mark_question_start(text, srai)
pre_processed = self.pre_process_text(client_context, text, srai)
question = self.get_question(client_context, pre_processed, srai)
conversation = self.get_conversation(client_context)
if len(conversation.questions) == 0:
if self.client.trigger_manager is not None:
self.client.trigger_manager.trigger(
SystemTriggers.CONVERSATION_START, client_context)
assert (conversation is not None)
if srai is False:
conversation.exception = None
if client_context.server_mode is True:
self.conversations.load_learnf_with_userid(client_context)
client_context.brain.rdf.apply_updates()
if client_context.deleteVariable is True:
conversation.clear_data_property()
self.conversations.remove_internal_data(client_context)
conversation.record_dialog(question)
try:
answers = self.process_sentences(client_context, question, srai,
responselogger)
except LimitOverException as excep:
if self._is_error is False:
self._is_error = True
conversation.exception = str(excep)
YLogger.exception(client_context, "Limit-Over Exception",
excep)
base = conversation.internal_base
conversation.add_internal_data(base, 'exception',
conversation.exception)
conversation.add_internal_variables(base, 'after_variables')
if srai is True:
conversation.pop_dialog()
raise
except Exception as excep:
if self._is_error is False:
self._is_error = True
conversation.exception = str(excep)
YLogger.exception(client_context, "Exception", excep)
base = conversation.internal_base
conversation.add_internal_data(base, 'exception',
conversation.exception)
conversation.add_internal_variables(base, 'after_variables')
if srai is True:
conversation.pop_dialog()
raise
if self._is_error is True:
question.exception = conversation.exception
answer = self.get_exception_response(client_context)
sentence = question.current_sentence()
sentence.response = answer
answers = [answer]
client_context.reset_question()
if srai is True:
conversation.pop_dialog()
conversation.save_sentiment()
if srai is False:
self.conversations.set_internal_data(client_context)
self.save_conversation(client_context)
if client_context.server_mode is True:
self.conversations.remove_conversation(client_context)
self.conversations.remove_learnf_with_userid(client_context)
if self.client.trigger_manager is not None and srai is False:
self.client.trigger_manager.trigger(SystemTriggers.QUESTION_ASKED,
client_context)
return self.combine_answers(answers, srai)
def process_sentences(self, client_context, question, srai,
responselogger):
self._utterances = None
utterances = []
answers = []
sentence_no = 0
for sentence in question.sentences:
YLogger.debug(client_context, "Sentence-Words %s", sentence.words)
question.set_current_sentence_no(sentence_no)
utterance = client_context.brain.tokenizer.words_to_texts(
sentence.words)
if utterance == '':
answer = ''
else:
answer = self.process_sentence(client_context, sentence, srai,
responselogger)
utterances.append(utterance)
answers.append(answer)
if self._is_error is True:
break
sentence_no += 1
self._utterance = self.combine_answers(utterances, False)
return answers
def process_sentence(self, client_context, sentence, srai, responselogger):
assert (client_context is not None)
assert (client_context.brain is not None)
try:
client_context.check_max_recursion()
client_context.check_max_timeout()
except LimitOverException:
raise
if srai is False:
self.check_spelling_before(client_context, sentence)
self._srai_count = 0
else:
self._srai_count += 1
if self._srai_count > client_context.bot.configuration.max_search_srai:
raise LimitOverException(
"Max search srai [%d] exceeded: [%s]" %
(client_context.bot.configuration.max_search_srai,
client_context.brain.tokenizer.words_to_texts(
sentence.words)))
response = client_context.brain.ask_question(client_context, sentence,
srai)
if response is None and srai is False:
response = self.check_spelling_and_retry(client_context, sentence)
if response is not None:
return self.handle_response(client_context, sentence, response,
srai, responselogger)
else:
return self.handle_none_response(client_context, sentence,
responselogger)
def handle_response(self, client_context, sentence, response, srai,
responselogger):
assert (sentence is not None)
YLogger.debug(client_context, "Raw Response (%s): %s",
client_context.userid, response)
sentence.response = response
sentence.calculate_sentinment_score(client_context)
answer = self.post_process_response(client_context, response, srai)
self.log_answer(client_context, sentence.text(), answer,
responselogger, srai)
return answer
def handle_none_response(self, client_context, sentence, responselogger):
assert (sentence is not None)
sentence.response = self.get_default_response(client_context)
sentence.calculate_sentinment_score(client_context)
if responselogger is not None:
responselogger.log_unknown_response(sentence)
return sentence.response