def get_transducer(self):
self.transducer_generated = self.get_generation_validity()
if not self.transducer_generated:
transducer = None
else:
from bracket_rule_transducer import BracketRuleTransducer
rule_transducer_factory = BracketRuleTransducer(self)
transducer = rule_transducer_factory.get_transducer()
return transducer
def test_bracket_markings(self):
self.initialise_segment_table("plural_english_segment_table.txt")
from bracket_rule_transducer import BracketRuleTransducer
from bracket_rule_transducer import LEFT_IDENTITY_BRACKET, RIGHT_BRACKETS, LEFT_BRACKETS, RIGHT_IDENTITY_BRACKET
rule = Rule([{"cons": "+"}], [{"voice": "-"}], [], [{"voice": "-"}], False)
word = 'zt'
prologued_word_transducer = get_prologued_word(word)
rule.extract_data_from_feature_bundle_lists()
bracket_rule_transducer_factory = BracketRuleTransducer(rule)
if rule.obligatory:
obligatory_dfa = bracket_rule_transducer_factory._get_obligatory_dfa([LEFT_IDENTITY_BRACKET],
RIGHT_BRACKETS)
obligatory_transducer = pyfst_from_dfa(obligatory_dfa)
prologued_obligatory = safe_compose(prologued_word_transducer, obligatory_transducer)
custom_obligatory_transducer = pyfst_from_dfa(
bracket_rule_transducer_factory._get_obligatory_dfa([RIGHT_IDENTITY_BRACKET], LEFT_BRACKETS))
prologued_obligatory = safe_compose(prologued_obligatory, custom_obligatory_transducer)
else:
prologued_obligatory = prologued_word_transducer
right_context_transducer = bracket_rule_transducer_factory._get_right_context_dfa()
prologued_obligatory_right = safe_compose(prologued_obligatory, right_context_transducer)
left_context_transducer = bracket_rule_transducer_factory._get_left_context_dfa()
prologued_obligatory_right_left = safe_compose(prologued_obligatory_right, left_context_transducer)
print(get_transducer_outputs(prologued_obligatory_right_left, limit=10))
replace_transducer = bracket_rule_transducer_factory.get_replace_transducer()
prologued_obligatory_right_left_replace = safe_compose(prologued_obligatory_right_left, replace_transducer)
print(get_transducer_outputs(prologued_obligatory_right_left_replace, limit=10))
def debug_generation(self):
self.hall_of_fame.update(self.population)
record = self.stats.compile(self.population) if self.stats else {}
self.logbook.record(gen=self.generation,
nevals=self.new_individuals_in_generation,
**record)
self.logger.info(self.logbook.stream)
best_hypothesis_str = hypothesis_to_string(self.hall_of_fame[0])
record.update({
'generation': self.generation,
'best_hypothesis': best_hypothesis_str
})
self.logger.log_stats_record(record, self.island_number)
if self.generation != self.initial_generation and self.generation % ga_config.DUMP_POPULATION_INTERVAL == 0:
self.dump_population()
if self.generation > 0 and self.generation % ga_config.CLEAR_KEY_VALUE_CACHE_INTERVAL == 0:
self.cache.flush()
if self.generation > 0 and self.generation % ga_config.CLEAR_TRANSDUCERS_CACHE_INTERVAL == 0:
from bracket_rule_transducer import BracketRuleTransducer
self.logger.info("Clearing rule set cache...")
RuleSet.clear_caching()
self.logger.info("Clearing bracket rule transducers cache...")
