class Frame_aligner(Block):
def __init__(self,
align_file_name="",
output_folder="",
output_name="",
obl_ratio_limit="0.5",
min_obl_inst_num=2,
**kwargs):
""" overriden block method """
super().__init__(**kwargs)
self.align_file = None
self.output_folder = output_folder
self.output_name = output_name
self.obl_ratio_limit = float(obl_ratio_limit)
self.min_obl_inst_num = int(min_obl_inst_num)
self.linker = Linker() # !!!
if align_file_name != "":
try:
self.align_file = open(align_file_name, 'r')
except FileNotFoundError:
#print( "Cesta: " + os.path.dirname(os.path.realpath(__file__)))
print("ERROR: Alignment file " + align_file_name +
" not found.")
exit()
#self.a_and_b = 0
#self.a_only = 0
#self.b_only = 0
#self.direction = 0 # 0 .. both, 1 .. a -> b, 2 .. b -> a
# to be overloaded
self.a_frame_extractor = Frame_extractor()
self.b_frame_extractor = Frame_extractor()
self.a_lang_mark = ""
self.b_lang_mark = ""
self.examiner = Modal_examiner()
def process_bundle(self, bundle): # void
""" overriden Block method """
#logging.info( "bundle id: " + str( bundle.bundle_id))
a_frame_insts = []
b_frame_insts = []
for tree_root in bundle.trees:
if tree_root.zone == self.a_lang_mark:
a_frame_insts = \
self.a_frame_extractor.process_tree( tree_root)
self.examiner.examine_sentence(tree_root, self.a_lang_mark)
elif tree_root.zone == self.b_lang_mark:
b_frame_insts = \
self.b_frame_extractor.process_tree( tree_root)
self.examiner.examine_sentence(tree_root, self.b_lang_mark)
word_alignments = self.align_file.readline().split()
frame_pairs = self.linker.find_frame_pairs(a_frame_insts,
b_frame_insts,
word_alignments)
print(len(frame_pairs))
for frame_pair in frame_pairs:
# linking frame types
# if the frame type link does not exist yet, create one
a_frame_type = frame_pair.a_frame_type
b_frame_type = frame_pair.b_frame_type
a_b_frame_type_link = a_frame_type.find_link_with(b_frame_type)
# could be done the other way around: b_frame.find( frst_frame)
if a_b_frame_type_link is None:
a_b_frame_type_link = Frame_type_link(a_frame_type,
b_frame_type)
# linking frame instances
a_frame_inst = frame_pair.a_frame_inst
b_frame_inst = frame_pair.b_frame_inst
a_b_frame_type_link.link_frame_insts(a_frame_inst, b_frame_inst)
# ADDED FOR MODALS, DELETE
return a_frame_insts, b_frame_insts, frame_pairs
def after_process_document(self, doc): # void
""" overriden block method """
self.examiner.print_stats()
a_dict_of_verbs = self.a_frame_extractor.get_dict_of_verbs()
b_dict_of_verbs = self.b_frame_extractor.get_dict_of_verbs()
#self.end_obl( a_dict_of_verbs)
#self.end_obl( b_dict_of_verbs)
#self._finalize_dictionary( a_dict_of_verbs, self.a_lang_mark)
#self._finalize_dictionary( b_dict_of_verbs, self.b_lang_mark)
#self._output_control()
self._pickle_dict(a_dict_of_verbs, b_dict_of_verbs)
#return
#print( "=== pocty slovies ===")
#print( self.a_lang_mark, len( self._a_dict_of_verbs))
#print( self.b_lang_mark, len( self._b_dict_of_verbs))
#print( self.a_and_b, self.a_only, self.b_only)
super().after_process_document(doc)
def end_obl(self, dict_of_verbs):
suma = 0
for vr in dict_of_verbs.values():
for ft in vr.frame_types:
for fta in ft.args:
if fta.deprel == "obl":
suma += len(fta.insts)
print(suma)
def _finalize_dictionary(self, dict_of_verbs, lang_mark):
#obl_ratio_limit = 0#.5
extraction_finalizer = \
Extraction_finalizer( dict_of_verbs, self.obl_ratio_limit,
self.min_obl_inst_num, lang_mark)
for verb_record in dict_of_verbs.values():
verb_record.finalize_extraction(extraction_finalizer)
extraction_finalizer.finalize_extraction()
def _pickle_dict(self, a_dict_of_verbs, b_dict_of_verbs): # void
""" called from after_process_document """
a_b_dicts_of_verbs = a_dict_of_verbs, b_dict_of_verbs
logging.info(sys.getrecursionlimit())
logging.info(sys.getsizeof(a_b_dicts_of_verbs))
sys.setrecursionlimit(50000)
logging.info(sys.getrecursionlimit())
#a_output_name = self.output_folder + self.a_lang_mark + \
# "_" + self.b_lang_mark + "_" + self.output_name
#b_output_name = self.output_folder + self.b_lang_mark + \
# "_" + self.a_lang_mark + "_" + self.output_name
a_b_output_name = self.output_folder + "_" + self.output_name
pickle.dump(a_b_dicts_of_verbs, open(a_b_output_name, 'wb'))
#pickle.dump( a_dict_of_verbs, open( a_output_name, 'wb'))
#pickle.dump( b_dict_of_verbs, open( b_output_name, 'wb'))
def _output_control(self): # void
""" called from after_process_document """
a_dict_of_verbs = self.a_frame_extractor.dict_of_verbs
for verb_record in list(a_dict_of_verbs.values()):
for ft in verb_record.frame_types:
for fi in ft.insts:
for fia in fi.args:
fial = fia.frame_inst_arg_link
print(fia.node.form, fial)
