def min_consonant_count_transducer(min_consonant_count=3, add_meta_arc=True):
"""Allows only strings with at least |min_consonant_count| consonants."""
t = pt.Transducer()
for i in range(min_consonant_count+1):
for l in pt.abc.ALL_SYMS:
t.add_arc(i, i, l, l)
if i > 0:
for l in pt.abc.CONSONANTS:
t.add_arc(i-1, i, l, l)
t[min_consonant_count].final = True
if add_meta_arc:
pt.AddPassThroughArcs(t)
return t
def ApplyLoanwords(self, ar_vocab_groups, loanwords_transducer,
sw_pre_transducer, add_meta_arc, with_syllabification):
time_a = time.time()
sw_word_transducer = pt.UnionLinearChains(self.sw_pron_list)
if add_meta_arc:
pt.AddPassThroughArcs(sw_word_transducer)
if with_syllabification:
pt.AddSyllabificationArcs(sw_word_transducer)
sw_word_transducer.arc_sort_input()
time_sw = time.time()
print(" building SW transducer took:", time_sw-time_a, "sec")
ar_transducer = pt.UnionLinearChains(self.ar_word_list)
ar_transducer.arc_sort_output()
time_b = time.time()
print(" building AR transducer took:", time_b-time_sw, "sec")
print(" sw_pre_transducer")
sw_vocab = sw_pre_transducer >> sw_word_transducer
sw_vocab.arc_sort_input()
time_c = time.time()
print(" applying sw_pre_transducer took:", time_c-time_b, "sec")
print(" loanwords")
combined = loanwords_transducer >> sw_vocab
combined.arc_sort_input()
time_d = time.time()
print(" applying loanwords took:", time_d-time_c, "sec")
print(" ar_vocab")
self.t_all = pt.Transducer()
for ar_vocab in ar_vocab_groups:
print(".", sep="", end="")
sys.stdout.flush()
self.t_all.set_union(ar_vocab >> combined)
print()
self.t_all.arc_sort_input()
time_e = time.time()
print(" ar_vocab >> combined took:", time_e-time_d, "sec")
print(" t_correct")
self.t_correct = ar_transducer >> self.t_all
self.t_correct.arc_sort_output()
self.t_all.arc_sort_output()
time_g = time.time()
print(" building t_correct took:", time_g-time_e, "sec")
print(" total ApplyLoanwords took:", time_g-time_a, "sec")
def ar_morphology_transducer(add_meta_arc=True, with_syllabification=False):
"""Removes one AR prefix and one suffix (optionally)."""
rule_name = "<<IT_MORPH>>"
if add_meta_arc:
operation_weight = None
else:
operation_weight = pt.abc.OT_CONSTRAINTS[rule_name]
t = strip_transducer(morphemes.AR_PREFIXES, operation_weight, add_meta_arc,
rule_name)
t.concatenate(pt.accept_all_transducer())
t.concatenate(
strip_transducer(morphemes.AR_SUFFIXES, operation_weight, add_meta_arc,
rule_name))
if add_meta_arc:
pt.AddPassThroughArcs(t)
if with_syllabification:
pt.AddSyllabificationArcs(t)
#t.arc_sort_input()
return t
def sw_morphology_transducer(add_meta_arc=True, with_syllabification=False):
"""Appends SW prefixes and suffixes (optionally)."""
rule_name = "<<MT_MORPH>>"
if add_meta_arc:
operation_weight = None
else:
operation_weight = pt.abc.OT_CONSTRAINTS[rule_name]
t = append_transducer(morphemes.SW_PREFIXES, operation_weight,
add_meta_arc, rule_name)
#t.concatenate(append_transducer(morphemes.SW_PREFIXES, operation_weight, add_meta_arc, rule_name))
t.concatenate(pt.accept_all_transducer())
t.concatenate(
append_transducer(morphemes.SW_SUFFIXES, operation_weight,
add_meta_arc, rule_name))
if add_meta_arc:
pt.AddPassThroughArcs(t)
if with_syllabification:
pt.AddSyllabificationArcs(t)
#t.arc_sort_output()
return t
def vowel_deletion_transducer(add_meta_arc=True):
"""Deletion of vowels."""
t = pt.Transducer()
for l in pt.abc.ALL_SYMS:
t.add_arc(0, 0, l, l)
next_node = 1
for l in pt.abc.VOWELS:
t.add_arc(0, next_node, l, pt.abc.EPSILON)
rule_name = "<<MAX-V>>"
if add_meta_arc:
t.add_arc(next_node, 0, pt.abc.EPSILON, rule_name)
else:
t.add_arc(next_node, 0, pt.abc.EPSILON, pt.abc.EPSILON, pt.abc.OT_CONSTRAINTS[rule_name])
next_node += 1
t[0].final = True
if add_meta_arc:
pt.AddPassThroughArcs(t)
return t
def AssertReachable(ar_str, sw_str, ar_post_transducer, loanwords_transducer,
sw_pre_transducer, add_meta_arc=True, with_syllabification=False):
print("Constructing input transducer for word:", ar_str)
in_t = pt.linear_chain(ar_str)
in_t = in_t >> ar_post_transducer
print("Constructing output transducer for word:", sw_str)
out_t = pt.linear_chain(sw_str)
if add_meta_arc:
pt.AddPassThroughArcs(out_t)
if with_syllabification:
pt.AddSyllabificationArcs(out_t)
out_t = sw_pre_transducer >> out_t
if add_meta_arc:
out_t = out_t >> pt.weights_transducer()
out_t.arc_sort_input()
in_t.arc_sort_output()
print("Combining loanwords with output")
test_t1 = loanwords_transducer >> out_t
if len(test_t1) == 0:
print("NOT REACHABLE: loanwords_transducer >> out_t")
print("Testing input with loanwords")
test_t3 = in_t >> loanwords_transducer
if len(test_t3) == 0:
print("NOT REACHABLE: in_t >> loanwords_transducer")
print("Combining with input")
test_t2 = in_t >> test_t1
if len(test_t2) == 0:
print("NOT REACHABLE: in_t >> test_t1")
else:
print("Reachable!")
print("Printing full paths")
pt.PrintFullPaths(test_t2, num_shortest=1)
print("AssertReachable done.")