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 get_left_to_right_application(self):
prologue_transducer = get_prologue_transducer()
if self.obligatory or self._should_fix_transducer():
obligatory_transducer = pyfst_from_dfa(
self._get_obligatory_dfa([LEFT_IDENTITY_BRACKET],
[RIGHT_IDENTITY_BRACKET]))
composed_transducer = safe_compose(prologue_transducer,
obligatory_transducer)
else:
composed_transducer = prologue_transducer
right_context_transducer = self._get_right_context_dfa()
replace_transducer = self.get_replace_transducer()
replace_transducer = uniform_encoding.get_weighted_replace_transducer(
replace_transducer, self)
left_context_transducer = self._get_left_context_dfa()
prologue_inverse_transducer = get_prologue_inverse_transducer()
composed_transducer = chain_safe_compose(composed_transducer,
right_context_transducer,
replace_transducer,
left_context_transducer)
if self.transformation_type == INSERTION:
insertion_obligatory_transducer = pyfst_from_dfa(
self._get_obligatory_dfa([RIGHT_IDENTITY_BRACKET],
[LEFT_IDENTITY_BRACKET]))
composed_transducer = safe_compose(
composed_transducer, insertion_obligatory_transducer)
# remove multiple paths
if self.transformation_type == ASSIMILATION or self.transformation_type == INSERTION:
RI_obligatory_transducer = pyfst_from_dfa(
self._get_custom_obligatory_dfa([RIGHT_APPLICATION_BRACKET],
[LEFT_IDENTITY_BRACKET]))
composed_transducer = safe_compose(composed_transducer,
RI_obligatory_transducer)
if self.transformation_type == DELETION:
JL_obligatory_transducer = pyfst_from_dfa(
self._get_custom_obligatory_dfa([RIGHT_IDENTITY_BRACKET],
[LEFT_APPLICATION_BRACKET]))
composed_transducer = safe_compose(composed_transducer,
JL_obligatory_transducer)
composed_transducer = safe_compose(composed_transducer,
prologue_inverse_transducer)
if not self.obligatory: # obligatory rules should be weighted only in the replace level
if not self._should_fix_transducer(
): # only non-right-context rules should be weighted this way
composed_transducer = uniform_encoding.get_weighted_rule_transducer(
composed_transducer, self)
return composed_transducer
def get_ignore_dfa(sigma, language_dfa, ignored_set):
new_sigma = sigma | ignored_set
intro_transducer = get_intro_transducer(sigma, ignored_set)
language_transducer = pyfst_from_dfa(language_dfa)
composed_transducer = safe_compose(language_transducer, intro_transducer)
language_dfa = pyfst_to_dfa(composed_transducer, new_sigma)
return language_dfa
def get_prologued_word(word):
from bracket_rule_transducer import get_prologue_transducer
word_transducer = get_transducer_acceptor(word)
prologue_transducer = get_prologue_transducer()
prologued_word_transducer = safe_compose(word_transducer, prologue_transducer)
prologued_word_transducer.remove_epsilon()
return prologued_word_transducer
def get_left_to_right_application(self):
prologue_transducer = get_prologue_transducer()
if self.obligatory:
obligatory_transducer = pyfst_from_dfa(
self._get_obligatory_dfa([LEFT_IDENTITY_BRACKET],
[RIGHT_IDENTITY_BRACKET]))
composed_transducer = safe_compose(prologue_transducer,
obligatory_transducer)
else:
composed_transducer = prologue_transducer
right_context_transducer = self._get_right_context_dfa()
replace_transducer = self.get_replace_transducer()
left_context_transducer = self._get_left_context_dfa()
prologue_inverse_transducer = get_prologue_inverse_transducer()
composed_transducer = chain_safe_compose(composed_transducer,
right_context_transducer,
replace_transducer,
left_context_transducer)
if self.transformation_type == INSERTION:
insertion_obligatory_transducer = pyfst_from_dfa(
self._get_obligatory_dfa([RIGHT_IDENTITY_BRACKET],
[LEFT_IDENTITY_BRACKET]))
composed_transducer = safe_compose(
composed_transducer, insertion_obligatory_transducer)
# remove multiple paths
if self.transformation_type == ASSIMILATION or self.transformation_type == INSERTION:
RI_obligatory_transducer = pyfst_from_dfa(
self._get_custom_obligatory_dfa([RIGHT_APPLICATION_BRACKET],
[LEFT_IDENTITY_BRACKET]))
composed_transducer = safe_compose(composed_transducer,
RI_obligatory_transducer)
if self.transformation_type == DELETION:
JL_obligatory_transducer = pyfst_from_dfa(
self._get_custom_obligatory_dfa([RIGHT_IDENTITY_BRACKET],
[LEFT_APPLICATION_BRACKET]))
composed_transducer = safe_compose(composed_transducer,
JL_obligatory_transducer)
composed_transducer = safe_compose(composed_transducer,
prologue_inverse_transducer)
return composed_transducer
def get_outputs_of_word(self, word):
word_transducer = get_transducer_acceptor(word)
rule_set_transducer = self.get_transducer()
if rule_set_transducer:
dot(word_transducer, "word_transducer")
dot(rule_set_transducer, "rule_set_transducer")
word_rule_set_transducer = safe_compose(word_transducer,
rule_set_transducer)
word_rule_set_transducer.remove_epsilon()
dot(word_rule_set_transducer, "word_rule_set_transducer")
if len(word_rule_set_transducer):
try:
outputs = get_transducer_outputs(word_rule_set_transducer)
except:
print("get_outputs_of_word failed with word: {}".format(
word))
else:
outputs = []
else:
outputs = get_transducer_outputs(word_transducer)
return outputs
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