BracketRuleTransducer.clear_caching()
if ga_config.LOG_POPULATION_INTERVAL > 0 and self.generation % ga_config.LOG_POPULATION_INTERVAL == 0 and self.generation > 0:
self.log_all_population()
if self.generation % ga_config.HALL_OF_FAME_DEBUG_INTERVAL == 0:
self.logger.debug('\n\n**** {} top {} hypothesis:****\n'.format(
self.island_name, ga_config.HALL_OF_FAME_HYPOTHESES))
for i in range(ga_config.HALL_OF_FAME_HYPOTHESES):
try:
hypo = self.hall_of_fame[i]
self.logger.debug('** #{} **'.format(i + 1))
log_hypothesis(hypo, self.logger.debug)
self.logger.debug('\n')
except IndexError:
break
def _clear_modules_caching(self):
from bracket_rule_transducer import BracketRuleTransducer
BracketRuleTransducer.clear_caching()
RuleSet.clear_caching()
def _rule_construction_helper(self, rule, word):
from bracket_rule_transducer import get_prologue_inverse_transducer
from bracket_rule_transducer import BracketRuleTransducer
from bracket_rule_transducer import LEFT_IDENTITY_BRACKET, RIGHT_IDENTITY_BRACKET, RIGHT_APPLICATION_BRACKET, LEFT_APPLICATION_BRACKET
remove_multiple_paths = False
second_obligatory = True
prologued_word_transducer = get_prologued_word(word)
rule.extract_data_from_feature_bundle_lists()
bracket_rule_transducer_factory = BracketRuleTransducer(rule)
if rule.obligatory:
obligatory_dfa = bracket_rule_transducer_factory._get_obligatory_dfa([LEFT_IDENTITY_BRACKET],
[RIGHT_IDENTITY_BRACKET])
obligatory_transducer = pyfst_from_dfa(obligatory_dfa)
prologued_obligatory = safe_compose(prologued_word_transducer, obligatory_transducer)
if second_obligatory:
custom_obligatory_transducer = pyfst_from_dfa(
bracket_rule_transducer_factory._get_obligatory_dfa([RIGHT_IDENTITY_BRACKET],
[LEFT_IDENTITY_BRACKET]))
prologued_obligatory = safe_compose(prologued_obligatory, custom_obligatory_transducer)
else:
prologued_obligatory = prologued_word_transducer
right_context_transducer = bracket_rule_transducer_factory._get_right_context_dfa()
prologued_obligatory_right = safe_compose(prologued_obligatory, right_context_transducer)
replace_transducer = bracket_rule_transducer_factory.get_replace_transducer()
prologued_obligatory_right_replace = safe_compose(prologued_obligatory_right, replace_transducer)
left_context_transducer = bracket_rule_transducer_factory._get_left_context_dfa()
prologued_obligatory_right_replace_left = safe_compose(prologued_obligatory_right_replace,
left_context_transducer)
print(get_transducer_outputs(prologued_obligatory_right_replace_left, limit=10))
# remove_multiple_paths
if remove_multiple_paths:
custom_obligatory_transducer = pyfst_from_dfa(
bracket_rule_transducer_factory._get_custom_obligatory_dfa([RIGHT_APPLICATION_BRACKET],
[LEFT_IDENTITY_BRACKET]))
prologued_obligatory_right_replace_left = safe_compose(prologued_obligatory_right_replace_left,
custom_obligatory_transducer)
custom_obligatory_transducer = pyfst_from_dfa(
bracket_rule_transducer_factory._get_custom_obligatory_dfa([RIGHT_IDENTITY_BRACKET],
[LEFT_APPLICATION_BRACKET]))
prologued_obligatory_right_replace_left = safe_compose(prologued_obligatory_right_replace_left,
custom_obligatory_transducer)
print(get_transducer_outputs(prologued_obligatory_right_replace_left, limit=10))
prologue_inverse_transducer = get_prologue_inverse_transducer()
prologued_obligatory_right_replace_left_inverse = safe_compose(prologued_obligatory_right_replace_left,
prologue_inverse_transducer)
prologued_obligatory_right_replace_left_inverse.remove_epsilon()
transducer_outputs = get_transducer_outputs(prologued_obligatory_right_replace_left_inverse, limit=10)
print(transducer_outputs)
return transducer_outputs